Leah vs Luminance: how they compare in 2026
This is a genuine head to head. Both sit in contract review and drafting. Both are enterprise platforms covering the contract from intake through negotiation to post execution analysis, both are sold well beyond the legal department, and both are headquartered in the United Kingdom. They are strong in opposite places. Luminance owns more of its own stack: it builds and hosts its own legal model, gives every customer a dedicated single tenant instance and offers deployment inside the customer's own environment. It also publishes a benchmark, ContractIQ Bench. Leah publishes the paperwork. It sets a tiered liability cap with figures and an enumerated list of uncapped claims. It gives a warranty with a thirty day fix period and published uptime tiers with refund. And it names its subprocessors, with thirty days notice and an objection right. The sharpest difference between them is training. Leah states that customer contract data is never used to train models. Luminance has stated in its own press material that its model learns from every NDA or supplier agreement negotiated within the platform. For a contract platform, that is the question to settle first.
Which one fits
Choose Leah if
- You need the allocation of loss on paper before you sign. Leah's master terms set a general cap at the fees paid in the twelve months before the first incident, and an enhanced cap at three times that figure where the security clause or the data processing addendum is breached. They enumerate uncapped claims covering indemnification, intellectual property and confidentiality. They add a customer indemnity, a warranty with a thirty day correction period and published uptime tiers of 99.00, 99.5 and 99.9 percent with termination and refund. That matters here because Luminance's product negotiates with counterparties on the customer's behalf.
- The training question has to be answered no. Leah's security FAQ states that customer contract data is never used to train models, and its governance page commits to zero retention with its named providers so that data is processed but never stored by them. Luminance's own press material states that its model learns from the negotiations conducted inside the platform. Luminance trains its own model on a corpus it puts at 220 million legal documents.
- You want to know who is in the chain and to hear about it when it changes. Leah's data processing addendum names Anthropic, OpenAI, Cohere and Google AI/ML with the jurisdictions for each, requires a new subprocessor to be added to the published list at least thirty days before it processes customer personal data, and gives a thirty day objection right with termination and refund of prepaid fees. Luminance names AWS and its own Luna model.
Choose Luminance if
- Isolation and location are hard requirements. Every Luminance customer receives a dedicated single tenant instance with complete isolation and no co mingling of data, and deployment inside the customer's own environment is offered as an alternative to the hosted option. Backups run nightly to a secondary AWS data center in the same region, encrypted and kept a minimum of fourteen days. Vendor staff cannot view customer documents without explicit authorization granted through the interface, with all access tracked and audited. Leah's default is shared, with single tenancy available for customers with strict isolation requirements.
- You want an accuracy claim you can at least argue with. Luminance publishes ContractIQ Bench, a benchmark of 189,000 manually annotated and reviewed data points assessing interpretation of named provision types including liability caps, termination for convenience and confidentiality obligations. It is tested on held out documents and concepts excluded from training, with blind evaluation by legal experts alongside. Luminance states that the model is trained to prioritize faithful extraction and to identify absence rather than invent an answer. Leah asserts that every action is measured against benchmarks for accuracy, bias and outcome.
- The buyer is not only the legal team. Luminance publishes dedicated pages for six industries and eight business functions including compliance, procurement, finance, human resources and marketing, and it serves both law firms and corporate departments. It is consistent throughout about where its boundary sits: contract work end to end rather than litigation or research. Leah reaches law firms only indirectly, through managed service partners.
At a glance
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.
Leah sells AI agents and an orchestration layer that sit on top of a contract lifecycle platform, and that platform works without them. The vendor describes it the other way round. It says other vendors bolted AI onto systems built for manual workflows, while Leah was designed from scratch with orchestration as the foundation. ContractPod Technologies has sold contract lifecycle management since 2012. Leah launched in March 2023 as an AI services hub within that platform, went standalone in May 2023, and Leah Intelligence followed in October 2024. Without the agents, the product is still a working CLM with guided intake, approval routing, DocuSign and Adobe Sign execution and a contract repository. That CLM has its own market and its own Gartner category placement. The orchestration layer on top is model driven.
