Lexlegis.ai vs Manupatra: how they compare in 2026

L
Lexlegis.ai profile
M
Manupatra profile
Last verifiedSeptember 27, 2026

Lexlegis.ai and Manupatra both put generative AI over Indian case law and statutes for lawyers, tax professionals and in house counsel. Lexlegis.ai sits in the top two bands on twelve of fifteen axes and Manupatra on ten of fifteen, identical on ten. Lexlegis.ai leads on where it can run and who vouches for it. It publishes five deployment modes, from Indian cloud to air gapped installations, holds ISO 27001 with SOC 2 reports on request, and names KPMG and the Government of India among customers. Manupatra's lead is in its contract and its corpus. Its agreement bars training on uploaded material, including anonymized use, without separate written consent, and requires deletion within 24 hours with breach notice inside the same window. Its own flags mark whether a judgment is still good law. Manupatra names no customer and claims no security certification. Both claim their AI eliminates hallucinations, and neither publishes a measurement behind the claim.

At a glance

Category
Lexlegis.aiLegal Research
ManupatraLegal Research
Founded
Lexlegis.aiNot published
Manupatra2000
Headquarters
Lexlegis.aiMumbai, Maharashtra, India
ManupatraNoida, Uttar Pradesh, India
Last verified
Lexlegis.aiSep 7, 2026
ManupatraSep 13, 2026

All 15 axes, side by side

The same grid applied to every vendor in the index, graded from public sources. Hover a grade to see what the letter means on that axis.

AI Centrality

How much of the product is actually AI. Whether the machine learning is the mechanism the buyer is paying for or a feature layered onto conventional software, and whether the vendor is specific about which is which.

Lexlegis.ai
AA on AI CentralityThe artificial intelligence is the product. Remove the models and there is nothing left to sell.

Every product sold is model work and the company describes itself as AI-first. Legal AID is three generative capabilities: Ask answers legal questions in the IRAC format with sources returned and explained, Interact reads uploaded contracts, orders and notices and answers across them, and Draft produces contracts, submissions, written arguments, oral argument outlines and authority bundles. MIRA sits on the same foundation as a skill-configurable agent that reads a brief, selects from a library of 215 production skills, orders them into a plan, reasons through each step and checks its own reasoning against a verification layer before producing output. The agreement defines the Service as the platform including Ask, Interact, Draft and MIRA, so the model products are the contracted thing rather than a layer over something else. The vendor states it trained its own model on its corpus and positions that against fine-tuning and retrieval-augmented generation, and its published model layer describes signed checkpoints, per-tenant guardrails and an inference audit trail, which are the controls of an organization running its own inference rather than reselling someone else's. The non-model asset is real and is described honestly rather than credited: a curated closed corpus of Indian laws, circulars and judgments built over decades of legal documentation work. No separately purchasable non-AI product was located on any surface read, and the corpus is an input to the assistants rather than a search subscription sold on its own. Terms of service v2.0, trust page and product pages read 7 September 2026.

Manupatra
BB on AI CentralityThe models are the engine of a core capability, layered on a product that would still function without them as a document or workflow system.

Artificial intelligence is a substantial, named, shipped layer on a research platform that existed for twenty-five years without it, which is the B band rather than the A. What ships is specific and dated. Manupatra AI Search launched in December 2025, is described by the vendor as powered by semantic AI and large language models, and retrieves judgments by legal concept rather than keyword; AI Summary generates judgment summaries; AI Compare sets judgments side by side. The separately branded ManuWorks toolkit adds OCR, summarization, translation, document question-answering, case timeline generation, document comparison and AI-assisted drafting, each with its own published credit rate. The product page carries the HTML title 'Powered by AI' and the headline 'India's trusted Legal Research partner, Powered by AI', and the features link states the platform uses NLP, AI and ML for information retrieval. What holds it off A is that the vendor itself positions the AI as one retrieval mode among several rather than as the product. Its own guidance directs users to MANU SEARCH, LEGAL SEARCH and CITATION SEARCH for precise, literal or Boolean queries and reserves AI Search for meaning-based ones, and the AI is metered separately from the content subscription through a credit wallet, so a subscriber can hold the platform without holding the AI. The underlying asset is the editorially enriched corpus and the citation and treatment apparatus, which predate the models and would survive their removal. Verified 13 September 2026.

Citation Accuracy and Hallucination Disclosure

Whether the vendor publishes measured accuracy on citations and assertions, grounds output to primary sources, and says plainly what its system does when it does not know. Legal has a documented public record of fabricated citations reaching filed briefs, so an untested claim of accuracy is not evidence.

Lexlegis.ai
BB on Citation Accuracy and Hallucination DisclosureGrounding is real and documented, with linked primary sources and a described retrieval method, short of published accuracy figures an outsider can test.

Real architectural controls sit under an unqualified accuracy claim that is never measured, which is the shape this band's middle grade exists for. The controls are documented and specific. Output is returned in the IRAC format with a list of the sources relied on, each openable and downloadable, with a summary of the source and a statement of why it was pertinent. Retrieval is from a closed curated database of laws, circulars and judgments rather than the open web, and the vendor describes the order as retrieve from the corpus, then reason, then verify. The published model layer names output hallucination detection and a meta reasoning gate, per-tenant guardrails and signed model checkpoints. The reasoning chain is stated in absolute terms: every factual claim in every output is traced back to its source, and if the trace is broken the claim is flagged, with no exceptions. MIRA is described as checking its own reasoning against a layer built specifically to catch its mistakes. What is absent is any measurement of the claim. The vendor markets research without hallucination and grounded, hallucination-free legal intelligence, and publishes no accuracy figure, no error rate, no hallucination rate, no test set, no benchmark and no evaluation result for any of it. An absolute negative claim with no measurement behind it is weaker evidence than a modest measured one, and it is recorded here because a reader should be able to see the gap between the claim and what supports it. The trust center lists an item titled A note on Hallucinations and a report titled AI Flow for Ask; neither was opened on this channel and both are the named rebuttal route on this row. Trust page, product pages and terms read 7 September 2026.

Manupatra
BB on Citation Accuracy and Hallucination DisclosureGrounding is real and documented, with linked primary sources and a described retrieval method, short of published accuracy figures an outsider can test.

Grounding is real, documented and architecturally specific, and no accuracy figure is published, which is the B band precisely. The grounding is described in the vendor's own AI brochure and is more than an assertion: AI Search runs against a proprietary, editorially enriched corpus rather than public web data, results are returned as the ten most relevant judgments organized into Facts, Issues, Arguments, Reasoning, Ratio and Summary, hovering Reference confirms the case source, clicking Citation opens the full judgment, and the vendor states that every insight is explainable and traceable to verified legal sources. The platform also surfaces judicial treatment directly, detecting overruled, distinguished and conflicting judgments and marking currency with blue, yellow and red flags, so the citation-status limb of the A band is satisfied even though the band overall is not. The index's rule governs the hallucination claim rather than the D limb: the brochure states that the editorial foundation ensures the AI operates strictly within the legal framework, significantly reducing inaccuracies and eliminating AI hallucinations, and that claim stands alongside real controls rather than in place of them, so 'bare' does not apply. The failure-modes limb of the A band is satisfied, and it is the most substantive thing this vendor publishes about the limits of its own AI. The brochure carries a section headed 'Guidance on Using AI Search' with an express 'When NOT to Use AI Search' list, telling users to avoid AI Search for precise or literal queries, party names, case citations, exact statutory phrases and verbatim quotations, stating that it does not follow Boolean logic, and routing those queries by name to MANU SEARCH, LEGAL SEARCH and CITATION SEARCH. A further page contrasts well-framed and poorly framed queries across nine worked examples. Very few records in this corpus publish where their own retrieval is weak, and this one does it in the vendor's own voice and in specific terms. What keeps the row at B is the limb the band leads on: no measured accuracy is published for AI Search, AI Summary or AI Compare, no test set is described, no error rate or evaluation exists, and nothing quantifies the elimination claim, which is an absolute assertion of the kind this axis exists to test. The remaining A limb also fails: nothing states that the system reports finding no support, and on a product that always returns a ranked ten, whether the tenth result is relevant or merely nearest is unanswered. The agreement pulls the other way and is recorded because it is the vendor's own instrument: clause 11.12 states that judgment text on the site is computer generated and that authenticity, correctness and preciseness must be verified from the certified copy, and clauses 15.1 to 15.3 disclaim any warranty of accuracy and place verification checkpoints on the user. Verified 13 September 2026.

Autonomy and Oversight Model

What the system decides on its own, what a lawyer must approve, and whether the vendor documents where the review point sits. A tool that drafts under review and a tool that files without one are different products and different risks.

Lexlegis.ai
BB on Autonomy and Oversight ModelA written commitment that the models work alongside a supervising lawyer, with real review surfaces, short of the full control structure: commonly the threshold at which the system stops or what happens after it is wrong.

The oversight position is contractual and an automated checkpoint is described, with the threshold at which the agent acts alone left unstated. Clause 8 of the terms is headed AI outputs and professional responsibility and is unusually direct: outputs are tools to support legal or compliance work and not legal advice, the customer remains responsible for reviewing and validating all outputs before relying on them, filing them or acting upon them, no output creates a lawyer client relationship with the vendor, and the customer's professional judgment is not replaced by the Service. Clause 7 adds a conduct obligation most agreements omit, that the customer will not use outputs to falsely attribute authority or to deceive any court or tribunal. On the product side an automated gate is described rather than merely asserted: the model layer publishes a meta reasoning gate and output hallucination detection through which every input and output passes, with the vendor stating that failure of any layer opens a ticket rather than a breach, and MIRA is described as verifying its own reasoning before producing output. Marketing states the position plainly, that the routine work is taken and the judgment stays with the user, end to end under supervision. What is missing is the limb this band names as commonly absent. Nothing states what MIRA does without a person, at what point a skill chain completes unattended, whether any step requires human approval before the next runs, what a user sees when the meta reasoning gate rejects a claim, or whether autonomy is configurable. Terms of service and trust page read in full 7 September 2026.

Manupatra
BB on Autonomy and Oversight ModelA written commitment that the models work alongside a supervising lawyer, with real review surfaces, short of the full control structure: commonly the threshold at which the system stops or what happens after it is wrong.

