Jhana.ai

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.

The models are the product and the corpus was built to feed them. The vendor describes itself as a frontier legal intelligence lab and names its proprietary work as verifier models, graph models for legal ontologies, a national legal archive of more than 16 million machine enhanced documents, and research and drafting agents. Machine enhanced is the operative phrase: the corpus was not licensed and wrapped, it was processed into a model asset, and the stated use of seed capital was to build proprietary datasets and models and to hire researchers in law and AI. Every product is an agent pipeline rather than a workflow tool with a model attached, including the public sector agents running extraction, synopsis and forensic scrutiny. Remove the models and there is no product left. Same shape as Reveal, reached from the opposite direction: Reveal acquired model capability into a platform, Jhana built the corpus and the models first and the interface after.

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.

Grounding is architectural and demonstrated, short of any published measurement. The vendor states plainly that its agents always cite their work, and names a verifier model as proprietary technology, which puts a checking mechanism in the architecture rather than in the marketing. Published product material shows source linked output where each conclusion sits one click from the underlying source page and each extracted figure is anchored to page and paragraph, which is a demonstrable behaviour a reader can inspect rather than a claim about accuracy. Held at B because nothing measured is published: no accuracy figure, no precision or recall on retrieval or extraction, no hallucination rate, no test set, no independent benchmark participation, and no statement of what the verifier catches or how often. Searched the home page, the product pages, the published blog material, the privacy policy and the terms and conditions on 29 Aug 2026. For a vendor whose named innovation is a verifier, the absence of a measured verification result is the gap.

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.

The positioning implies an oversight model and no document describes one. Calling the product a paralegal rather than a lawyer is a deliberate and meaningful choice: a paralegal works under supervision, and the naming carries that. But naming is not a published model. Nothing states where a human must review agent output before it is used, whether any step can complete without review, what the agents may do unattended in the Courtroom and PUBSEC pipelines running extraction, synopsis and forensic scrutiny for courts, or what happens when an agent is wrong inside a registry workflow. No confidence threshold, escalation path or human confirmation requirement was located. Checked the home page, the product pages, the public sector material, the terms and conditions and the blog on 29 Aug 2026. The gap matters more here than for a research tool sold only to private practice, because the deployed users include judges and registrars.

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.

Adoption is quantified and unusually institutional. Stated: more than 10,000 users, over 150 judges and registrars across five or more courts, more than 100,000 paralegal sessions and over 200,000 searches. Judicial adoption is the strongest element and the second instance on this index of a public authority using the product rather than merely appearing as a logo, after FinregE and the FCA Handbook. Also dated and attributable: a $1.6m seed led by Together Fund with named participants, a January 2026 partnership with CADRE ODR for online arbitration, and support from Jio GenNext, the AWS Public Sector Startup Hub, Microsoft Founders Hub and the Google Cloud Startups Program. Held at B on three gaps. Every usage figure is self reported with no methodology, period or definition, so a paralegal session is whatever the vendor counts. No court, institution or firm is named. No outcome measure of any kind is published: nothing on time saved, error reduction or research quality, only volume.

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.

Real documentation exists and none of it is about privilege, and one disclosed practice cuts against the posture. The privacy policy is a genuine product document framed against the Digital Personal Data Protection Act 2023 and the Information Technology Act 2000, with erasure on request by email, retention carve outs for fraud prevention and for tax, legal reporting and audit obligations, and a security section. That is more than several better funded vendors on this index publish. What is absent: any treatment of legal professional privilege or work product, any statement about the confidentiality of case files uploaded to AI Paralegal or Document Intelligence, and any segregation model. Named as a concern rather than buried: the privacy policy states that personal information may be disclosed to data processors including marketing and advertising agencies and web analytics companies. For a platform holding client matter documents, disclosure to advertising and analytics categories is a real posture question, and the policy does not distinguish account level personal data from uploaded matter content.

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.

Not addressed in any located material. The product generates propositions, advisories and memos from natural language input, which is output shaped like advice, and the user base explicitly extends beyond advocates to founders, tax professionals and compliance teams. That combination is precisely where the question bites: a non lawyer receiving an advisory generated from case law is the unauthorised practice scenario, and nothing published addresses it. No statement that output is not legal advice, no positioning on the supervising advocate's role, and no engagement with Bar Council of India rules on advocate conduct or advertising. Checked the home page, the product pages, the terms and conditions, the privacy policy and the blog on 29 Aug 2026. The paralegal naming is the closest thing to a position and it is a brand choice rather than a disclosure.

AI Governance and Bias Disclosure

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

Nothing published about how the models are governed, evaluated or monitored. No AI policy, no evaluation methodology, no bias or fairness assessment, no accuracy monitoring, no drift statement, no model card, no named governance body and no external standard such as ISO 42001. The absence is conspicuous against the vendor's own subject matter expertise, since its published blog material analyses the DPDP Act and DPDP Rules compliance obligations in structured detail for its customers, and none of that analytic rigour is turned on its own system. Also unaddressed: whether a corpus described as machine enhanced introduces systematic error into the archive itself, which for a research product built on a proprietary dataset is the governance question that matters most. Checked the home page, the company and funding material, the product pages, the blog and the terms and conditions on 29 Aug 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.