Luminance uses models throughout contract interpretation, negotiation and portfolio analysis. Luna Crescent specializes in extraction, classification and other repetitive legal tasks, while additional models handle work that benefits from broader generative capabilities. A routing layer selects the model for each task, and separate validation models test important answers. The repository connects agreements and amendments into context for later questions, and negotiation uses prior concessions and approved positions to generate markups. Workflows turn those findings into contract actions across legal and business teams. The central service depends on model generated understanding of contractual language and relationships, rather than a template library alone.
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.
The Contract Helpdesk answers questions from the customer's agreements and links each response to the clause or document supplying the answer. Its illustrated payment example identifies the agreement version and clause number, while Leah Discovery extends retrieval across a portfolio. Users can open the underlying contract and initiate a reminder, amendment or escalation from the answer. This is a described route from a question to its contractual source. However, the product's categorical claim that Leah does not hallucinate is not supported by a published test set or error breakdown. The AI terms expressly disclaim accuracy, completeness and currency, and linked contract text does not validate external legal authority.
Luna Crescent is tested against ContractIQ Bench, with 189,000 manually annotated data points covering provisions such as liability caps, termination rights and confidentiality. Testing includes unseen concepts and blind review by legal specialists. Luminance claims 5 percent higher accuracy than leading general models without an absolute result or named comparators. Separate validation checks important outputs against model agreement, source evidence and task requirements. A failed check can lead to a retry and a visible request for human review. The model is also trained to identify missing information rather than fill the gap. Public sample tasks, a reproducible evaluation set and detailed error rates are not available.
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.
Customer policies determine which Leah agents may act, the data they may access, thresholds, and escalation requirements. The orchestrator applies approvals, escalations, and rejections during execution, then records the action, rationale, policy, inputs, and outcome. Policies are configurable rather than fixed defaults. The AI terms add a categorical limit: automated decisions with legal or similarly significant effects on individuals require adequate human review and legal compliance. Customers must also check outputs themselves and cannot present them as provider approved or wholly human generated. Trials and betas are restricted to internal evaluation. These contractual limits accompany the product's policy, execution, and audit loop.
Business users can generate and progress routine agreements under legal approved templates and role permissions, with nonstandard terms routed to legal. Proposed negotiation language appears in Word. For important AI outputs, a separate model or panel checks agreement, supporting evidence and task requirements. Every required critic check must pass; a failed check prompts another attempt using the feedback, followed by a review flag if the problem persists. Customer decisions and workflow actions retain an audit trail. The agreement requires manual interpretation of all results, and the AI terms require interfaces that support human oversight of the system's operation and limitations.
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.
Leah publishes qualitative quotes from four named people. Noelle Perkins is EVP and Chief Legal Officer at Cushman and Wakefield, and Lidia Kamleh is Chief Legal Officer at Dubai Future Foundation. Frances Bain-Cumberbatch is Chief Legal and External Affairs Officer at Ansa McAL, and Zillia Knight is Senior Legal Officer at Terumo Europe. Three results are published with the customer unnamed. A major American logistics company cut contract review time by 91 percent. A global manufacturer protected more than $18 million of revenue, and an American retail REIT tracked more than $2 million of savings. About 54 enterprise logos appear, including Philips, MUFG, Sandoz, Pernod Ricard, Alaska Airlines and Wood PLC. PwC and KPMG appear among them. PwC entered a commercial alliance in March 2024, and Epiq resells Leah in its Service Cloud. Integreon is quoted as an early adopter that resells it, and Pinsent Masons adopted it for managed legal services in July 2025. Partners and customers are shown together without distinction, and the Chief Product Officer of Execo, another services partner, is among the testimonials.
Current customer examples tie specific organizations to contracting work. AMD reports finding precedent language in under two minutes, Emma Sleep reports twice as many contracts reviewed per quarter, and Avianca claims 90 percent less document review time. Repository examples include Yokogawa's 99 percent faster response to business questions and Trench Group's two hour tariff analysis, compared with two weeks and about €11,000. Workflow examples include ProSapient reducing eight steps to three and Hitachi moving from NDA generation to signature in under five minutes. These are vendor presented customer results, without a common study protocol or dated measurement period. The current customer base exceeds 1,000 enterprises.
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.
