CaseMine vs Descrybe: how they compare in 2026
CaseMine and Descrybe both sell AI case law research at the accessible end of the market, and both publish what they charge. They differ in what the AI is actually doing. CaseMine puts a retrieval layer over a case law repository it has built since 2013, covering India, the United States and the United Kingdom court by court. It adds its own citator and states ISO 9001:2015 and ISO/IEC 27001:2022 certification. On Descrybe the corpus is itself model output: judicial opinions summarized by AI, 10,969,112 of them across 492 courts and tribunals on the vendor's July 2026 coverage page. A citator produces issue level treatment flags. A connector layer exposes citation lookup, quote verification and treatment review to Claude, ChatGPT and Perplexity, so the research tools reach into the assistant a lawyer already uses rather than requiring another workspace. Both publish coverage court by court with the limits stated. CaseMine covers three jurisdictions, and Descrybe covers one at a finer grain, down to a date range per court. On compliance they part as certification against disclosure. CaseMine states ISO certification and says little about data handling. Descrybe holds no certification of its own and publishes a security page naming its model provider, its hosting region, its training exclusion and its retention and deletion practice.
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.
The models drive everything that is sold, and a real content product sits beneath them. CaseMine says keyword and citation search are free without an account, while AMICUS, CaseIQ, summaries, argument generation, drafting and contract analysis need a subscription. Without the AI, a searchable case law repository across three jurisdictions remains, a product this market has bought for decades and CaseMine has built since 2013. The AI layer is substantial and is what the buyer pays for: retrieval augmented generation over that repository, an extraction engine that takes whole documents as queries, and search by intent.
The content itself is model output, which is as central as AI can be. The product was built by AI summarizing judicial opinions, now 10,969,112 opinions across 492 courts and tribunals per the July 2026 coverage page, and that summarization is how the database was built, not a feature on top of it. On top sit DescrybeLM, a purpose built legal reasoning engine; Cytator, an AI citator giving treatment analysis by issue; and an engine layer offering retrieval and verification tools to outside assistants. Without the models there would be no summaries, no citator and no product, only public domain opinions the company never had rights to sell.
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 real and described as a mechanism. CaseMine says AMICUS uses retrieval augmented generation to pull case law, statutes and principles from its database before answering, and contrasts this with general tools, saying it ensures the assistant does not just sound convincing. Answers are linked to source authorities, put on the home page as cited authority out with the source kept visible. CaseIQ is described as extractive rather than generative. No accuracy figure, test set, evaluation or benchmark result appears. Trade press covering a new AMICUS model in March 2026 reports an internal benchmarking framework covering research, document analysis, drafting, issue spotting and complex reasoning, but that is a vendor claim relayed by a third party, with no results and no vendor page describing it.
Grounding is well structured, without a published measurement. DescrybeLM answers carry a table of authorities linked to primary law, the reasoning is shown at each step, and quote verification and citation resolution are built in as tools. The vendor also discloses against its own interest: the terms say Descrybe does not review or validate outputs for accuracy, truthfulness, relevance or reliability, or to confirm they are not misleading, and earlier material warned users to check summaries against the original opinion. No accuracy figure of any kind is published: no summary fidelity rate, citation resolution accuracy, hallucination rate or test set. A claim in third party material that DescrybeLM beats named frontier models on bar exam benchmarks was not found in vendor material with any method or results, and is not relied on.
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.
Verification is offered as the control and nothing describes an oversight model around it. The published position is that answers carry linked source authority and that the source stays visible, so a user can open what the system relied on, and the vendor frames this as what makes responses verifiable and dependable. That is a real review affordance for a research product. Beyond it, nothing. No published material states what any tool does without a human, where a review point sits, what happens when the system is uncertain, or what a user is expected to check before relying on a draft. This matters because AMICUS does not stop at research: CaseMine states it drafts contracts, memos, motions, petitions and transactional documents, and can include citations to judgments in those drafts. A product that writes filings with citations in them and publishes no oversight model is asserting verifiability without describing where verification happens.