The vendor speaks to stewardship directly, which most of this roster does not, and what it says does not hold together. The home page states that customer data is not used for training, and in the next sentence that free users have the choice to opt out of sharing data for model training. Those two statements cannot both be complete. The privacy policy supports erasure on request, discloses processor categories, and states that appropriate security measures and generally accepted industry standards are applied. Held at C because the operative rule is unclear from public material: no retention period is stated for prompts or generated output, no distinction is drawn between account data and uploaded matter content, and the training position depends on which sentence a reader stops at. The signal row records the contradiction with the quote. Credited above D because a vendor that addresses training at all, and gives free users a control, is materially ahead of one that is silent.

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.

Terms exist, were read, and run one way. The terms and conditions contain an indemnity under which the user holds the company and its officers, agents and employees harmless against third party claims, demands, damages, penalties, losses and actions including reasonable attorneys' fees, together with an entire agreement clause incorporating the privacy policy. No corresponding vendor commitment was located: no warranty as to output, no accuracy undertaking, no service level, no liability position, and no remedy where a generated citation, advisory or memo is wrong. This is graded on published material rather than on absence, which is why the basis is Vendor Published: the document exists and allocates risk in one direction. Checked the terms and conditions and the privacy policy on 29 Aug 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.

Interoperability is claimed at the right level for the public sector product and documented nowhere. The vendor states that Courtroom and PUBSEC are interoperable with existing and new systems and are transacted with over API, which is the correct shape for selling into court registries, and the January 2026 CADRE ODR partnership is a named integration into an online arbitration workflow. What does not exist: any API documentation, any authentication or scope detail, any statement of what moves in which direction, and any named integration for private practice. No document management system, no practice management platform, and nothing addressing how a firm gets its own matter files in and its work product out. Checked the home page, the public sector material, the product pages and the blog on 29 Aug 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.

One genuinely strong deployment disclosure sits beside a complete absence for everyone else. For the public sector products the vendor states that the agents are owned and hosted by the courts themselves, which is customer hosted deployment stated plainly and is the strongest such position located in this pull. It is also the correct answer for a judiciary that cannot put case data on a startup's infrastructure. For the commercial product nothing equivalent exists: no hosting provider is named, no region or data residency commitment is published, and no private or single tenant option is described for firms. Cloud vendor programme memberships with AWS, Microsoft and Google are startup support programmes and are not a hosting disclosure, and are not credited as one. Data residency is a live question in India under the DPDP framework the vendor itself writes about. Checked the home page, the public sector material, the privacy policy and the terms and conditions on 29 Aug 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.

No certification of any kind was located, and the security language is the unfalsifiable shape this index does not credit. The privacy policy states that appropriate security measures are taken against unauthorised access, alteration, modification, disclosure or destruction, and that generally accepted industry standards are followed. Neither statement names a standard, an auditor, a scope or a date, and generally accepted industry standards is a phrase that cannot be checked or falsified by any reader. Located nothing on SOC 2 of either type, ISO 27001, ISO 42001, CERT-In empanelment or any Indian assurance framework, and no trust centre, security page or documentation request route exists. Checked the privacy policy, the terms and conditions, the home page, the product pages and the site footer on 29 Aug 2026. Under the three tier test the artifact is absent rather than gated. Weighed against the deployment position: a vendor whose court customers host the software themselves has offloaded the hardest part of this question for those customers, and published nothing for the rest.

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.

Processor categories are disclosed, a list is promised, and no entity is named. The privacy policy states that personal information may be disclosed to entities that may qualify as data processors under the DPDP Act, listing categories including hosting service providers, cloud computing entities, IT service firms, marketing and advertising agencies and web analytics companies, and states that the vendor will make efforts to disclose an updated list of key processors. A promise to publish is not a publication, and no such list was located. On the model layer specifically: the vendor claims proprietary verifier and graph models, which is a supply chain statement of a kind, and never says whether any third party foundation model sits underneath the agents, which is the question a buyer needs answered. Compare Onspring at B, the only record on this index that names its model provider outright.

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.

A free tier is quantified and the paid ladder is behind a login. Published free tier limits are specific: ten paralegal sessions, fifteen searches and thirty file document intelligence per month, which tells a buyer both the unit of consumption and the shape of the meter, and is more structural disclosure than most of this roster offers. Beyond that, plan details require signing in, and no price, currency or range for any paid tier was located. Held at C rather than higher because the headline is visible and the actual cost is not, and a buyer cannot compare against an incumbent subscription without creating an account. Source basis recorded as Third Party Estimated: the free tier limits come from an independent review platform's product assessment dated July 2026 rather than from a vendor pricing page read directly, and that assessment itself notes pricing is not fully public and may be out of date.

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.

Corpus scale and jurisdiction are stated precisely and the vendor is candid that the scope is one country. More than 16 million judgments and statutes in a corpus described as India's National Legal Archive, with the explicit positioning that this is purpose built for the Indian legal context rather than a general model with a legal skin. Named user populations span private practice lawyers, law firms, in house teams, judges and registrars across five or more courts, founders, tax professionals and compliance teams, and published subject matter reaches SEBI regulations, AML and CFT, the DPDP Act and Rules, the IT Act and Supreme Court authority on Section 65B evidence certification. Single jurisdiction scope is a limitation for a buyer outside India and is not a disclosure failure. Held at B rather than A because the corpus is not broken down: no list of which courts, tribunals and forums are included, no statement of historical depth, and no update lag or refresh frequency, so a practitioner cannot confirm their own forum is covered or how current it is.