Customer inputs and outputs are confidential Customer Data under the May 2026 agreement. Their use is limited to providing and maintaining the service, with a separate AI clause permitting training or improvement solely for that customer's benefit. The trust portal rules out generalized model training, while the DPA identifies four model providers that store or retain no Customer Data. Role permissions and a private Azure OpenAI environment support access isolation. Mutual confidentiality obligations and prior notice of legally compelled disclosure apply. The terms do not expressly preserve privilege or work product, and Leah describes no matter-specific retrieval walls inside a customer's organization.
Each customer has an isolated environment with separate workloads, data and layered access controls. The agreement treats private learning and outputs as customer property and confidential information, and limits support access to service delivery. It also licenses customer data for product improvements, private learning and usage insights; there is no general prohibition on learning from customer content. Private learning stays with that customer, while nonidentifying usage insights can be retained for improvement. Inputs sent to external models are permitted disclosures directed by the customer. No specific clause protects attorney client privilege or work product, and the model list gives no provider retention limits.
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.
Attachment D requires independent review of every AI output for its intended use and prohibits representing output as provider approved or wholly human generated. Automated decisions with legal or similarly significant effects require adequate human review and compliance with law. The platform nevertheless promotes autonomous work across legal, procurement, finance, and contracting functions. These general AI use restrictions establish customer responsibilities, but do not explicitly distinguish legal advice from software assistance, restrict use by jurisdiction, or map professional competence duties. The broad legal automation positioning therefore has a contractual review requirement without a dedicated professional responsibility explanation for practicing lawyers.
The master agreement prohibits treating the product as legal advice or a substitute for relevant independent or internal legal advice. All results require manual interpretation, and customers must tell third party users about those limits. AI Systems Terms add human oversight duties and describe the intended use as legal document analysis, research, contract analytics and legal workflows. The product also serves procurement, finance, sales and other business functions, with exceptions routed to legal. Luminance sets out no lawyer competence procedures and no boundaries for supported jurisdictions. The license permits professional services firms to use the software when serving their own clients.
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.
A dedicated AI governance page names six failure modes the vendor says it engineered out. They include black box decisions that cannot be defended to a regulator or board, and compliance frameworks retrofitted after the fact. Against them the page sets three pillars and a loop of policy, execution and audit. Each action is logged with its rationale and governing policy, in records described as tamper resistant and immutable. The page also says every action is measured against benchmarks for accuracy, bias and outcome, and that accountability is structural rather than aspirational. It names no person or role accountable for model behavior and describes no testing before release. It gives no benchmark method or schedule and discloses no bias measurement result.
AI Systems Terms require ongoing risk management, records of risks and tests, and data governance that considers bias, representativeness and missing data. The model evaluation process includes task benchmarks, complete workflow testing through ALumNI, blind expert review and a human sign off before a reversible production rollout. Luna Crescent releases must clear ContractIQ Bench and expert review before replacing a current version. These controls extend beyond the information security program and address model behavior before release. Luminance names no individually accountable AI governance owner and gives no results on uneven performance across matter types, parties or populations.
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 DPA names model, hosting, support, signature, and integration subprocessors with their locations and purposes. New subprocessors require 30 days' notice and permit an objection, with termination and unused-fee refunds if unresolved. Security incidents affecting personal data require notice within 48 hours of awareness. The trust portal describes role restrictions, identity-provider authentication, monitoring, incident exercises, and penetration testing. Contract data can be exported during the term and deleted following a request after termination, subject to backup and legal exceptions. Leah gives no routine duration for prompt history. Provider-level zero retention does not define the lifetime of records held by Leah itself.
Customers can delete data during use, with removal from active stores within 30 days and backups within a further 30 days; faster deletion can be requested. Routine backups occur every 24 hours, remain within the hosting region and are kept for 14 to 28 days under the master agreement. Access follows least privilege, with multifactor authentication, optional single sign on and isolated customer workloads. The subprocessor list names hosting, model, support and operations services. New providers require 30 days notice and allow 15 days to object. The data agreement requires notice of a personal data breach within 24 hours, mitigation and information about the affected records and likely consequences.
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.
Infringement by AI output is excluded from the provider's intellectual property indemnity under the May 2026 Cloud Terms. Attachment D also disclaims output accuracy, completeness and currency, and puts the job of checking results on customers. Ordinary cloud service remedies remain: a performance warranty, correction or termination with a refund after an unsuccessful 30 day fix period, and an indemnity for covered service infringement claims. The general liability ceiling is fees paid or payable in the preceding twelve months, with a threefold ceiling for security or DPA breaches and specified uncapped claims. Trials and betas have no indemnity or warranty and a $1,000 cap. These software remedies do not remove the explicit AI output exclusions.