Substantive published material on how the tool is meant to be used and where it should not be relied on, short of the A band's thresholds and review surfaces. The strongest item is unusual enough to name in full. The vendor's AI brochure carries a section headed 'Guidance on Using AI Search' with an express 'When NOT to Use AI Search' list, telling users to avoid AI Search for precise or literal queries, party names, case citations, exact statutory phrases and verbatim quotations, stating that AI Search does not follow Boolean logic, and routing those queries to the named non-AI modes MANU SEARCH, LEGAL SEARCH and CITATION SEARCH. That is a published negative boundary attaching to a named mode, and it is rarer in this corpus than a confidence score. Around it sit real verification affordances: Reference hover to confirm the source, Citation click to open the full judgment, and a follow-up chatbot for narrowing rather than for acting. The agreement reinforces it, clause 15.3 placing responsibility for sufficient procedures and checkpoints on the user and clause 11.12 requiring verification against the certified copy. What is absent is what A asks for. No confidence score, no threshold, no statement of what the system does with a query it cannot answer well, no review surface before output, and nothing at all on oversight of the ManuWorks AI Drafter, which generates legal drafts at 50 credits per query with no described review step. R124(2) is the reason this is B and not A: the constraint published here attaches to a named mode but describes which queries the mode suits rather than what its output may not be used for, which is the limb the Verbit tier constraint satisfied. Recorded against parking item 2 as the closest instance in this lane. Verified 13 September 2026.

Operational and Outcome Evidence

Named, dated evidence that the product works in production at real firms or legal departments. Case studies with figures and identified customers count. Unattributed testimonials and launch announcements do not.

Lexlegis.ai
BB on Operational and Outcome EvidenceReal deployment evidence with substance, short of full attribution or measurement: a named customer without figures, or figures without the named customer.

A named customer roster with real institutional weight, and no outcome measured or attributed. The roster is published on the homepage and spans professional services, industry and government: KPMG, Dhruva Advisors, Greaves Cotton, Thermax, K Raheja Corp, Aurtus, MyGate, Artha Energy and the Government of India, with the vendor stating more than 100 enterprises. A government body and a Big Four firm as named users are meaningful signals in this market. Public milestones are dated and checkable: the platform was showcased at NVIDIA GTC 2026 in March 2026, where the company announced a preview beta launch in the United States, and the skill library is quantified at 215 production skills. What is absent is everything past the logo. No case study is published, no customer is quoted by name and role, no deployment is described, and no outcome is measured: no time saved, no volume processed, no adoption figure, no accuracy or quality result from any customer. The efficiency claims that do appear are unattributed and unquantified, being research memos in hours rather than days and authority bundles built automatically. A reader can establish who is said to use it and nothing about what it did for them. Homepage, about page, trust page and the March 2026 announcement read 7 September 2026.

Manupatra
CC on Operational and Outcome EvidenceCustomer logos and unattributed testimonials stand in for evidence, or results are quoted with no basis stated.

Scale is claimed at length and no customer is named and no outcome is measured, which is the C band. What is published is substantial as usage evidence: more than 150,000 active users, more than 10 million legal documents and judgments indexed, twenty-five years of operation, a team of more than 175, and offices across India and Bangladesh. The About page's 'Who We Serve' section enumerates five clientele categories, being legal professionals and law firms, in-house legal departments in corporations and banks, academic institutions, government bodies and regulatory agencies, and law students and researchers across Asia. Categories are not customers. Not one law firm, company, court, tribunal or university is named anywhere on the surfaces read, there is no client wall, no case study, no testimonial and no logo strip, which is a marked contrast with the penetration disclosure typical of vendors at this scale in the corpus. Nor is any outcome quantified: nothing states time saved, research hours displaced, accuracy improved or matters affected, and the AI brochure's benefit claims are functional descriptions rather than measurements. The 150,000 figure is itself unattributed and undated, appearing as a counter on the product page and in the About page metadata. The seed recorded 100,000 users, which the vendor's own current pages contradict; the figure written here is the vendor's. Verified 13 September 2026.

Privilege and Confidentiality Posture

How client confidences are handled: attorney client privilege and work product treatment, segregation of one client matter from another, whether client data trains any model, and what the vendor commits to in writing rather than in marketing.

Lexlegis.ai
BB on Privilege and Confidentiality PostureSubstantive published commitments on confidentiality and training use, short of the full picture: commonly silence on segregation between users or matters, or on what the underlying model provider may retain.

Substantive published commitments across every limb this axis tests except the one that decides the top grade. What is published, and most of it contractually: clause 10 of the terms binds each party to hold the other's confidential information in confidence and use it only for the purposes of the agreement, names Customer Content as the Customer's confidential information, and survives termination for five years or in perpetuity for trade secrets. Clause 5 leaves all rights in customer content with the customer and grants the vendor a license to process it solely to provide the Service. The privacy policy states that content data is governed by strict access controls and is accessible only to the account it belongs to. The trust page publishes matter level isolation on multi-tenant software as a service, which is a direct answer to the question this axis asks and one most records cannot give, alongside role based access control with custom roles, single sign-on required on all enterprise deployments, least privilege, access log management, and session recording and replay on enterprise. Customer managed keys with hardware security module storage are available on the non-SaaS modes, and retention is configurable to zero on enterprise and on the on-premises and air-gapped modes. What is absent is express privilege and work product treatment. The words privilege and work product appear nowhere on any surface read; clause 8's statement that no lawyer client relationship arises with the vendor addresses the vendor's own position, not the protection of the customer's privilege. That limb is required, so the record holds below the top grade. Terms, privacy policy and trust page read in full 7 September 2026.

Manupatra
BB on Privilege and Confidentiality PostureSubstantive published commitments on confidentiality and training use, short of the full picture: commonly silence on segregation between users or matters, or on what the underlying model provider may retain.

Substantive published commitments across every limb this axis tests except the one the index's rule makes decisive, which is what holds an otherwise exceptional record at B. The commitments sit in the published customer agreement rather than a policy page, and they are unusually precise. Clause 4.1 defines Confidential Information to include documents, data, personal data, case files, names, facts, legal strategies and client information uploaded through the website, the AI tools or any related service. Clause 4.2 requires that it be treated as strictly confidential, used solely for the limited purpose of providing the specific service requested, and not copied, stored, retained, archived, analyzed, mined, disclosed or shared for any other purpose, with access restricted to need-to-know personnel under equivalent confidentiality obligations. Clause 4.3 bars disclosure to any third party including affiliates, subcontractors, AI engines and cloud providers without prior express written consent. Clause 4.7 requires permanent deletion of all Confidential Information and any copies, backups, AI memory, logs, caches or traces immediately after the request is fulfilled and in no event later than 24 hours, with written certification of deletion available on request. Clause 5.8 bars employee viewing of Subscriber Content or AI outputs except to resolve a specific technical issue the Subscriber has raised, with such access logged and minimized. Clause 4.8 makes these obligations survive termination indefinitely. Training is consent-gated and graded separately. The limb that fails is privilege and work product, and it fails as a matter of language rather than of substance: neither privilege, professional secrecy, work product nor any equivalent is named anywhere, and nothing addresses what happens to privileged material if disclosure is compelled. Clause 4.1's enumeration of case files and legal strategies is the closest the corpus has come outside Verbit's express work-product clause, and it is not the same thing. Verified 13 September 2026.

UPL and Professional Responsibility Posture

Whether the vendor is clear that it supplies a tool rather than legal advice, who its audience is, and how it addresses unauthorized practice of law, competence and supervision duties, and jurisdiction limits. ABA Formal Opinion 512 is the reference point. Where the advice line is not the duty a product raises, the axis is read through the nearest professional duty it does raise: judicial conduct rules and the reviewing duty for products sold only to courts, and the duty to bill for time actually spent for products that draft time entries.

Lexlegis.ai
BB on UPL and Professional Responsibility PostureA real position is published on advice versus tooling, short of full treatment: commonly a disclaimer without the supervision and competence dimension, or silence on jurisdiction limits.

The boundary is drawn in the agreement, backed by an access gate, and no governing professional instrument is named. What is published: clause 8 states that outputs are tools to support legal or compliance work and not legal advice, that the customer remains responsible for reviewing and validating outputs before relying on, filing or acting on them, that no output constitutes a lawyer client relationship with the vendor, and that professional judgment is not replaced. Clause 7 carries a prohibition rarely seen in a legal AI agreement and directly on this axis: the customer will not use outputs to falsely attribute authority or to deceive any court or tribunal. Clause 2 gates the product itself, restricting paid plan eligibility to legal or compliance professionals or those employed in a role requiring legal research and drafting tools, which is an access control rather than a disclaimer and is a real answer to the unauthorised practice question. The vendor also states that every skill is designed, tested and validated by senior lawyers. What is absent is any named authority. No rule of the Bar Council of India, no provision of the Advocates Act, no state bar council guidance and no court practice direction is cited anywhere, in a jurisdiction where the Supreme Court has publicly taken up the consequences of AI-generated authority in filings. Terms of service and about page read in full 7 September 2026.

Manupatra
BB on UPL and Professional Responsibility PostureA real position is published on advice versus tooling, short of full treatment: commonly a disclaimer without the supervision and competence dimension, or silence on jurisdiction limits.

Substantive published commitments short of alignment with any conduct rule, which is the B band. The agreement addresses the question head on rather than by implication. Clause 16.1, headed 'No Legal Advice', states that material contained on or made available through the site is not intended to and does not constitute legal advice and does not in any manner establish a client-advocate relationship. The disclaimer published on the legal page repeats that the company does not warrant any form it provides and is not creating a lawyer-client relationship by providing forms, and adds that forms should be used as a guide and modified to suit requirements and laws, at the user's own risk. Clause 15.3 places on the user the responsibility for implementing sufficient procedures and checkpoints to satisfy accuracy requirements for input and output. Clause 11.12 goes further than most records in this corpus by requiring that the authenticity, correctness and preciseness of judgment text be verified from the certified copy of the judgment, which is a verification duty rather than a disclaimer. Taken together that is a real posture on who is answerable for the output. What is absent is any engagement with the professional rules themselves. No conduct rule of any jurisdiction is named, the Bar Council of India is not mentioned, and no clause requires the subscriber to use the product consistently with professional obligations. The gap is sharper than usual on this record and is named rather than passed over: the Supreme Court of India held in July 2026 that citing unverified AI-generated precedent is professional misconduct and directed the Bar Council of India to frame norms, and this vendor publishes that ruling as editorial content on its own updates service while publishing nothing about what its own AI features mean for that duty. Verified 13 September 2026.

AI Governance and Bias Disclosure

Published governance over model behavior: who owns it inside the vendor, what is tested before release, and what is disclosed about disparate output across matter types, parties, or populations.

Lexlegis.ai
CC on AI Governance and Bias DisclosureResponsible AI principles are published without a mechanism, a testing regime, or anything a buyer could audit.