Oversight is described as product behavior rather than as policy, which is the useful form. DescrybeLM is documented to clarify jurisdiction, posture and key facts with the user before proceeding, which puts a human checkpoint at the start of the task rather than only at the end, and it is documented to tell the user when it needs more information and to flag uncertainty rather than filling gaps with guesses. Reasoning is visible at each step and work product is exported deliberately to Word rather than acted on. In the engine configuration the user can see which Descrybe tools an external assistant called and read the data each returned, which is oversight of a tool chain and not merely of an answer. None of it is quantified or bounded: no statement of when the system proceeds without clarification, no confidence threshold, and no description of what the uncertainty flag actually triggers.
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.
An unattributed volume claim stands in for evidence. CaseMine's own competitor comparison page says the platform is used by thousands of US lawyers every day, with no named customer, figure, date or case study attached. The FAQ, pricing page, CaseIQ page, privacy policy, about page, bespoke AI solutions page and comparison page name no customer, quote no identified user, and tie no measured outcome to a named firm. The one institutional relationship publicized is a legal technology certificate program with an Indian law school, which is a teaching partnership, not a deployment. The site blocks automated readers, so the home page could not be read directly and may say more.
Institutional adoption claims are specific, but none was found in vendor material. Third party research says Descrybe was added to the NSLT curriculum in place of Casetext and approved as a NELLCO electronic resource reaching nearly 150 law libraries. Those would be strong evidence if the vendor published them, since library consortiums and curricula evaluate research tools carefully, but as found they are third party claims. Dated and checkable: coverage by a named legal technology journalist across four product stages from 2024 to 2026, an ABA Woman of Legal Tech award for cofounder Kara Peterson in 2024, and an Anthem Award for Responsible Technology. No customer, deployment, usage figure or outcome measure published by the vendor was found, and the coverage milestones describe the vendor's own build, not customer results. The source basis is recorded as third party estimated.
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.
One pathway is addressed with real precision and the rest is general. The precise part is CaseIQ: the privacy policy states that nothing uploaded there is stored on CaseMine's server and that any uploaded content is immediately removed after processing, protected by 256-bit end-to-end encryption. That is the strongest single confidentiality commitment on the record. Against it sits the treatment of everything else, in the same document: files, documents and data uploaded to a CaseMine account may be retained on its servers even after deletion or termination of the account, for backup and legal compliance purposes. Those two positions are both published and they point in opposite directions depending on which door a lawyer's document came through. Three questions are unaddressed entirely. Nothing describes segregation between customers, users or matters. Nothing addresses privilege or work product treatment. And nothing states what the underlying model provider may retain, which matters because the assistant is described as powered by GPT. The no-training commitment is a policy statement and is qualified, as set out on the training signal.
Published commitments cover what firms ask first, but not privilege itself. The security page, updated 30 July 2026 and read 11 September 2026, says customer prompts, files and history are not used to train Descrybe owned models; OpenAI's API is the main model provider, opted out of training, with abuse logs kept up to 30 days; uploaded briefs are not kept once text is extracted; users can delete research history at any time; and the product uses individual accounts with no shared or matter workspaces, so separation is per user and matter walls do not exist. The binding terms of service, described as the legal equivalent of a signed contract, say nothing on these points, so the commitments are policy, not contract. Privilege and work product are not addressed anywhere, including the privacy policy (updated 29 September 2025) and the security page, though the product now takes in users' briefs and pleadings. Where OpenAI processes requests is not stated.
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.
A real advice line is published, and the vendor's marketing contradicts it. The AMICUS product page says its information should not be considered legal advice, that it may include errors in cases, titles or citation codes, that users should consult qualified legal professionals and check original records, and that CaseMine is not liable for decisions or actions based on it. Naming citation codes as a known failure and pointing users back to original records goes beyond a footer. But the home page markets AMICUS as built to always give dependable advice backed by credible sources and as a trustworthy assistant giving accurate, precise and strategic answers, so the same product is sold as dependable advice and disclaimed as possibly inaccurate non advice. No jurisdiction limits are stated, though the platform is sold in India, the United States and the United Kingdom at once, and no bar, Bar Council or law society guidance is engaged. The site blocks automated readers on every path, so this text was read through search results rather than the page itself, and no quote is recorded.
Not addressed, on a product where it most obviously should be. The vendor says the product brings legal reasoning to anyone who needs it, lawyers and nonlawyers alike, and its founding mission is access to justice for people without legal training. A guided reasoning engine that takes a member of the public's question and returns a cited analysis raises the unauthorized practice question directly, and no position was found: no statement that output is not legal advice, no guidance on when to consult a lawyer, and no engagement with any bar rule. The closest is the terms disclaiming review or validation of outputs, which allocates risk without addressing the professional question.