The October 2026 master agreement provides an IP indemnity for the product, with exclusions for unauthorized combinations, continued infringing use and unsupported versions. A performance warranty allows correction, replacement or a refund of unused prepaid fees. General liability is capped at the previous 12 months' fees, rising to three times that amount for specified data obligations. Confidentiality and IP ownership breaches are outside those caps, and certain legally nonexcludable losses remain protected. Claims require notice within nine months. The agreement separately disclaims losses caused by output errors or omissions and treats customer directed model transmissions as permitted disclosures. No insurance commitment is specified.
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.
Contract workflows connect with Salesforce, Microsoft Dynamics and HubSpot opportunities, while an Outlook add in starts contracts, uploads attachments and triggers approvals that synchronize with the CLM. Signature connections let customers configure signing order, copied recipients and expiry dates, then return executed agreements to the repository. Word supports review and tracked changes. The current CLM page also lists SharePoint, OneDrive, Box, Google Drive, Dropbox and NetDocuments, replacing the earlier absence of a named legal document system. ERP connections include SAP, Oracle, Workday and Coupa. Leah does not document field mappings, sync direction or permission inheritance for each connector, least of all for the document systems.
Negotiation generates context based markups directly in Microsoft Word using approved language, earlier concessions and comparable agreements. The workflow product connects contract actions, tasks and approvals across business teams and lets other assistants access contract capabilities through MCP. Users can initiate analysis, summaries and reviews from connected applications. The platform also names Outlook, DocuSign and LexisNexis connections. Its technical terms allow uploads from document management systems and virtual data rooms. Connector specific configuration, permission inheritance and the exact fields moving in each direction are not explained. These connections bring AI contract work into existing applications while retaining the central agreement record.
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.
Shared cloud delivery is standard, with single tenant deployment and a private Azure OpenAI environment available for additional isolation. The May 2026 subprocessor schedule adds region-specific hosting arrangements across Microsoft Azure, Google Cloud, and Oracle Cloud. Azure lists Europe, the United States, and Australia; Oracle also includes India, while Google lists Japan and several European jurisdictions. Model providers have their own location lists, selected according to the customer's hosting region where applicable. The schedule separates hosting from model services, but gives no complete per-customer map of storage, inference, backups, and support access across each product configuration.
The standard hosted product runs in an isolated customer environment on AWS. Customers choose an available AWS data center through the order form, with backups in the same region and recovery objectives of 24 hours for a complete primary data center failure. Additional inference can run on AWS, Azure or Google infrastructure. Azure and Google processing follows an EU or UK hosting choice; for other hosting choices the processing sits in an international location that Luminance does not name. Direct OpenAI API processing takes place in the United States. Luminance publishes neither the full list of available hosting regions nor the routing of each model task, and a storage choice alone does not fix every processing location.
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.
A-LIGN's ISO/IEC 27001:2022 certificate covers the information security management system for design, development, operation, maintenance, and support of Leah CLM and Leah Legal as SaaS. Certificate ISMS-CO-032326 was issued March 23, 2026 and expires March 23, 2029. The trust portal provides the certificate directly, alongside a SOC 3 report covering the contract management and Copilot systems from October 2024 through September 2025. That report addresses security, availability, confidentiality, and privacy, with Azure and customer control dependencies excluded from the auditor's examination. Penetration-test material is also available. These attestations concern system controls, not the correctness of an AI legal recommendation.
ISO 27001:2022 covers the information security management system for development, sales, support and operation, including the Cambridge technical headquarters, remote workers and cloud processing. SOC 2 Type II examinations cover security, availability and confidentiality, alongside independent penetration tests. Contractual measures include encrypted storage and transport, nightly operating system updates, daily backups, least privilege access and a dedicated data protection officer. Senior leaders review the security program quarterly, and a security advisory board supports it. Audit information and certificates are available through the contractual review process. The current assurance statements leave out the auditor names, report dates and examination periods.
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.