Technical controls over the model are published in detail and no governance apparatus is described. What is genuinely governance-adjacent and credited here rather than on the security row: per-tenant custom guardrails, which put configuration of model behavior in the customer's hands, and the vendor's statement that every skill in the library is designed, tested and validated by senior lawyers before it ships, which is a human review gate on the product's content. The trust center also lists a code of ethics, anti-bribery and anti-modern-slavery items, and a product security entry titled A note on Hallucinations. What is missing is everything that would make this governance of the AI rather than security around it. No AI governance framework or standard is published or claimed and no ISO/IEC 42001 or equivalent is asserted, in contrast to the five security and privacy frameworks that are asserted with precision. Nobody is named as accountable for model behavior. No pre-release evaluation protocol, benchmark, red-team exercise or published result exists for any model or skill. And no bias, fairness or representativeness disclosure appears at any level, which is worth naming for a product whose corpus is Indian judicial and tax material and whose outputs are used in a jurisdiction where courts have begun scrutinizing AI-derived authority. Trust page, about page and trust center index read 7 September 2026.

Manupatra
CC on AI Governance and Bias DisclosureResponsible AI principles are published without a mechanism, a testing regime, or anything a buyer could audit.

Something is published and it is a paragraph, which is the C band rather than the D. The legal page carries a section headed 'Use of AI' reading, in full, that the vendor is committed to using artificial intelligence responsibly and prioritizes data privacy, bias mitigation, transparency, human oversight, accountability, user consent, and collaboration with regulations, with the stated aim of maximizing AI's benefits while ensuring ethical practices and minimizing potential risks. That is a list of the right words and it is the whole of it. Nothing behind any of the seven named priorities is published: no framework is adopted or certified, no accountable owner or committee is named, no governance process is described, no bias testing is described and no results of any kind are disclosed, and nothing states how human oversight is exercised in a product where the oversight the vendor actually documents is guidance to the user rather than a control in the system. Bias in particular is asserted as a priority and never returned to, on a product whose AI ranks and summarizes judicial authority and surfaces judicial treatment, where a systematic tilt in which precedents surface would be invisible to the researcher. Recorded and not credited as governance, because they are commercial or contractual rather than governance instruments: the agreement's AI section is a strong commitment and is graded on the confidentiality and training rows, and the responsible-AI paragraph is not tied to it. The Grievance Redressal Officer named on the same page is a statutory Indian IT Act appointment for complaints handling, not an AI accountability owner, and is recorded as such. Verified 13 September 2026.

AI Safety and Data Stewardship

Retention, deletion, access control, and what happens to prompts and documents after they are processed. Whether the vendor states its subprocessors and its incident practice, or leaves the buyer to assume.

Lexlegis.ai
AA on AI Safety and Data StewardshipRetention, deletion, access control, subprocessors and incident practice are all published, current, and specific enough to hold the vendor to.

Published policy across every limb this axis asks about, in the agreement and in a structured trust architecture, with one gap named. Certifications are asserted with precision: ISO/IEC 27001 certification stated as the 2022 revision, SOC 2 Type 2, ISO/IEC 20000-1:2018, plus GDPR, CCPA and India's DPDP alignment, with CERT-In empanelled audits and annual audit stated where applicable. Encryption is specified as TLS 1.3 in transit and AES 256 at rest, with customer managed keys and hardware security module key storage available on the non-SaaS modes. Access control is detailed: SAML 2.0, OIDC and SCIM 2.0 provisioning, single sign-on required on enterprise, role based access control with custom roles, multi-factor enforced through the customer's identity provider, least privilege and access log management. Network controls include virtual private cloud isolation, private link, web application firewall, denial of service protection, IP allow listing and geographic restriction, with no inbound traffic at all on the on-premises and air-gapped modes. On the model itself, prompt injection detection and sanitization and model poisoning defense via signed checkpoints. Training, retention and deletion are all committed: no customer content in shared training corpora, retention configurable and reducible to zero on enterprise and the isolated modes, data erasure documented, and data exportable at any time with return or destruction at customer election on termination, with attestation. Incident practice exists on both sides, a data breach notification item in the trust center and a coordinated disclosure program committing acknowledgment within 24 hours, triage within 72 and remediation to severity-based service levels, with safe harbor for good faith researchers. The gap, named because every partial names its limitation: the privacy policy states that a current sub-processor list is published at the trust center, and no such list appears on that page. Privacy policy, terms, trust page and trust center read 7 September 2026.

Manupatra
BB on AI Safety and Data StewardshipSubstantive published policy covering most of the ground, short of the full set: commonly no named subprocessor list or no stated incident practice.

Substantive published policy covering most of the ground, missing the named subprocessor list, which is the B band and the single limb it names. What is published is contractual rather than promotional and it is specific. Clause 4.4 requires industry-standard technical and organisational security measures. Clause 4.5 commits to notifying the Subscriber in writing within 24 hours of becoming aware of any actual or suspected unauthorised access, disclosure, loss, corruption or breach, to providing full details and impact, to containing, investigating and remediating at its own cost, and to cooperating in legal, regulatory or remedial action. A named notification window with a stated content requirement is the limb most records in this corpus lack entirely. Clause 4.6 accepts the role of data processor with the Subscriber as data fiduciary under the Digital Personal Data Protection Act 2023 and commits to the Information Technology Act 2000. Clause 4.7 requires deletion within 24 hours with written certification. Clause 5.3 requires that any storage be temporary, encrypted and automatically deleted after use. The privacy policy adds encryption technologies, secure servers and firewalls, access control mechanisms, employee confidentiality agreements and training, and states these are regularly reviewed, and records compliance with the SPDI Rules 2011. What is missing is the subprocessor limb and it is missing in an interesting way. No processor is named anywhere, and clause 4.3 explains why the vendor might think none needs naming, since it bars disclosure to any third party including subcontractors, AI engines and cloud providers without express written consent. The privacy policy nonetheless discloses that third parties perform payment processing, email delivery and hosting, and that the products may integrate cloud storage services including Dropbox and Google Drive, so processors exist and none is identified. No certification supports any of it. Verified 13 September 2026.

AI Liability and Recourse

What the vendor stands behind contractually when its output is wrong. Indemnities, caps, carve outs, insurance, and whether any of it is published or only reachable through a negotiated agreement.

Lexlegis.ai
BB on AI Liability and RecourseA real published position on liability, short of the full picture: commonly a stated indemnity without scope or caps.

More of what this axis asks for than most records carry, with one absence that keeps it off the top. What is published: an affirmative warranty, clause 11 warranting that the Service will be provided with reasonable care and skill, which is a positive obligation rather than the pure as-is disclaimer that is the norm in this corpus. A service level with a number, clause 9 carrying a 99.5 percent monthly uptime target on the SaaS mode and bespoke service levels on enterprise and non-SaaS modes set out in the order form, with a named customer success manager for enterprise. Insurance disclosed rather than absent, the trust center listing both cyber insurance and professional indemnity insurance in its legal section. A liability cap at clause 12 set at fees paid in the twelve months preceding the claim, mutual exclusion of indirect, incidental, consequential and punitive damages, and express carve-outs preserving liability for gross negligence, wilful misconduct and anything that cannot be limited by law. Termination at clause 13 on 30 days uncured material breach, with customer content returned or destroyed at the customer's election and, per the trust page, with attestation. What is absent is indemnity. No indemnity of any kind runs to the customer: no intellectual property indemnity, no defense obligation, no remedy ladder, nothing. And the warranty that does exist covers the manner of provision rather than the product's output, clause 11 stating expressly that outputs are not warranted to be accurate, complete, error free or fit for purpose, so on the question this axis exists to ask the customer still bears the loss when an output is wrong. Terms of service and trust center read in full 7 September 2026.

Manupatra
BB on AI Liability and RecourseA real published position on liability, short of the full picture: commonly a stated indemnity without scope or caps.

A published agreement addresses three of the four limbs with real specificity and the overall allocation is heavily one-sided, which is B rather than A. Warranty is addressed by exclusion: clauses 25.1 and 25.2 provide the information on an 'As Is' basis and disclaim all warranties including merchantability, fitness for purpose and non-infringement, and clauses 15.1 and 15.2 disclaim any representation that the service will be correct, accurate, reliable, uninterrupted or timely. Liability is addressed and largely excluded: clauses 25.3 and 25.4 exclude liability in contract or tort for any loss howsoever arising in connection with the service, expressly including where caused by the vendor's own negligence, and clause 25.5(i) is the provision that matters most on this axis, excluding liability for any claim relating to the subscriber's inability or failure to perform legal research work properly or completely, or to any decision made or action taken in reliance on the Data. A cap exists at clause 26.1, limited to the aggregate charges paid during the subscription period, though it is drafted as applying to claims relating to copyright in the data rather than as a general cap, which is a real gap named here rather than smoothed over. What lifts this above the floor is that the agreement also runs the other way, which is uncommon in this corpus. Clauses 21.2 and 26.3 require the vendor to indemnify, defend and hold harmless the Subscriber, its clients, employees and affiliates against claims, losses, damages, liabilities, penalties and legal costs arising from any breach of confidentiality or misuse of Subscriber Content, any unauthorised access or data breach, any negligence, wilful misconduct or violation of law by the vendor or its personnel, and any infringement of intellectual property rights. Clause 5.9 adds immediate termination, indemnification and injunctive relief without proof of special damages for misuse of Subscriber Content, and clause 12.1 a prorated refund. Insurance is not addressed anywhere. Verified 13 September 2026.

Practice Systems Integration Depth

How deeply the product reaches into the systems legal work already lives in: document management such as iManage and NetDocuments, Word and Outlook, contract lifecycle management, matter management, e-billing, and court filing systems.

Lexlegis.ai
CC on Practice Systems Integration DepthIntegrations are listed as logos or marked as coming, with no documentation an implementer could use.

Enterprise identity integration is specified in unusual detail and no legal practice system is named. What is established: the agreement defines the Service as including web, API and on-premise components made available under an order form, so a programmatic surface is contracted for rather than merely advertised; and the identity layer is published with named standards, being SAML 2.0, OIDC and SCIM 2.0 provisioning, with single sign-on required on all enterprise deployments and multi-factor enforced through the customer's own identity provider policies. On the isolated modes the platform runs inside the customer's own cloud estate using the customer's identity and access management and keys. Those are real integrations and they are integrations with the customer's security stack, not with the systems a legal team works in. What is absent is the substance of this axis. No document management system, case or matter management platform, word processor, email client or e-signature product is named on any surface read. No integrations page, connector directory or partner list was located in the site navigation. No API or developer documentation was located, so the API's existence is established from the agreement and its scope and depth are not. Nothing describes what data moves in which direction between the platform and any system a firm already runs. Terms of service, trust page, privacy policy and site navigation checked 7 September 2026.

Manupatra
CC on Practice Systems Integration DepthIntegrations are listed as logos or marked as coming, with no documentation an implementer could use.