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.
Nothing published addresses governance of model behavior: no responsible AI page, framework, accountable owner, account of testing before model changes, or anything on uneven output across jurisdictions, court levels, party types or populations. The vendor's certifications cover other subjects: ISO 9001:2015 for quality and ISO/IEC 27001:2022 for information security. Trade press reports an internal benchmarking framework used to evaluate AI before release, but no vendor page describes it and no results are published, so it is noted under citation accuracy.
Nothing is published on governance: no AI policy, evaluation method, bias or fairness assessment, accuracy monitoring, drift statement, model card, governance body or external standard. The gap matters here because every summary is machine generated, so any systematic bias in summarization flows into every search result and citator treatment, unseen by a user who reads the summary rather than the opinion. The vendor has already rebuilt all its summaries once on newer models, showing that quality depends on the model, and published no evaluation of either version. On 11 September 2026 the privacy policy and the security page, updated 30 July 2026, were also read; the security page says models may be updated for quality, safety, performance or reliability without notice unless the change is material, which concerns the supply chain, not governance.
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.
Two of the five elements are published, two are absent and one is generic. Retention is genuinely specific on one pathway, with CaseIQ uploads immediately removed after processing and never stored, and genuinely open-ended on the other, with account content retained on servers even after account deletion or termination for backup and legal compliance. Deletion and data subject rights are set out properly, with access, rectification, erasure, restriction, portability and objection, a thirty day response commitment and appointed EU and UK Article 27 representatives named as Euverify Ltd. Encryption is stated in transit and at rest, with 256-bit AES and SSL or TLS named on the pricing page, and ISO/IEC 27001:2022 certification is claimed. What is missing is what a security reviewer asks for next. No subprocessor list of any kind exists: third parties are given only as categories, being data storage and infrastructure hosting, database and account management, web analytics and payment processing, with not one entity named. And no incident or breach notification practice is published on any surface, which is a notable gap alongside an ISO 27001 certification. Access control is described only as authorized personnel bound by confidentiality.
Most of the ground is covered, specifically enough to hold the vendor to. The security page, updated 30 July 2026 and read 11 September 2026, says what is kept (prompts, research history, outputs, metadata and logs) and what is not (uploaded files are discarded once text is extracted). Users can delete history at any time and accounts on request; no zero retention mode is offered; production access is limited to authorized technical staff on a need to know basis; and data is encrypted in transit and at rest on DigitalOcean managed services in the New York region. It names subprocessors with links to their attestations: OpenAI's API, DigitalOcean, Clerk and Stripe, with Google Analytics in the privacy policy. OpenAI API training is opted out. Two gaps remain: no incident or breach practice is published, and no retention period is given for logs and metadata outside the user's deletion control.
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.
Liability is addressed once, only to disclaim it. The AMICUS product page says CaseMine is not liable for damages from decisions or actions based on AMICUS information, and acknowledges output may contain errors in cases, titles or citation codes. That places the loss on the user. No liability cap, indemnity, warranty, service level or insurance is published anywhere reachable. The Terms of Use referred to by the privacy policy could not be found, so what a signed customer actually receives is unknown, and a product page disclaimer is not the agreement. The Terms of Use, or any published cap or indemnity, could change this. The site blocks automated readers on every path, so this text was read through search results rather than the page itself, and no quote is recorded.
A published liability position exists and runs almost entirely one way. The terms of service make the user indemnify and hold Descrybe harmless for liability, loss, claims and expenses, including reasonable attorneys' fees, related to use of the service or its outputs, and limit any Descrybe liability to what the user paid in the twelve months before the claim. At published prices, that caps vendor exposure at $300 to $600 per user. The terms also say Descrybe does not review or validate outputs for accuracy. A specific, readable cap is real disclosure a buyer can price. No warranty, accuracy undertaking, service level or correction route was found.
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.
No integration with the systems legal work lives in was found, and none is claimed. Checked the FAQ, pricing page and feature list, CaseIQ page, privacy policy, about page and bespoke AI solutions page: no integrations page, connector, API or developer documentation, and no mention of document management, practice management, Word, Outlook or electronic filing. Asked whether anything needs installing, the vendor says CaseMine is fully cloud based and used in a browser on desktop and mobile with no setup, which describes a self contained product. Document upload is a file transfer, not a connection, and the bespoke solutions page offers to build tools around a firm's documents and workflows as consulting rather than a published integration.