The May 2026 DPA lists Anthropic, OpenAI, Cohere, and Google AI/ML for Leah functionality, each with a no-storage or no-retention statement and defined processing jurisdictions. Azure supports hosting, translation, and Office services; Oracle and Google add hosting options. Zuva handles document intelligence, and ABBYY is limited to older CLM deployments. The platform routes tasks among language models and allows customer customization. New subprocessors require 30 days' notice, with an objection and termination route. Leah does not give model versions or say which provider produced a particular generated passage. The regional provider list describes available processing arrangements rather than a complete execution record.
Luna Crescent is a family of proprietary legal models developed from a general foundation model and specialist contract training. It works alongside other models selected by task through benchmarks and expert review. The subprocessor register identifies AWS, Microsoft Azure and Google Cloud for inference inside cloud environments, plus direct OpenAI API processing in the United States. The AI terms distinguish models hosted within infrastructure from requests sent to an external model provider. New subprocessors require advance notice, but Luminance does not name the individual foundation models, their versions or the task routing. No equivalent notice commitment covers every model substitution within an existing provider.
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.
The cloud service uses negotiated orders with a default 12-month subscription, automatic renewal, and 30-day nonrenewal notice. Standard, Gold, and Platinum support plans set different support hours, response targets, and availability commitments. Professional services require an order or statement of work, with possible travel expenses and separately quoted configuration work. AI access is described as unlimited subject to fair use, including restrictions on excessive compute, API calls, storage, and transfer. No subscription amount, seat rate, or implementation price is stated. For applicable European switching arrangements, an early termination charge can include the remaining subscription fees despite the right to transfer data.
Subscription allowances measure the number of files held in the product, and each version counts as a separate document. Annual uploads are subject to a fair use ceiling of twice the contracted document allowance, regardless of later deletion. Fees, payment timing and the initial term are set in the order form rather than a public rate table. Renewal ordinarily repeats the initial term unless either party gives 90 days notice. The standard increase ceiling is the annual Consumer Prices Index change plus three percentage points, unless the order form says otherwise. Luminance gives no prices for assisted implementation or extra modules.
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.
Leah publishes dedicated industry pages for CPG and manufacturing, energy and utilities, financial services, healthcare, and pharma and medical devices. It describes its customers as Fortune 500 enterprises in regulated industries. By function it publishes pages for legal leadership, legal operations, sales and revenue, procurement, and finance. The pages carry distinct propositions written for the General Counsel, the contract operations team, the Chief Procurement Officer and the finance leader. The customer roster spans banking, airlines, pharmaceuticals, consumer goods and engineering. No published page says which practice areas, contract types or matters the platform does not support, and none addresses smaller organizations. Law firms appear only indirectly, through managed service partners, rather than as a served segment.
Legal, procurement, finance, sales, executive and HR teams use negotiation, repository analysis and workflows across commercial agreements. The license also permits professional services organizations to use the product for their clients and includes terms for United States government customers. The legal suite defines Corporate, Diligence and Discovery applications, while the AI terms describe legal document analysis, research, analytics and related legal workflows. Examples span NDA creation, supplier review, liability positions, tariffs and renewal obligations. Luminance does not set out firm size segments, supported research jurisdictions, language limits or excluded practice areas. Current product positioning extends across an enterprise's contract work.
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?
Attachment D permits use of inputs and outputs to train or improve AI features solely for the customer's benefit. The same agreement treats both as Customer Data. The trust portal excludes generalized model training, and the DPA says Anthropic, OpenAI, Cohere, and Google store or retain no Customer Data for Leah functionality. Private customer improvement is therefore permitted even though generalized training and retention by those model providers are restricted.
The master agreement licenses customer data for product improvements, private learning, usage insights and outputs. Private learning means recognizing new clauses and contract types from customer training; it belongs to the customer, remains confidential and is not shared across customers. Nonidentifying usage insights may be retained indefinitely for improvement. No general training prohibition overrides those rights. Inputs used with additional model features may also be transmitted to the named provider at the customer's direction.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
Inputs and outputs are Customer Data under the AI terms. Export is available during the subscription, and deletion follows a request after termination within 60 days, with backup and legal exceptions. The DPA permits access, return, or deletion of personal data during service but sets no routine prompt history duration. Zero retention at named model providers does not define how long Leah keeps its own conversation, output, or audit records.