Integrations are claimed and some host applications are named, with no documented catalog and no configuration detail, which is the C band. What is named is thin and scattered rather than gathered on an integrations page, and no integrations page exists in the navigation, which is itself a page-inventory finding. The research platform publishes a 'Plug In' item in both the Subscription and Research menus, resolving to ManupatraAlert, an alerting plug-in on the manupatrafast.com property. The privacy policy discloses that the products may integrate or offer third-party services including data visualization services and cloud storage solutions, naming Dropbox and Google Drive, for uploading, storing or sharing documents. The wider suite implies internal connection between the research platform and the Mykase practice and case management product, Manucomply and Manucontract, but each sells on its own domain with its own login and nothing published describes data moving between them. What is absent is everything the higher bands ask for. No practice management system, document management system or CRM used by Indian firms is named as supported, so a buyer cannot establish whether its own system connects. No public API reference, developer documentation, connector register or authentication model was located. No direction of flow, prerequisite or configuration step is described for any connection, including the plug-in. Nothing states whether AI Search or the ManuWorks toolkit can be reached from inside another application or only from the vendor's own interface. Verified 13 September 2026.

Deployment Model and Data Residency

Where the software runs and where the data sits. Multi tenant cloud, single tenant, private deployment, on premises, and whether region of residence is a published option or an enterprise conversation.

Lexlegis.ai
AA on Deployment Model and Data ResidencyDeployment options and data residency are published, including the regions available, what changes between tiers, and where processing happens as distinct from where data is stored.

Both limbs are published in full, per mode, and the record states what changes between tiers, which is the question this axis asks and which most records cannot answer. Five deployment modes are named and described. Mode A is multi-tenant software as a service hosted in India on ISO 27001 certified infrastructure, with matter level isolation stated for the multi-tenant case. Mode B is single tenant on the L&T Vyoma sovereign Indian cloud, with virtual private cloud isolation. Mode C runs inside the customer's own AWS, Azure or Google Cloud estate in a region of the customer's choosing, with the customer's own identity and access management and keys, virtual private cloud isolation and private link or private service connect. Mode D is on premises, including an NVIDIA DGX Spark unit on a desk or a full deployment in the customer's data center. Mode E is air gapped, with the vendor stating there is no inbound traffic at all on modes D and E. Residency is addressed as architecture rather than a setting: data location is stated per mode in the privacy policy, and residency can be locked to a chosen region on modes B, C, D and E. Customer managed keys with hardware security module storage are available on modes C, D and E. Cross-border transfers for customers outside India are stated to be governed by standard contractual clauses and equivalent mechanisms. What a buyer cannot establish is a residency choice on the entry SaaS mode, which is fixed in India, and no sub-processor list was located to test the mode statements against. Privacy policy, trust page and homepage read in full 7 September 2026.

Manupatra
BB on Deployment Model and Data ResidencyDeployment model is stated clearly with partial residency detail, or residency is offered without the processing location being addressed, or the tenancy model is stated on its own with no residency detail published.

Cloud delivery with a published residency commitment and no tenancy model stated, which is B because tenancy and region are co-equal limbs and publishing either clears C. Residency is answered in unusually committal terms for this corpus. The privacy policy carries a section headed 'Data Localization' stating that all personal information is stored and processed on servers located within the user's country of residence, that for users in jurisdictions requiring data localization the vendor ensures personal data is stored and processed on servers physically within that country's territorial boundaries, and that it does not transfer such data outside the country unless in accordance with applicable law and with appropriate safeguards. The About page describes the company as a cloud-first platform, and the AI brochure states that the models are trained on Indian data and built in India, which speaks to where the processing sits even though it is written as a provenance claim. The limits are real and are named rather than left to inference. The localization commitment is written about personal information, and the material this axis most concerns on an AI product is Subscriber Content uploaded to the AI tools, which the agreement governs on confidentiality terms without stating where it is processed; the two are not the same category and nothing joins them. No tenancy model is described anywhere, so a buyer cannot establish whether a firm's uploaded documents sit in a shared or isolated environment. No data center, region, availability zone or cloud provider is named, no on-premises or private deployment is offered or refused, and nothing distinguishes storage from processing for the AI features specifically. Verified 13 September 2026.

Security Certifications and Trust Center

Independent attestation a buyer can pull without a sales call: SOC 2, ISO 27001, penetration test summaries, a trust center with current reports and named scope rather than a badge image.

Lexlegis.ai
BB on Security Certifications and Trust CenterCertification is real and stated, short of accessible evidence: a named standard without scope, date, or a way to obtain the report.

A real trust center with a named artifact inventory, and the reports behind a request whose terms the portal does not state. What is ungated: six compliance positions in the trust center, being ISO/IEC 27001, ISO/IEC 20000-1:2018, SOC 2 Type 2, SOC 2, GDPR and CCPA, with the vendor's own trust page adding DPDP and stating the ISO certification as the 2022 revision; CERT-In empanelled audits, which is the Indian government-recognised assessor regime and a meaningful local signal; a risk profile giving data access level, impact level and a four hour recovery time objective; a control inventory spanning product security, data security, data privacy, access control and environmental, social and governance items; and a coordinated disclosure program with published response times of 24 hours to acknowledge and 72 to triage. What sits behind the access flow: the ISO/IEC 27001 certificate, the SOC 2 Type 2 report, a vulnerability assessment report, a penetration test report, a security compliance and deployment FAQ, and a document titled AI Flow for Ask. The portal offers to start a security review, to view and download sensitive information and to request access, and does not state whether access is instant on an email address, a self-service non-disclosure click-through, or approval after a sales conversation, so the lower tier is taken under the standing rule and that is why this is not the top grade, alongside the absence of any named auditor, audit period or scope statement. One discrepancy is recorded: the on-site trust page lists five frameworks including DPDP and omitting ISO 20000-1, while the trust center lists six including ISO 20000-1 and omitting DPDP. No request was submitted. Trust page and trust center read 7 September 2026.

Manupatra
DD on Security Certifications and Trust CenterNo independent security attestation located.

No certification and no trust center were located on any surface, which is the D band, and the absence is the vendor's rather than this index's. The page inventory was taken from the navigation and footer under R20 and there is no security page, no trust page, no compliance page and no certification badge anywhere on the estate: the footer's Company column runs About, Contact, Careers and Legal, and the Legal page's six tabs are Corporate, Copyright, Privacy Policy, Disclaimer, IT Compliance and Public Record. No ISO 27001, ISO 27701, SOC 2, StarAudit, CERT-In empanelment, PCI DSS or any other standard is claimed, no auditor is named, no report period is given, and no certificate, attestation or penetration test summary is offered for download or on request. A targeted search for certification claims returned nothing attributable to the vendor. What exists instead is a description of controls without external validation, and it is recorded so the grade is read as a disclosure finding rather than as a claim about the vendor's actual security: the privacy policy lists encryption technologies, secure servers and firewalls, access control mechanisms, and employee confidentiality agreements and training, stated to be regularly reviewed and updated, and the agreement commits at clause 4.4 to industry-standard technical and organisational measures. The statutory position is disclosed, being compliance with the Information Technology Act 2000, the SPDI Rules 2011 and the Digital Personal Data Protection Act 2023, with a named Grievance Redressal Officer, but a statutory compliance assertion is not an independent attestation and is not credited here. Verified 13 September 2026.

Model Supply Chain Disclosure

Which models sit underneath, whose they are, where they run, and whether the vendor commits to telling customers when that changes. A legal buyer inherits every dependency it cannot see.

Lexlegis.ai
CC on Model Supply Chain DisclosureThe vendor refers to advanced or proprietary models without identifying what sits underneath.

The architecture is described and nothing in it is named, and the one list the vendor says it publishes could not be found where it says it is. What is disclosed: the vendor states it built and trained its own model on its legal corpus and positions that against fine-tuning and retrieval-augmented generation, and its published model layer describes signed model checkpoints, per-tenant guardrails, prompt injection sanitization and an inference audit trail, all of which are the controls of an operator running its own inference. So a reader learns that the model is claimed to be the vendor's own, which is a real answer to part of the question. What is not disclosed is any name or any boundary. No model or model family is named. No third-party model, API or inference provider is named, and nothing states whether any is used at all, so a reader cannot tell whether customer content reaches an outside model on any mode. The infrastructure that is named is either the vendor's own hosting in India, the L&T Vyoma sovereign cloud, or the customer's own AWS, Azure or Google estate, and a customer's own cloud is not a supply chain disclosure by the vendor. Change notification is committed and is the strongest element here: the privacy policy states that customers are notified of material changes to the sub-processor list, with a right to object for enterprise customers. But the list itself is stated to be published at the trust center and does not appear on that page, so the commitment attaches to a document that could not be located. Privacy policy, terms, trust page and trust center read 7 September 2026.

Manupatra
CC on Model Supply Chain DisclosureThe vendor refers to advanced or proprietary models without identifying what sits underneath.

The supply chain is partly disclosed, which is the B band's territory on partiality but falls to C because what is disclosed is the technology class and the build location rather than anything about the models themselves. What the vendor does say is more than silence and is recorded precisely. The AI brochure and the December 2025 launch material both state that Manupatra AI Search is powered by semantic AI and large language models. The brochure states that the system is trained on Indian data and built in India, and that it runs on a proprietary, editorially enriched legal knowledge ecosystem developed over decades rather than on generic AI systems trained on public or non-curated data. That answers what the models were pointed at and where the work was done. It does not answer the questions this axis asks. No model is named, no version is given, and no provider is identified, so a buyer cannot establish whether the large language models are the vendor's own, licensed, or accessed as a third-party service, and the phrase 'built in India' describes the product rather than the model. Nothing states where inference runs. No commitment to notify subscribers of a model change was located. There is no subprocessor list that would answer the question by another route. One provision cuts against the silence and is noted because a careful reader will find it: clause 5.5 of the agreement bars the vendor from making Subscriber Content available to third parties including, expressly, AI vendors and cloud providers without prior express written consent, which contemplates that such vendors exist in the stack while naming none of them. Verified 13 September 2026.

Commercial Transparency

Whether a buyer can learn what this costs without entering a sales process: published rates, the unit being charged, what sits behind an enterprise tier, and what implementation adds.

Lexlegis.ai
AA on Commercial TransparencyA buyer can learn what this costs without entering a sales process: published rates, the unit being charged, and what implementation adds.