Integration is the product, documented at tool level. Descrybe Legal Engine offers citation lookup, quote verification, treatment review and source retrieval to outside assistants over MCP, naming Claude, ChatGPT and Perplexity as supported hosts, plus an Open Connector for building more tools. A user guide describes the flow: the assistant calls one or more Descrybe tools, Descrybe runs the search, lookup, verification or retrieval and returns focused results, and the user can see which tools were called and what each returned. Work exports to Word. The integrations are to AI assistants, not legal practice systems: there is no connection to document management, practice management or a firm's matters.
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.
Cloud delivery is stated plainly, and neither region nor tenancy follows. The vendor says the platform is fully cloud based and used in a browser with no installation. No region is published: the privacy policy says data is on secure servers without saying where, and the only geography given is corporate, with operations in India serving India, the United States and the United Kingdom. No residency option is offered, processing is not distinguished from storage, and nothing says where a UK or US customer's data sits. No tenancy model is stated. Euverify Ltd is appointed as Article 27 representative for the EU and UK, which suggests cross border processing without describing it.
Tenancy and storage region are stated; processing location is half stated. The security page, updated 30 July 2026 and read 11 September 2026, says customer data is stored on DigitalOcean servers in the New York City region, the product runs on provider managed multitenant infrastructure, it is built around individual user accounts, and no single tenant, private or shared workspace option exists. Research requests are processed by OpenAI's API, and nothing says in which region. Queries routed through an outside assistant such as Claude, ChatGPT or Perplexity add a leg whose location is also not described. No residency options are offered.
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.
Certification is real and named to the version. CaseMine says it is certified to ISO 9001:2015 for quality management and ISO/IEC 27001:2022 for information security, and lists 256 bit AES and SSL or TLS encryption on every tier, with end to end encryption for CaseIQ uploads. No certification body, auditor, certificate number, issue or expiry date, or scope is published, and there is no route to obtain evidence, whether self serve, on request or under an NDA. There is no trust center, no penetration testing or vulnerability program is described, and no SOC 2 report is claimed. A buyer is asked to accept two certifications on the strength of a sentence.
The vendor states plainly that it holds no attestation of its own. The security page, updated 30 July 2026 and read 11 September 2026, says Descrybe does not hold its own SOC 2 or ISO 27001 certification and has begun work toward SOC 2. It relies on providers' programs, linking to trust portals for DigitalOcean (SOC 2 Type II, SOC 3), OpenAI (SOC 2 Type 2, ISO 27001, 27017, 27018, 27701), Stripe (PCI Level 1, SOC 2 Type 2) and Clerk (SOC 2 Type 2). Data is encrypted in transit over TLS and at rest on DigitalOcean, and production access is need to know. A provider's report covers its infrastructure, not the application layer where Descrybe's code, access controls and data handling sit, which is what a firm's security questionnaire asks about. Descrybe is a self funded two founder company charging $25 to $50 a month, and a SOC 2 examination is a material cost against that model. The security page sits in the site footer.
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 architecture is described well and the supply chain behind it is not. Retrieval augmented generation is named, with case law, statutes and principles pulled from CaseMine's database before any answer, and CaseIQ is identified as extractive rather than generative. The model is named only by family: AMICUS is repeatedly described as powered by GPT, and the premium tier promises the latest and most advanced models. No provider entity, model or version is named, nothing says where inference runs or what the provider may keep, and there is no change notification commitment, though the premium tier is partly sold on newer models. The privacy policy lists third parties only by category.
The provider is named, and change notification is the gap. The security page, updated 30 July 2026 and read 11 September 2026, says Descrybe uses OpenAI API models for summaries and research outputs, that OpenAI's API is its main LLM provider, that the exact model may vary by feature and change over time for quality, safety, performance or reliability, and that users are not notified of each update unless it materially affects terms, privacy or data processing. This settles earlier conflicting signals: the terms disclose third party models without naming them, and a third party directory had called DescrybeLM the company's own model. The page also says OpenAI training is opted out and abuse logs are kept up to 30 days, linking to OpenAI's data controls. No specific model or version is named, there is no commitment to notify of model changes, and where OpenAI processes requests is not stated.