Customers can delete data at any time. Active data stores remove it within 30 days, with backup removal within a further 30 days and expedited deletion available on request. The agreement also provides a 30 day period for retrieving and deleting data after the term ends. Routine backups have a separate 14 to 28 day retention period. No setting guarantees zero retention across application records, logs and external model processing.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
Leah describes role permissions, restrictions on the data agents may access, single tenant deployment and a private Azure OpenAI environment. Leah claims these controls separate customer information and limit access. Its trust portal also describes least privilege and reviews of staff access. The materials do not explain how retrieval enforces a barrier between separate legal matters inside one customer organization, or how permissions inherited from an external document system constrain an agent's answer. Corporate tenant isolation should not be assumed to provide a law firm's matter conflict controls.
Each customer instance isolates workloads and data through logical, technical and access controls. Administrators configure roles, multifactor authentication and single sign on, while support access is limited to service delivery and may follow additional customer policies. The product applies its own access controls around the contract workspace. No current commitment describes inheriting an external document system's permissions at retrieval time or enforcing a law firm's matter level ethical walls.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
Section 19 permits legally required disclosure of confidential information, including Customer Data. Where law permits, the recipient must give reasonable advance notice and cooperate at the discloser's expense to seek confidential treatment. Inputs and outputs are covered as Customer Data under the AI attachment. The obligation therefore reaches customer AI material as well as uploaded contracts, without a separate transparency report or history of government demands accompanying it.
The confidentiality clause requires notice of legally required disclosure where lawful, giving the customer a reasonable opportunity to respond. Luminance must cooperate with reasonable efforts to resist, limit or delay disclosure at the customer's expense. A separate clause also requires prompt notice and cooperation when responding to a subpoena or judicial, administrative or arbitral order. No aggregate transparency report is specified.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
Answers and agent guidance draw on the customer's contracts, internal policies and playbooks. Those materials are an internal reference base, not an identified external legal corpus. The vendor also refers to established legal precedents, without naming a case law database, jurisdiction, publisher or licensing arrangement. No update schedule accompanies that reference. Source links to a customer's agreement can show where a commercial term came from, but say nothing about the provenance or rights basis of any external legal authority used in an answer.
Contract answers draw on customer agreements, amendments, negotiation history and the platform's standard legal concepts and clauses. Luna Crescent separately learns from more than 3.4 million proprietary legal data points, with an evaluation set of 189,000 annotated points. The customer retains its data and private learning. The sources and rights behind the general training collection, and any primary law collection used for legal research, are not identified individually or accompanied by an update schedule.
Good Law Verification
Does the product tell you when the authority it just cited has been overruled?
No citator, treatment signal or currency check is described, and nothing says whether legal authority is reviewed for later history. The platform manages contracts, obligations and procurement workflows rather than retrieving case law, so a citator is not part of what it sells. The vendor does refer to Leah operating against established legal precedents, without identifying any source. That is the one place the product invokes primary authority, and no verification step is described for it.
Negotiation and repository answers are grounded in contracts, amendments and earlier agreed positions. The AI terms also include legal research among the product's intended functions. No named commercial citator, subsequent history calculation or treatment flag accompanies that research description. Source grounded validation checks whether a response follows the supplied evidence, which is separate from checking whether a legal authority remains valid.
Refusal and Uncertainty Behavior
What does the product do when the answer is not in the corpus?
The home page and the AI governance page describe no explicit path for Leah to decline to answer or abstain, and no confidence or grounding rating. The governance loop does produce rejections. Approvals, escalations and rejections are applied automatically according to the customer's rules. Those are policy outcomes set by configured guardrails, not the model declining because it cannot ground a response. Neither page says what Leah does when the customer's own contract set or playbook does not cover the question in front of it.
Luna Crescent is trained to identify information that is absent from the contract rather than invent it. Differential Validation separately checks important responses through independent model agreement or a source grounded critic. A failed check can trigger a retry using the identified problem; continued failure results in a review flag. The current explanation describes those behaviors without a public evaluation of their refusal rate or missed unsupported answers.
Fabricated Citation Record
Does a public court record exist addressing fabricated or hallucinated legal citations in output from this product?