Published figures with a stated unit, a genuine free trial, and the commercial mechanics in the agreement rather than withheld. Two paid tiers carry figures on the vendor's own site: Professional from 9,000 rupees per user per month and Enterprise from 17,250. A seven day free trial of Legal AID is published on the site and confirmed at clause 4 of the terms, expressly requiring no payment instrument to start, with the vendor reserving the right to limit trial features and usage caps. Eligibility is stated rather than left implicit, clause 2 restricting paid plans to legal or compliance professionals or those in a role requiring legal research and drafting tools. The mechanics sit in the agreement: subscription basis, fees and billing frequency and currency set out in an order form or product plan, all fees exclusive of GST, VAT and equivalent taxes added at prevailing rates, and late payment attracting interest at 1.5 percent per month or the statutory rate if lower. Clause 9 publishes a service level with a figure, a 99.5 percent monthly uptime target on the SaaS mode. Clause 13 states the termination position and what happens to customer content on exit. Three gaps are named rather than smoothed: the four non-SaaS deployment modes and MIRA carry no published price and route to an order form; no usage quota, seat minimum or overage term is published for either paid tier; and trial usage caps are expressly discretionary. Homepage, pricing statements and terms of service read 7 September 2026.

Manupatra
AA on Commercial TransparencyA buyer can learn what this costs without entering a sales process: published rates, the unit being charged, and what implementation adds.

Figures, units, structure and the governing commercial terms are all published and a buyer can plan against them without a sales conversation, which is the A band. The AI Solutions page publishes an ungated price table. Unlimited AI Summary access is 5,000 rupees a year for Supreme Court judgments, 5,000 rupees a year for High Court judgments and 10,000 rupees a year for both, all stated exclusive of GST. AI wallet plans covering AI Search, AI Compare and AI Summary are published at 1,000 credits for 6,000 rupees, 1,500 credits for 9,000 rupees and 2,000 credits for 12,000 rupees. The unit of charge is not only stated but decomposed: the ManuWorks toolkit publishes per-feature consumption at 15 credits per page for OCR, 3 per page for summarization, 20 per page for translation, 5 per query for question-answering, 8 per page for timeline generation, 15 per document for comparison and 50 per query for AI drafting, with credit value fixed at 1 credit to 1.25 rupees, a 10,000 rupee minimum first purchase and a 5,000 rupee minimum top-up. Activation is stated to be pro-rated and a credit dashboard showing top-ups, usage and balance is described. So a buyer can model the cost of a specific workload in advance, which almost nothing else in this corpus permits. The commercial terms behind the price are published too, in the agreement: advance payment, non-refundability, renewal and deactivation mechanics, the vendor's right to modify charges with effect from the next renewal term, a seven-day window to dispute transactional charges, and taxes payable in addition. Two limits are named rather than glossed. The content subscription itself, as opposed to the AI add-ons, routes to a subscription plans page and the agreement states that the product price card is available on request, so the platform's own base price is less exposed than its AI pricing. And clause 3.9's transactional pricing statement implies a usage-metered tier whose rate card is not published. Verified 13 September 2026.

Firm and Practice Coverage

Who the product is actually built for. AmLaw, midlaw, small firm and solo, in house departments, government and courts, and which practice areas are supported rather than merely claimed.

Lexlegis.ai
BB on Firm and Practice CoverageSegment and practice coverage is described with substance, short of the boundaries: what is supported is clear, what is not is left open.

Practice scope is named with specificity and the corpus behind it is not inventoried. What is published: four practice domains addressed by name, being direct tax, indirect tax, corporate law and general laws, over a corpus described as a closed database of laws, circulars and judgments and characterized as India's largest repository of legal documents. Buyer types are named across the range this index cares about: general counsel, law firms and government bodies in the vendor's launch material, litigators and transactional and advisory counsel on the practitioner pages, and compliance professionals through the eligibility clause. Depth of capability is quantified where the vendor can quantify it, at 215 production skills composable into workflows. Currency of Indian law is addressed at least once in substance, the vendor distinguishing the Bharatiya Nyaya Sanhita from the Indian Penal Code as a failure mode of general-purpose assistants, which is a real coverage claim in a jurisdiction that has recently recodified its criminal law. What is absent is any inventory or boundary. No court or tribunal is named, no date range or historical start point is given, no depth statement distinguishes full text from headnote, and nothing states which parts of Indian law are thinly covered or out of scope. The United States is entered as a preview beta announced in March 2026 with no coverage description at all, so a reader cannot establish what the American corpus contains. Product pages, homepage, launch material and terms read 7 September 2026.

Manupatra
BB on Firm and Practice CoverageSegment and practice coverage is described with substance, short of the boundaries: what is supported is clear, what is not is left open.

Coverage is documented with real substance across jurisdiction, content type and buyer role, short of the stated limits the A band requires. Jurisdictional coverage is the strongest part and is specific: Indian case law from the Supreme Court, all High Courts and all tribunals, together with legislative, regulatory and procedural material, notifications, reports, commentaries and secondary sources across all Indian jurisdictions and Commonwealth countries, with a dedicated content coverage page, more than 10 million indexed documents, and daily addition of new laws, notifications and judgments. A separate Bangladesh database, BDlex, is sold alongside. Buyer roles are addressed explicitly and across the range: the AI pricing page names students, lawyers, law firms and corporate legal teams as the audiences for the credit model, and the About page enumerates legal professionals and individual practitioners, in-house legal departments in corporations and banks, academic and educational institutions, government bodies, regulatory agencies and policy makers, and law students, researchers and educators across Asia. Practice coverage is addressed through the content rather than through practice-area packaging, which suits a general research database, and the index's rule governs that limb: a horizontal case-law platform does not vary by practice area and is neither credited nor penalized for the absence of practice-specific editions. What holds it off A is that no limit is stated anywhere. Nothing identifies a jurisdiction, court level, tribunal or content type that is out of scope, nothing states the earliest coverage date or any gap in the historical record, nothing says which Commonwealth jurisdictions are included and which are not, and nothing states whether the AI features cover the whole corpus or only parts of it, which matters because AI Search is described in terms of judgments while the database also holds statutes, rules and commentary. Verified 13 September 2026.

The 12 legal signals, side by side

Recorded rather than graded. These are the questions a practitioner has to answer before a tool touches a client matter, and the answers are taken from public material only.

Client Data in Training

Can material a lawyer puts into this product be used to train a model?

Lexlegis.ai
Never, in the contract

The prohibition is in the agreement, and the vendor is the only record located so far that states the qualifier and then explains what it leaves open. Clause 5 of the terms of service v2.0, effective 1 January 2026 and last updated 19 April 2026, provides that the customer retains all rights in documents, queries and other content submitted, grants only a limited license to process that content solely to provide the Service, and states that Customer Content is not used to train shared models.

The privacy policy repeats it twice, saying content data is not used to train shared models and that the vendor does not use personal data to train shared models. The trust page states that these commitments are contractual rather than aspirational. The qualifier is the word shared, and unlike the other records carrying that shape this vendor addresses what sits outside it rather than leaving a reader to infer: fine tuning on customer data, if offered, happens only under a signed data processing addendum and only inside the customer's own deployment boundary.

That is a bounded carve-out with two conditions attached, and it is recorded here because it is the difference between a qualifier that hides an exception and one that defines it. Two structural facts reinforce the commitment rather than being counted twice: content retention is customer-configurable and reducible to zero on enterprise and the isolated modes, and on the on-premises and air-gapped modes the content never leaves the customer's estate at all. Terms, privacy policy and trust page read in full 7 September 2026.

Manupatra
Opt in

Training is barred by default and available only on the subscriber's express written consent, which is this value and the strongest instance of it in the corpus. Clause 5.2 of the published agreement, headed 'Absolute Prohibition on AI Training and Internal Use', bars the vendor from using Subscriber Content in whole, in part, anonymized, pseudonymised, aggregated, derived or otherwise for training, fine-tuning or improving any AI or machine learning model, for developing or enhancing algorithms, features or products, for internal analytics or database enrichment, for quality improvement, troubleshooting or research and development, or for teaching, demonstration or testing.

The anonymized and aggregated carve-out that has qualified most training clauses in this corpus is closed here rather than left open. Clause 5.7 requires that any contemplated use for training or improvement be authorized by prior express written consent obtained through a standalone written agreement, and states that silence or continued use of the service shall not constitute consent, which forecloses the deemed-consent route.

Clause 14.3.1 repeats the prohibition in the ownership section and clause 5.9 makes breach a material breach carrying indemnification and injunctive relief. The value is opt-in rather than contractual-never because the agreement does not commit never to train; it commits not to train without a separate signed permission the subscriber may withhold at its sole discretion. That follows the Agiloft precedent on a consent-governed no-training clause.

Recorded for completeness: the privacy policy states the same rule in weaker terms for user data generally, and R43(1) is discharged, the agreement having been located and read in full.

Prompt and Output Retention

How long does the product keep what a lawyer typed, and can that be set to zero?

Lexlegis.ai
Customer set, zero available

The customer sets the retention window and zero is reachable, with the tier at which it becomes reachable stated rather than hidden. The privacy policy separates three buckets and gives each its own rule. Content data, defined as documents uploaded, queries submitted and outputs generated, has its retention set by the customer, defaulting to the active life of the account plus ninety days for export. Usage data, being query logs, session identifiers, device metadata and telemetry, is retained for up to 24 months by default and is configurable on enterprise.

Account data is held for the relationship plus seven years, which the vendor attributes to Companies Act and Income Tax Act record-keeping rules rather than leaving it unexplained. The trust page states the floor: retention windows are configurable to zero on enterprise and on the on-premises and air-gapped modes. So zero exists, and the honest limitation is that it is an enterprise and isolated-mode capability rather than something available on the entry subscription, where the default window governs.

Exit is addressed alongside retention: all customer data is exportable at any time in standard formats, and on termination it is returned or destroyed at the customer's election with attestation, which clause 13 of the terms carries into the agreement. A data erasure control is listed separately in the trust center. Privacy policy, terms and trust page read in full 7 September 2026.

Manupatra
Disclosed fixed window

A fixed maximum retention period is published in the customer agreement, which is this value, and the period is the shortest located in this corpus. Clause 4.7, headed 'No Retention / Mandatory Deletion', requires the vendor to permanently delete and remove all Confidential Information uploaded by the client and any copies, backups, AI memory, logs, caches or traces immediately after fulfilling the specific request and in no event later than 24 hours after use, and to provide written certification of deletion on request.

The enumeration is what makes it gradeable rather than aspirational: AI memory, logs and caches are the places retention actually survives a deletion promise, and they are named. Clause 5.3 adds that the vendor shall not store, cache, retain, archive or otherwise preserve Subscriber Content beyond what is strictly necessary to generate the requested output, and that all storage must be temporary, encrypted and automatically deleted immediately after use.

Clause 5.6 extends the same treatment to AI outputs containing or derived from Subscriber Content, which are to be treated as the subscriber's confidential information and not stored, reused or made available to any other user or system. Two limits are recorded rather than smoothed. The 24-hour rule governs Subscriber Content uploaded to the service and does not by its terms reach the query text of a research search or the account and usage telemetry the privacy policy describes collecting.