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 buyer can learn the cost without a sales process. Three prices are published against named tiers with full inclusions: 99 US dollars a month for AMICUS AI with 250 credits, CaseIQ, CiteText, 50MB uploads, up to 20 cases and 2GB storage; 149.99 US dollars a month for 500 credits plus 100MB uploads, web search, the latest models, a dedicated account manager and unlimited cases; and an annual rate of 1,499.99 US dollars shown against a struck through 1,799.99. The unit is explicit: AI credits per month, upload size, storage and a cap on matters. A free tier below that offers keyword, citation and statute search and full judgments without an account, plus a free student registration. The vendor states no commitment and no fees. The site blocks automated readers, so these figures were read through search results; the Essential tier's price was not among them, so whether it is free or paid is unknown, and two readings of the page differ on top tier storage, 5GB or unlimited. A third party directory says paid plans start at 149.99 with a free forever plan; the vendor's own figure of 99 dollars is used.
Exact prices are published for every tier: $25 per month for Descrybe Legal Engine and $50 per month for the full Descrybe Platform with the Engine included. The unit is a monthly subscription per user. The vendor says every feature is included at either tier, so the headline price does not buy a limited subset. It also explains its pricing: it is self funded, without outside investor timelines or enterprise sales costs, and prices for the work rather than the market. A buyer can work out total cost from the public site without contacting anyone.
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.
Coverage is documented by jurisdiction and court. India covers the Supreme Court, all High Courts, tribunals and commissions. The United States covers all fifty states and federal courts, including the Supreme Court, circuit and district courts and the bankruptcy and immigration courts, with complete appellate and supreme court coverage but trial court rulings in only some states, an honest limit. The United Kingdom covers the Supreme Court, Court of Appeal, High Court and major tribunals. Buyers are named per market: advocates, attorneys, solo practitioners, solicitors, barristers, law firms, legal researchers and students. In house legal teams and government are absent from every buyer list. Practice areas are not listed, which suits a case law platform built to serve any practice but leaves no statement of where the product stops.
Coverage is listed court by court, with limits stated. The coverage page (descrybe.com/coverage, data as of 18 July 2026) lists 492 courts and tribunals and 10,969,112 opinions: federal (137 courts, 3,221,272 opinions), state (201 courts, 7,324,869), bankruptcy (97 courts, 80,941), specialized and agency tribunals (47 tribunals, 340,850) and tribal courts, plus statutes, regulations, constitutions, session laws and court rules. Each court shows its opinion count and first and latest dates, so a researcher can check whether a recent decision should be present. Most active supreme and appellate courts run to the first days of July 2026, and gaps are shown, such as Connecticut Superior Court ending in December 2014 and the Alabama Court of Appeals in 1999. No refresh schedule is stated beyond the as of date, and a few rows, including the New Hampshire district court and the Minnesota and Arizona appellate courts, show latest dates in the future, which look like source metadata. The vendor says it serves lawyers and nonlawyers alike.
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?
The commitment sits in the security answer of the public FAQ, and its qualifier matters. The promise is that data is never used to train external models, which covers the third party layer and leaves CaseMine's own models unaddressed. Since the assistant is described as powered by GPT, the natural reading is that customer content does not reach the base model provider's training, a real commitment; nothing says whether customer content informs CaseMine's own retrieval, embedding or fine tuning.
The privacy policy is silent on training, and the Terms of Use it mentions could not be found on 2 September 2026. AMICUS is described as trained entirely on authoritative legal sources and comprehensive legal archives, which concerns its source material, not user data. The position rests on the published policy statement, with its qualifier.
Stated in policy, in the form that matters. The security page, updated 30 July 2026 and read 11 September 2026, says Descrybe does not use customer prompts, files or conversational history to train Descrybe owned models, that its main model provider is OpenAI's API, that API inputs are not used for OpenAI training by default, and that Descrybe has opted out. That covers both the vendor's own models and the provider underneath.
It is policy rather than contract: the terms of service only disclose that third party AI models process primary law and carry no matching term. The privacy policy, updated 29 September 2025, was also read and is silent on training.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
Two published routes point in opposite directions, so the position is acknowledged rather than settled. The quoted commitment goes on to say that any content uploaded to CaseIQ is removed immediately after processing, protected by 256 bit end to end encryption. But the same privacy policy says files, documents and data uploaded to a CaseMine account may be kept on secure servers even after deletion or account closure, for backup and legal compliance, with no period.