Neither Leah nor ContractPodAi appears in the AI Hallucination Cases database maintained by Damien Charlotin. Its entries concern court decisions involving hallucinated content, with tool names recorded where available. Leah's principal workflows cover commercial contracting, procurement, and related legal work rather than preparation of court submissions. That product setting helps explain the limited relevance of litigation records, but a missing entry is not a measure of the accuracy of its contract answers or redlines.
The AI Hallucination Cases database contains no court or disciplinary entry naming Luminance as the source of fabricated legal citations as of October 15, 2026. Luminance's principal workflows concern contracts, but its AI terms also include legal research. The AI Systems Terms require human oversight, and the master agreement assigns the customer responsibility for manually interpreting and using output, including checking its relevance to the particular legal task.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
Leah publishes nothing that engages with bar or ethics guidance. That includes ABA Formal Opinion 512, state bar guidance in the United States, and Solicitors Regulation Authority or Law Society material. The company is headquartered in London and sells into legal departments across North America, Europe, Asia and Australia. Its published compliance material covers regulation and security frameworks, namely GDPR, CCPA, HIPAA, SOC and ISO. None of it addresses the professional conduct obligations that bind the lawyers using the product.
Professional services customers can identify mandatory requirements imposed on technology providers by their governing body before the order is signed. The master agreement commits Luminance to meet those requirements when compatible with applicable law, and permits regulatory access and assistance in specified circumstances. All output still requires manual interpretation. The agreement does not map these commitments to a named bar opinion or a particular lawyer ethics rule.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
Leah serves corporate legal, contracting, procurement, and finance teams, and also supports legal services delivered to clients. Pinsent Masons Vario adopted the platform for its Managed Legal Services and Consulting work, including contracting support for customers' internal legal departments. Corporate efficiency figures concern operating cost, while the firm's use can affect client service delivery. The platform records agent actions and policy decisions, but gives no specific method for adjusting client fees or disclosing saved lawyer time.
Luminance licenses its product both for internal business work and for services delivered to clients. Current customer examples include faster precedent retrieval, shorter review and fewer workflow steps, while corporate deployments also report operational savings. The contract permits customers to earn revenue from the resulting output. No guidance connects the time saved to a law firm's client invoice, and the workflow audit trail is not presented as a legal time billing record.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
Attachment B includes the DPA, processing details, transfer mechanisms, and a subprocessor list by purpose, location, and product. Model providers are Anthropic, OpenAI, Cohere, and Google, alongside hosting, signatures, support, and integration suppliers. Codestone, Freshworks, Oracle, and Totango appear in the May 2026 list. New providers require 30 days' notice and permit objections. The AI terms separately disclose customer-specific training permission and require independent output review.
The current list identifies AWS, Azure, Google Cloud and OpenAI for hosting or inference, plus named support and operations providers. It distinguishes processing inside cloud infrastructure from inputs sent directly to OpenAI and states broad processing locations. The data agreement requires 30 days advance notice of a provider change, with 15 days for objections. No separate client consent or outside counsel AI disclosure template accompanies the legal suite.
Court Disclosure Support
If a judge’s standing order requires an AI disclosure, can the product produce one?
The audit stage of Leah's published governance loop logs every decision. Each entry records what the agent did, why, under which policy, with what data and with what outcome. The records are described as tamper resistant, immutable and ready for any audit. That gives the action, the rule, the inputs and the result for each action. The published description of the log does not include the model. The platform chooses among several language models for each task and identifies no model or version, so the log does not show which system produced a given passage.
No export built for court disclosure or AI use certification is described. The audit framing is regulatory and internal rather than judicial.
Workflows keep an audit trail of actions and decisions, and the AI terms require proportionate event logs retained for the agreement's duration. Important responses can carry a validation result or a request for review, with contract evidence behind the answer. The legal agreement also provides assistance to regulators. Luminance describes no per document export that combines the exact model used, the retrieved sources and a person's verification.
The questions both sides leave open
Derived from the records above rather than written, so it cannot favor either vendor. Take these into both conversations and ask each side the same question.
- Good Law Verification
In summary
Leah
Leah is an agentic platform for contracting, legal, procurement and finance at large enterprises, sold by ContractPod Technologies of London and formerly known as ContractPodAi. It runs multiple large language models selected dynamically per task. Its data processing addendum names Anthropic, OpenAI, Cohere and Google AI/ML as providers under zero retention terms. Around them sits a published governance loop: the customer defines which agents may act, on which data and within which thresholds, and every action is logged with its rationale, the governing policy and the outcome. According to the AI Legal Index, Leah's strongest published material is that oversight model and its allocation of loss. The master terms publish a general cap at twelve months of fees, an enhanced cap at three times that figure and an enumerated list of uncapped claims. It is sold on a custom quote.