And the privacy policy's own retention section is the opposite shape, committing only to keep personal information as long as necessary and directing readers to ask the data protection manager for periods; where the two differ on subscriber material the agreement governs.

Ethical Walls and Matter Segregation

Does retrieval respect the firm’s ethical walls, or can the model read across them?

Lexlegis.ai
Own model, documented

The vendor maintains its own segregation model, documents it, and states it at matter level rather than only at account level, which is rare on this signal. The data layer published on the trust page lists matter level isolation on multi-tenant software as a service as one of five data controls, alongside encryption, customer managed keys and hardware security module key storage. The privacy policy supports it from the other side, stating that content data is governed by strict access controls and is accessible only to the account it belongs to.

The identity layer adds the machinery a firm would need to operate a wall: role based access control with custom roles, single sign-on required on all enterprise deployments, multi-factor enforced through the customer's own identity provider policies, SCIM 2.0 provisioning, and session recording and replay on enterprise. The isolated deployment modes take the question further by removing multi-tenancy altogether. What is not published is how the isolation is administered by the customer.

Nothing states whether a matter boundary is created automatically or configured, whether a user can be screened from a specific matter, how a wall is applied to an existing matter, or what an administrator sees. Nothing inherits from a document or case management system, because none is integrated. So the model is the vendor's own, documented in outline, and a firm would have to keep it aligned with the walls it maintains elsewhere. Trust page and privacy policy read in full 7 September 2026.

Manupatra
Claimed, not documented

Access control is committed to in contract and no segregation model is described, which is this value, though the reason is unusual enough to state carefully. What is documented is vendor-side access rather than customer-side walls, and it is documented well: clause 4.2(d) restricts access to need-to-know personnel bound by equivalent confidentiality obligations, and clause 5.8, headed 'Monitoring / Human Review', bars any employee from viewing Subscriber Content or AI outputs except where necessary to resolve a specific technical issue the subscriber has raised, requiring any such access to be logged, minimized and subject to confidentiality obligations.

A logged-access commitment of that kind is rare in this corpus. There is also a real user-level access regime: clause 11.10 requires IP-based and remote-access subscribers to supply a list of named Authorized Users before the subscription starts, clause 3.5 makes the subscriber liable for all actions under its credentials and reserves termination for password sharing, clause 3.6 makes an organization responsible for its constituents, and clause 2.3 limits display to one person at a time on non-IP plans.

What is not addressed is the thing this signal names. Nothing describes segregation between matters, between clients, or between users within a subscribing firm, no permission model, role model or ethical wall is described, and nothing states whether documents one user uploads to the ManuWorks toolkit are reachable by a colleague on the same organisational subscription. The product class softens this only partly: a shared research database has no matter structure, but the AI toolkit takes uploaded case documents and that is precisely where a wall would matter.

Third Party Request and Subpoena Notice

If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?

Lexlegis.ai
Not addressed

No located public material addresses what happens when a third party demands customer data, and the omission stands out against how complete the rest of this estate is. The surfaces that would ordinarily carry it were read in full and none does. The terms of service run to fifteen clauses including a mutual confidentiality clause at clause 10, and that clause contains no compelled-disclosure carve-out and no notice obligation, which is unusual because the carve-out is standard drafting in exactly that position.

The privacy policy names legal obligations as a lawful basis for processing under DPDP and GDPR and describes sharing with sub-processors, but has no law enforcement or government request section, no statement of what the vendor does on receipt of an order, and no notice commitment. The trust center's open page lists a data privacy group containing cookies, data breach notifications and items, and no compelled disclosure item.

No transparency report or law enforcement guidelines page exists on any surface. So nothing is established either way: neither a commitment to tell the customer nor a statement that disclosure would occur without notice. Two adjacent facts are recorded because they bear on the practical exposure without answering the question: on the on-premises and air-gapped modes the vendor does not hold the content, and a data processing addendum is named in clause 15 as forming part of the agreement but is not published. Terms, privacy policy and trust surfaces checked 7 September 2026.

Manupatra
Disclosure addressed, notice absent

Compelled disclosure is addressed and customer notice is absent, which is this value. The privacy policy states that the vendor may be required to disclose personal data to third parties where required under applicable laws, judicial orders or governmental regulations, alongside transfer in a merger, acquisition, divestiture or dissolution, and disclosure to protect rights, property or safety. Nothing anywhere states that the subscriber would be told a demand had been received, given an opportunity to object or to seek relief, or informed after the fact, and no transparency report or law-enforcement guidelines page exists on the estate.

A genuine tension between the two instruments is recorded rather than resolved silently, because a reader who finds it unaided and does not find it here would trust the record less. The customer agreement is drafted more protectively than the policy and contains no compelled-disclosure carve-out at all: clause 4.3 states that the vendor shall not disclose or make available Confidential Information to any third party without the subscriber's prior express written consent, and clause 5.5 repeats it for Subscriber Content.

Read literally the agreement admits no exception for a court order, which cannot be the intended effect and which no clause reconciles. The agreement governs where the two conflict, but the agreement's silence on legal process is silence rather than a notice commitment, so it cannot lift this value. The practical position for a buyer is that the vendor has said what it may disclose under compulsion, in the weaker of its two instruments, and has said nothing at all about telling the customer.

Primary Law Corpus Provenance

Where does the law in this product come from, and does the vendor have the right to use it?

Lexlegis.ai
Sources named, basis unstated

The corpus is identified by type and the rights basis behind it is asserted only as to the vendor's own holding, with no cadence published. What is identified: a closed database of laws, circulars and judgments, described by the vendor as India's largest repository of legal documents, addressed to four named practice domains being direct tax, indirect tax, corporate law and general laws. The vendor distinguishes its corpus from open-web retrieval as a design position, and traces the collection to decades of legal documentation and knowledge-structuring work by named individuals in its founding lineage, which is more origin story than most records offer.

Clause 6 of the terms states that the corpus, along with the software, models, skill library and metadata schemas, is owned by the vendor or licensed to it. That is an assertion about the vendor's own title and it does not state the basis on which the underlying legal materials are used: nothing identifies the reporters, publishers or official sources the judgments and circulars come from, and nothing states whether Indian primary law is treated as government material outside copyright or licensed from a publisher.

Two further gaps are named. No update cadence is published anywhere, so a reader cannot establish how quickly a new judgment, circular or amendment reaches the corpus, which matters in a jurisdiction that has recently recodified core statutes. And no coverage inventory exists to test the size claim against. Terms, product pages and about page read 7 September 2026.

Manupatra
Sources named and licensed

Sources are identified and the licensing basis for third-party material is stated in the agreement, which is this value and only the second instance of it in this pull. Provenance is addressed in three places. Clause 10.2 of the agreement states that certain information and data made available are the property of content providers and are identified as such, and that the vendor has been licensed by those content providers to store, catalog and distribute the information to subscribers, with no rights transmitted to the user.

The Public Record section of the legal page states that public records are created and maintained by government agencies and open for public inspection, that all data in the public records databases is supplied by courts and government agencies, that the vendor collects public records from established sources in the government and judiciary, and that in some cases this occurs under agreement with those suppliers whose terms of use and information protection requirements may be more stringent than applicable law.

The copyright section adds that third-party content has been reproduced after taking prior permission from the party concerned, whose copyright is retained. Against that the AI brochure grounds the models on the vendor's own editorially enriched corpus, built through citations, judicial treatment and interpretive summaries developed over decades, which is owned rather than licensed material. So a buyer can establish the classes of source and the basis on which each is held.

What is not published is a supplier-by-supplier register, any license term or duration, or a statement of whether the licenses extend to AI processing of the licensed material as distinct from its distribution, which is the live question on a licensed corpus feeding a language model.

Good Law Verification

Does the product tell you when the authority it just cited has been overruled?

Lexlegis.ai
Not addressed

No located public material addresses whether authority is checked for subsequent history, and the verification the product does perform answers a different question. What the product verifies is existence and support: every response returns a list of the sources relied on, each openable and downloadable with a summary and a statement of why it was pertinent, and the reasoning chain traces every factual claim back to its source and flags a claim whose trace is broken.

That is credited on the Citation Accuracy axis and is not counted again here. It establishes that a cited judgment is real and says what the vendor claims it says. It does not establish that the judgment still stands. Nothing published states whether a decision has been overruled, reversed on appeal, stayed, distinguished, declared per incuriam or superseded by statute, and no citator is licensed, no treatment or noting-up signal is computed or surfaced, and no currency flag attaches to any authority in an output.

The vendor does show awareness of the underlying problem in a related form, contrasting general-purpose assistants that treat the Bharatiya Nyaya Sanhita as though it were the Indian Penal Code, which is a statutory currency point rather than a judicial treatment one, and no mechanism is described for either. In a jurisdiction where a judgment's standing turns on later benches, the absence is material and is recorded as an absence established on readable surfaces rather than a retrieval limit. Product pages, trust page and terms checked 7 September 2026.

Manupatra
Own treatment signal

The vendor operates its own treatment apparatus rather than licensing one, which is this value. The product page carries a section headed 'Be sure you are using Good Law' stating that with effective-date highlights and blue, yellow and red flag indicators the platform ensures every statute and judgment accessed is good law. A three-colour flag scheme applied to both statutes and judgments, with effective-date signaling, is a citator in substance and is the vendor's own: nothing indicates any part of it is licensed from another publisher, and the underlying treatment data is described in the AI brochure as part of the proprietary editorial enrichment built over decades through citations, judicial treatment and cross-references.

The AI layer is wired into it rather than bypassing it, which is the part that matters for this signal on an AI product. The brochure states that AI Search lets a user instantly determine whether a judgment has been overruled, distinguished or followed, identify conflicting judgments, and understand how courts have interpreted a statute over time, and the product page lists authority checks and citation maps among the built-in analytics.

Two limits are named. No methodology is published for how treatment is determined, whether editorially by the licensed advocates who enrich the corpus or algorithmically, and no coverage statement says which courts, which years or which content types carry flags, which matters on a database spanning all Indian tribunals and Commonwealth jurisdictions. And nothing states what a flag means in borderline cases or how quickly a new adverse judgment propagates to the flag.

Refusal and Uncertainty Behavior

What does the product do when the answer is not in the corpus?

Lexlegis.ai
Documented

What the system does when it cannot support a claim is stated, in terms, and without a demonstration behind it. The commitment is explicit: every factual claim in every output is traced back to its source, and if the trace is broken the claim is flagged, with the vendor adding no exceptions. That is a described behavior on failure rather than a description of grounding, which is why it is graded here while the grounding architecture itself, being corpus-first retrieval, IRAC output and the sources list, is credited on the Citation Accuracy axis.