Luminance
Luminance is a contract lifecycle platform covering generation, negotiation, analysis, compliance and investigation. It was founded in Cambridge in 2015 by mathematicians and is sold to legal, compliance, executive, sales, procurement, finance, human resources and marketing functions across more than 700 organizations in over 70 countries. It runs a multi model architecture it calls a Panel of Judges, in which several models analyze each clause independently and reach consensus. In June 2026 it introduced Luna Crescent, its own legal intelligence model deployed inside its own AWS environment. According to the AI Legal Index, Luminance publishes a benchmark, ContractIQ Bench, built on 189,000 manually annotated and reviewed data points and tested on held out documents. Its subscription is quoted, with renewal increases capped in the agreement.
Questions buyers ask
Are Leah and Luminance alternatives to each other?
Yes, directly. Both are filed under contract review and drafting, both are enterprise platforms covering the contract from intake through negotiation to post execution analysis, both are sold to procurement and finance alongside legal, and both are headquartered in the United Kingdom. They diverge on architecture rather than on scope. Luminance builds and hosts its own legal model and gives each customer a single tenant instance. Leah orchestrates four named third party providers underneath a policy layer the customer configures. From the AI Legal Index, based on each vendor's own published materials as of September 14, 2026. No vendor pays for placement.
Does Luminance train its model on customer contracts?
Its own published material says the model learns from work done inside the platform. A press release of 17 February 2022 states that the AI learns from every NDA or supplier agreement negotiated within Luminance, every clause that causes an M&A transaction to fall apart and every piece of data culled during ediscovery. The Ask Lumi release of May 2023 repeats that its legal model learns solely from legally verified documents, then put at more than 150 million. Those statements date from 2022 and 2023. Leah, by contrast, states that customer contract data is never used to train models. From the AI Legal Index, based on each vendor's own published materials as of September 14, 2026. No vendor pays for placement.
How do Leah and Luminance handle liability?
Leah's master terms set a general cap at the amounts paid in the twelve months before the first incident, an enhanced cap at three times that figure for breach of the security clause or the data processing addendum, and list uncapped claims covering indemnification, intellectual property and confidentiality. It adds a customer indemnity, a warranty carrying a thirty day fix period and published uptime tiers with termination and refund. Luminance publishes its terms and conditions openly. From the AI Legal Index, based on each vendor's own published materials as of September 14, 2026. No vendor pays for placement.
Which one can be deployed inside our own environment?
Luminance. It offers deployment within the customer's own environment as an alternative to the hosted option, and every hosted customer receives a dedicated single tenant AWS instance with complete isolation and no co mingling of data. Leah's default is shared, with single tenant deployment available for customers with strict isolation requirements and a separate dedicated private environment offered within Azure OpenAI Studio. From the AI Legal Index, based on each vendor's own published materials as of September 14, 2026. No vendor pays for placement.
Neither vendor paid for inclusion, placement or a grade, and neither reviewed this page before it published. Everything on it comes from public material on the dates shown. How the index grades.
Three things to weigh. First, both vendors assert measurement and only one publishes a number, and that number is thinner than it looks. Luminance reports 5 percent higher accuracy than leading general purpose models on contract understanding, with no absolute rate and no comparator named, so a reader cannot tell what it is 5 percent better than. The benchmark is proprietary, with no sample tasks or rubric published. Leah states that every agent action is measured against benchmarks for accuracy, bias and outcome and publishes nothing behind it. Second, the training record on Luminance rests on dated vendor statements: a press release of 17 February 2022 and the Ask Lumi release of May 2023. Its current pages describe the corpus without repeating the sentence about learning from customer negotiations, which reads as age rather than withdrawal. Third, the documents themselves have limits. The Leah terms quoted on this page are version 3.0c of the master terms, and the current version is 4.0. Luminance publishes no integrations page, trust portal or downloadable report, which is missing documentation rather than a stated refusal. Neither vendor reviewed this page.