Two further mechanisms sit alongside it. The published model layer names output hallucination detection and a meta reasoning gate through which every input and output passes, with the vendor stating that failure of any layer opens a ticket rather than a breach, so a detection event is treated as an operational signal that someone sees. And MIRA is described as verifying its own reasoning against a layer built specifically to catch its mistakes before it produces output.

What is missing is the demonstrable half. No evaluation, benchmark, worked example or published result shows the flagging behavior operating, no rate is given for how often a trace breaks, and nothing states what the user actually sees when a claim is flagged, whether the claim is withheld, marked, or returned with a caveat. A trust center item titled A note on Hallucinations was not opened on this channel and is the named rebuttal route. About page, trust page and product pages read 7 September 2026.

Manupatra
Not addressed

No located public material describes what the system does when it cannot produce a reliable answer, which is the floor, and the note records why that verdict sits alongside genuinely useful published guidance rather than against silence. The vendor publishes more about the boundaries of its AI than most records in this corpus. Its AI brochure carries a section headed 'Guidance on Using AI Search' with an express 'When NOT to Use AI Search' list, telling users to avoid AI Search for precise or literal queries, party names, case citations, exact statutory phrases and verbatim quotations, stating that it does not follow Boolean logic, and routing those queries to the named MANU SEARCH, LEGAL SEARCH and CITATION SEARCH modes.

A further page sets out how to frame a good query and contrasts well-framed and poorly framed examples. That is a published account of where the tool is weak, which is rare and is recorded as such. It is not this signal. Every one of those statements addresses what the user should do before running a query; none addresses what the system does after one. Nothing states that AI Search declines a query, flags low confidence, marks an uncertain result for verification, returns fewer than ten results when fewer are relevant, or tells the user it has found nothing responsive rather than returning the closest available matches.

On a retrieval product that always returns a ranked ten, whether the tenth result is relevant or merely nearest is exactly the question, and it is unanswered. The surfaces read on the date shown were the AI brochure in full, the product page, the AI solutions page, the AI Search FAQ page, whose four accordion bodies render client-side and did not return, the agreement and the legal page.

Fabricated Citation Record

Does a public court record exist addressing fabricated or hallucinated legal citations in output from this product?

Lexlegis.ai
None located

No court order, opinion or disciplinary record naming Lexlegis.ai or Lexlegis Solutions Private Limited was located as of 7 September 2026. Searches were run on the company and product names against Indian and international sanction and hallucination language and against the AI Hallucination Cases database maintained by Damien Charlotin. The Indian decisions located name no tool. This is a statement about the public record and not a finding about the product.

The environment is recorded because it bears on how the statement should be read rather than on the vendor: Indian courts are actively seized of the question, the Supreme Court having taken cognisance in a February 2026 order of a trial court that relied on non-existent AI-generated judgments, describing it as a matter of institutional concern bearing directly on the integrity of the adjudicatory process and indicating that such a decision would be misconduct with legal consequences to follow, and a separate bench having raised petitions drafted with AI citing non-existent authority.

A vendor selling into that market on an express hallucination-free claim is exposed to the question in a direct way, and nothing on the public record to date names it.

Manupatra
None located

Searched on 13 September 2026 against the company name and the AI product name, across reporting and trackers covering Indian decisions on AI-generated fabricated citations, including the Supreme Court of India's July 2026 ruling and the 2026 line of cases preceding it. None located. No decision, order or tribunal finding names Manupatra or Manupatra AI Search as the source of a fabricated citation. Context is recorded because it bears on how this absence should be read, and it cuts in the vendor's favor rather than against it.

India has the most active fabricated-citation record of any jurisdiction in this corpus: a Supreme Court bench set aside National Company Law Tribunal and appellate orders resting on non-existent citations, held that citing unverified AI-generated precedent is professional misconduct for an advocate, and directed the Bar Council of India to frame norms, with earlier instances including a Bengaluru tax tribunal order recalled over four non-existent citations and a Bombay High Court assessment quashed over three invented precedents.

In the commentary that followed, Manupatra is named repeatedly as one of the authorized databases against which a citation should be verified, which is the opposite posture to the one this signal records. This signal records fabricated legal citations in filings and nothing else, so the vendor's wider litigation or regulatory history, if any, would not appear here.

Bar Guidance Alignment

Has the vendor engaged in public with the ethics opinions its buyers are bound by?

Lexlegis.ai
Generic reference

Professional responsibility is engaged squarely and in the agreement, and no bar or court instrument is named. The engagement is more concrete than most records at this value. Clause 8 is headed AI outputs and professional responsibility and states that outputs are tools to support legal or compliance work rather than legal advice, that the customer remains responsible for reviewing and validating all outputs before relying on them, filing them or acting upon them, that no output creates a lawyer client relationship with the vendor, and that professional judgment is not replaced.

Clause 7 goes further than the usual disclaimer and imposes a positive prohibition aimed at the courtroom, that the customer will not use outputs to falsely attribute authority or to deceive any court or tribunal, which addresses the precise conduct that professional guidance on AI has been written to prevent. Clause 2 restricts paid plan eligibility to legal or compliance professionals or those employed in a role requiring legal research and drafting tools, so the obligation attaches to a gated population.

What is absent is any named authority. No rule of the Bar Council of India, no provision of the Advocates Act 1961, no state bar council circular and no court practice direction or standing order on AI-assisted filings is cited, mapped or linked anywhere on the surfaces read. That gap is sharper here than in most markets, because Indian courts have publicly begun addressing AI-generated authority and a vendor engaging the conduct question this directly might be expected to name the source of the duty. Terms of service read in full 7 September 2026.

Manupatra
Generic reference

Professional responsibility is engaged in general terms without any guidance being named, which is this value. What exists is contractual and real as far as it goes. Clause 16.1 of the agreement, headed 'No Legal Advice', states that material on or made available through the site is not intended to and does not constitute legal advice and does not in any manner establish a client-advocate relationship, and the disclaimer on the legal page repeats that the vendor is not creating a lawyer-client relationship by providing forms.

Clause 11.12 requires that the authenticity, correctness and preciseness of judgment text be verified from the certified copy, and clause 15.3 places responsibility for sufficient procedures and checkpoints on the user. Those are duties pointed at the right question. What is absent is any named authority. No bar body, conduct rule or ethics opinion of any jurisdiction is cited, the Bar Council of India is not mentioned, and nothing maps any AI feature to a professional obligation.

The gap is sharper on this record than on most and is stated plainly. In July 2026 the Supreme Court of India held that citing unverified AI-generated precedent is professional misconduct and directed the Bar Council of India to constitute a committee to frame norms, with draft regulations on AI use in courts out for consultation. The vendor publishes that ruling as editorial content on its own legal updates service.

It publishes nothing connecting its own AI Search, AI Summary or AI Drafter to the verification duty the same ruling imposes, at a moment when its buyers are being told by their highest court that the duty is theirs personally.

Billing and Fee Posture

Does the vendor address what happens to the bill when the work takes an hour instead of six?

Lexlegis.ai
Savings claims only

Time savings are marketed to a buyer who bills clients, and nothing addresses the client's side of the bill. The product sits inside a fee relationship for at least part of its buyer set: law firms and litigators are named buyers alongside general counsel and government bodies, and the work being compressed is legal research and drafting, which a firm bills. The savings claims are explicit and aimed at that work, promising research memos in hours rather than days, hearing preparation without the all-nighter, authority bundles built automatically, and answers in seconds.

Against that, no published material addresses billing, fee treatment or client disclosure. Nothing tells a firm how to treat research hours that collapse, no per-matter record of AI-assisted work is offered as a basis for a fee narrative, and no guidance on disclosing AI assistance to a client was located on any surface. Clause 3 of the terms governs only what the customer pays the vendor, covering subscription basis, order forms, tax exclusivity and late payment interest, and never reaches the customer's own invoicing.

The omission is worth naming because the vendor engages the lawyer's other professional obligations at length, gating eligibility to professionals and prohibiting the use of outputs to deceive a court, so this is not a general silence about the duties of the buyer. Practitioner pages, homepage and terms of service checked 7 September 2026.

Manupatra
Not addressed

Nothing published addresses what happens to the bill when AI-assisted work takes an hour instead of six, which is the floor, and all four of the higher values are false of this record rather than nearly true. The product sits squarely inside a lawyer-to-client fee relationship: the buyers named on the vendor's own pricing page include lawyers, law firms and corporate legal teams, and legal research is conventionally either billed as time or passed on as a disbursement in Indian practice.

Nothing states which. No per-matter record of AI-assisted work is described, nothing identifies output as AI-assisted for the purposes of a bill, no guidance on fee or disclosure treatment is published, and no saving is claimed in billable terms, the AI brochure's benefit language being about finding material that keyword search misses rather than about time recovered. Recorded and expressly not credited, because R21 and R24 govern and technology cost is a different object from AI-assisted work: the credit wallet is unusually granular about cost, publishing per-page and per-query credit rates for OCR, summarization, translation, question-answering, timeline generation, comparison and drafting, a fixed credit value, and a dashboard showing top-ups, usage and balance.

That machinery would let a firm attribute AI cost to a matter with more precision than almost anything else in this corpus, and clause 3.9 of the agreement provides a detailed statement of transactional charges for verification. None of it is presented as a client-billing or disclosure instrument, and the vendor never makes the connection. The gap is a genuine one and the summary carries it.

Outside Counsel Guideline Readiness

Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?

Lexlegis.ai
On request only

The artifacts a firm would forward are named and governed, and the one that is said to be published could not be found. What is committed: clause 7 of the privacy policy states that the vendor relies on a small set of sub-processors, that a current list is published at the trust center, that customers are notified of material changes to that list, and that enterprise customers have a right to object. A notification commitment with an objection right is stronger than most records carry.

Clause 15 of the terms names a Data Processing Addendum and a Security Schedule as forming part of the entire agreement, so both exist and are contractually incorporated, and the trust center carries a documents flow through which they and the certification reports can be requested. What is not available is the disclosure itself. The sub-processor list does not appear on the open view of the trust center the privacy policy names as its location, so the commitment to notify changes attaches to a document a prospective buyer cannot read.

No model provider is named anywhere, so a firm cannot tell its client whose model, if any beyond the vendor's own, processes matter content. The Data Processing Addendum and Security Schedule are named and not published. That places the forwardable material behind a request rather than in a firm's hands, which is what this value records. No access request was submitted. Privacy policy, terms and trust center read 7 September 2026.

Manupatra
Not addressed

None of the three artifacts this signal looks for exists, which is the floor, and the note explains why the floor understates what a firm could actually do with this vendor's paperwork. No subprocessor list is published anywhere. No model provider is named, the AI being described only as semantic AI and large language models trained on Indian data and built in India, so a firm asked which third party processes its uploaded documents could not answer from a list.

No consent or notification pack drafted to be forwarded to a client exists, and no data processing addendum is published as a separate instrument. Partial means some providers named across the product's AI, and none is named, so subprocessors-listed is unavailable; nothing is gated behind a request either, so on-request would be false. What a firm does have is a published, forwardable customer agreement that answers the underlying question in the negative and in strong terms.

Clause 5.5 bars the vendor from providing access to, disclosing or transferring Subscriber Content to any third party including affiliates, subcontractors, service providers, AI vendors, cloud providers or other users without prior express written consent, clause 4.3 says the same for Confidential Information, clause 5.2 bars training, and clause 4.7 requires deletion within 24 hours. On the question a client AI clause actually asks, that is a better answer than most subprocessor lists give.

It is not the artifact the signal names, and the value records the artifact. Recorded as the clearest instance in this pull of a vendor whose contract would satisfy a client where its disclosure estate would not.

Court Disclosure Support

If a judge’s standing order requires an AI disclosure, can the product produce one?

Lexlegis.ai
Partial record

Real elements of a record exist and none is framed or offered as something a lawyer could produce. The strongest element is specific to AI work rather than to data access, which is why it is graded here: the published model layer states an inference audit trail in which every call is logged and attributable. That is a record of what the model did and who invoked it, not a log of who opened a file, and it is materially different from the general audit logging controls that appear on most trust centers.

Around it sit further elements. Every output carries the sources it relied on, openable and downloadable with a statement of why each was pertinent, and the reasoning chain flags any claim whose trace to a source is broken; those are credited on the Citation Accuracy and uncertainty rows respectively and are named here only because a reader assembling a record would reach for them. Clause 7 of the terms engages the courtroom directly by prohibiting use of outputs to falsely attribute authority or deceive a tribunal, which shows the scenario was considered.

What is absent is the disclosure half. Nothing states that the inference audit trail is exportable, available to the customer at all rather than to the vendor, retained for any period, or presentable in a form a court would accept. No certification, template, declaration or guidance addressing a practice direction on AI-assisted filings is published, in a jurisdiction where courts have begun asking. Trust page, terms and product pages read 7 September 2026.

Manupatra
Not addressed

Nothing published helps a lawyer disclose AI involvement to a court, which is the floor, and on this record that absence is more consequential than on most. Two adjacent features exist and neither is a disclosure instrument, so both are recorded and not credited. The AI brochure states that the right panel provides an AI-generated analysis of the query which the user can download or share, and that individual results can be emailed, printed or downloaded.

That is an export of AI output, but nothing states that an exported analysis is labeled as machine-generated, carries a timestamp, records the query that produced it, or is distinguishable from a manually compiled note once it leaves the platform. Clause 11.12 of the agreement requires that judgment text be verified against the certified copy, which is a verification duty rather than a disclosure mechanism and speaks to the accuracy of the source rather than to the involvement of the model.

What is absent is everything the higher values describe: no audit trail of AI use, no per-matter or per-query record a firm could produce, no certification template, no export designed to evidence machine involvement, and no guidance on when or how AI assistance should be disclosed. The consequence is concrete here in a way it is not for most products. The Supreme Court of India held in July 2026 that citing unverified AI-generated precedent is professional misconduct and that a decision resting on hallucinated material is no decision in law, so an Indian advocate may shortly need to evidence how a cited authority was found.

This platform produces AI-generated case analysis and provides no means of showing afterwards which words were the model's.

Which one fits

Choose Lexlegis.ai if

  • You need the AI inside your own walls. Lexlegis.ai publishes five deployment modes, from multi tenant cloud in India through a single tenant sovereign Indian cloud and your own AWS, Azure or Google estate to on premises and air gapped installations, with region locking and customer managed keys above the entry mode.
  • Your procurement team wants reports. Lexlegis.ai states ISO 27001 and SOC 2 Type 2 with audits by government empaneled assessors, and its trust center offers the certificate, the SOC 2 report and a penetration test report through a request flow.
  • You want an agent for multi step legal work. Lexlegis.ai's MIRA selects from 215 production skills, orders them into a plan and checks its reasoning before output, and every factual claim in an output is traced to a source, with broken traces flagged.

Choose Manupatra if

  • You need to know whether a judgment is still good law. Manupatra marks statutes and judgments with its own blue, yellow and red flags and effective date highlights, and its AI Search can show whether a judgment has been overruled, distinguished or followed.
  • You want strict handling of uploaded case files in the contract. Manupatra's agreement bars training on subscriber content, including anonymized or aggregated use, without standalone written consent, requires deletion within 24 hours including AI memory and logs, and indemnifies you for confidentiality breaches.
  • You want to price AI by the task. Manupatra publishes AI Summary at ₹5,000 a year per court level and a credit rate card for its toolkit, from 3 credits a page for summaries to 50 a query for drafting, at ₹1.25 a credit.

In summary

Lexlegis.ai

Lexlegis.ai, from Lexlegis Solutions Private Limited of Mumbai, is an Indian legal AI platform built on a closed corpus of Indian laws, circulars and judgments. Legal AID answers questions on direct tax, indirect tax, corporate and general law with sources, reads uploaded documents and drafts submissions, and MIRA runs multi step work from 215 skills. The AI Legal Index grades it in the top two bands on twelve of fifteen capability axes, with A grades on AI centrality, data stewardship, deployment and pricing. It offers five deployment modes up to air gapped, states ISO 27001 and SOC 2 Type 2, prices from ₹9,000 per user a month, and names KPMG and the Government of India among customers. As of 7 September 2026 the index located no accuracy measurement.

Source: AI Legal Index, 2026

Manupatra

Manupatra, from Manupatra Information Solutions Pvt. Ltd., founded in 2000 and based in New Delhi and Noida, is India's oldest online legal research platform, covering the Supreme Court, all High Courts and tribunals across more than 10 million documents, with its own good law flags. Its AI Search, launched in December 2025, retrieves judgments by concept, alongside AI Summary, AI Compare and the ManuWorks toolkit. The AI Legal Index grades it in the top two bands on ten of fifteen capability axes, with an A on pricing. Its agreement bars training without written consent and requires deletion within 24 hours. As of 13 September 2026 the index located no named customer, security certification or accuracy measurement.

Source: AI Legal Index, 2026

Questions buyers ask

Lexlegis.ai vs Manupatra: which is better for Indian legal research?

Lexlegis.ai sits in the top two bands on twelve of fifteen AI Legal Index capability axes and Manupatra on ten of fifteen, identical on ten. Lexlegis.ai leads on deployment options, security reports and named customers. Manupatra brings a 25 year old corpus with its own good law flags and stricter contract terms on uploaded material. Researchers who need to check whether a judgment still stands have more to use in Manupatra.

Does either tool check whether a judgment is still good law?

Manupatra does. It marks statutes and judgments with blue, yellow and red flags and effective date highlights, and its AI Search can show whether a judgment was overruled, distinguished or followed. Lexlegis.ai traces each claim to its source and flags broken traces, which confirms a source exists, but it publishes no check on later treatment. Graded by AI Legal Index against 15 capability axes and 12 legal signals, including privilege handling and citation accuracy, from each vendor's own published materials, verified September 27, 2026. No vendor pays for placement.

Can Lexlegis.ai run on premises?

Yes. Lexlegis.ai publishes five deployment modes: multi tenant cloud in India, single tenant on an Indian sovereign cloud, inside the customer's own AWS, Azure or Google estate, on premises, and air gapped with no inbound traffic. Region locking and customer managed keys are available above the entry mode. Manupatra is delivered as a cloud service and describes no tenancy model. Graded by AI Legal Index against 15 capability axes and 12 legal signals, including privilege handling and citation accuracy, from each vendor's own published materials, verified September 27, 2026. No vendor pays for placement.

How much do Lexlegis.ai and Manupatra cost?

Lexlegis.ai publishes Professional from ₹9,000 and Enterprise from ₹17,250 per user per month, with a seven day free trial. Manupatra publishes its AI add ons: AI Summary at ₹5,000 a year per court level, credit packs from ₹6,000 for 1,000 credits, and a per action rate card for its toolkit. Manupatra's base research subscription price is available on request. Graded by AI Legal Index against 15 capability axes and 12 legal signals, including privilege handling and citation accuracy, from each vendor's own published materials, verified September 27, 2026. No vendor pays for placement.

What do Lexlegis.ai and Manupatra both leave unpublished?

A measurement of their accuracy and a named model. Both claim their AI avoids hallucinations and neither publishes an error rate or evaluation. Neither names the model provider behind its AI or an integration with a legal practice system, and neither names Bar Council of India guidance on AI use. Graded by AI Legal Index against 15 capability axes and 12 legal signals, including privilege handling and citation accuracy, from each vendor's own published materials, verified September 27, 2026. No vendor pays for placement.

Disclosure

Three readings to weigh. Both vendors claim their AI is free of hallucinations and neither publishes an accuracy or error measurement. Lexlegis.ai's commitment is not to train shared models, with fine tuning on customer data possible under a signed addendum inside the customer's own deployment; its subprocessor list could not be found where its policy places it. Manupatra claims no security certification, and its privacy policy permits disclosure under legal compulsion without customer notice. Lexlegis.ai was verified on 7 September 2026 and Manupatra on 13 September 2026. Neither vendor reviewed this page.

Neither vendor paid for inclusion, placement or a grade, and neither reviewed this page before it published. Everything above comes from public material on the dates shown. How the index grades.

Contact

Correct a record, or ask how something was graded

Every grade and every signal on this index is drawn from public sources and dated. If a record is wrong, out of date, or missing an artifact the index did not locate, send the source and it will be reviewed and the record redated. Vendors are welcome to submit documentation. Nothing on this index is for sale, including a listing, a placement, or a grade.

AI Legal Index

The AI Legal Index is an independent index that tracks changes to AI vendors in legal. It holds 303 vendors across 9 categories, each graded on the same 15 capability axes and recorded against 12 legal signals, from public sources. No vendor pays for inclusion, placement, or rating.

Index Status
Last index update
September 26, 2026
The AI Legal Index is an editorial reference. It is not a regulatory body, not a law firm, and nothing published here is legal advice or a recommendation to retain or avoid a vendor. Records are verified against published sources, bar guidance and public court records. Where a record reads not addressed, the material was not located in public sources on the date shown. See the Methodology page for evaluation standards and limitations.
© 2026 AI Legal Index
3801 N Capital of Texas Hwy, Ste E240 · Austin, TX 78746