Nothing says how long prompts, chats or generated drafts persist normally, and no retention setting is offered. Deletion is available on request with a thirty day response commitment, and erasure rights are set out. A document dropped into CaseIQ is discarded at once, while the same document saved to an account may outlive the account.
Retention is described and the user controls it by deletion rather than by a window. The security page (updated 30 July 2026, read 11 September 2026) states that Descrybe retains account and product data needed to operate the service, which may include prompts, research history, generated outputs, metadata and operational logs; that the only upload surface is brief review and uploaded files are not kept once the text is extracted; that no customer facing zero retention mode is offered; and that users can delete their historical entries at any time.
On the provider side it states that OpenAI may retain API abuse monitoring logs for up to 30 days. Recorded as customer controlled with no zero option: deletion on demand is real control over the research history, and the vendor says plainly that zero retention is not available. What the deletion statement does not reach is the operational log and metadata retention, for which no period is stated. Corrected from not addressed on 11 September 2026.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
Matter folders and case limits exist, but no isolation boundary between customers, users or matters is described. Nothing describes segregation between customers, between users inside a firm, or between matters. No tenancy or isolation statement appears anywhere. What exists is adjacent and concerns the vendor's own staff rather than the customer's structure: access to personal information is stated to be limited to authorized CaseMine personnel who require it and who are bound by confidentiality obligations, and a separate clause reserves access to account contents by employees or service providers when investigating suspected illegal activity or terms violations.
The product does have matter-shaped features, with case management capped by tier at five, twenty or unlimited cases and personalized folders for storing documents, but nothing describes those as a security boundary.
The permission model is documented and it is the simplest one there is: one user, one account, nothing shared. The security page (updated 30 July 2026, read 11 September 2026) states that Descrybe is structured around individual user accounts rather than shared organization workspaces, that saved searches, history and product data are associated only with that user's own account, that access is protected through application and database access controls, and, stated rather than left to inference, that no shared organization workspaces, matter based workspaces or separate file and vault permissioning are offered, on provider managed multi tenant infrastructure.
So walls at matter level do not exist in the product and the vendor says so. A firm that needs them keeps them outside the tool, by giving each user a separate account. Recorded as own model, documented, because the isolation model is published and enforced by the vendor; the summary carries the limit. Corrected from not addressed on 11 September 2026.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
The privacy policy addresses compelled disclosure squarely and never reaches notice. It states that personal information may be disclosed when required by law or in response to the legal processes quoted, to comply with legal obligations in the jurisdictions where CaseMine operates, and to law enforcement or government authorities where the company determines disclosure is necessary to protect the safety, rights or property of users, employees, the public or CaseMine itself.
That last limb is a unilateral determination rather than a legal compulsion. Nothing anywhere commits to telling the customer a demand has arrived, and nothing reserves discretion over notice either, so the question is not reached rather than answered adversely. Checked the privacy policy in full, the FAQ, the CaseIQ page and the pricing page on 2 September 2026. The Terms of Use referenced by the privacy policy could not be located or retrieved and might address it.
Not addressed. No clause on government or law enforcement requests, no commitment to notify a user before producing their data, and no transparency report were found. The terms of service cover termination, permitted use and liability, not third party requests. The privacy policy, updated 29 September 2025, covers account information, named third party services and children's data, not this. The security page, updated 30 July 2026, lists legal and compliance obligations among reasons authorized staff may access production data, but says nothing about disclosure to authorities or notice to the user. Checked the terms of service, privacy policy, security page, FAQ and site footer on 11 September 2026.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
Sources are identified court by court, which for primary case law identifies the sources themselves. India covers the Supreme Court, all High Courts, tribunals and commissions. The United States covers the Supreme Court, federal circuit, district, bankruptcy and immigration courts, and all fifty states, with complete appellate and supreme court coverage but trial court rulings in only some states. The United Kingdom covers the Supreme Court, Court of Appeal, High Court and major tribunals.
Updates are described as continuous, through an automated pipeline for each jurisdiction. No license, contract or permission basis is stated for any jurisdiction, no publisher or data supplier is named, and nothing covers the commentaries and columns the platform also carries. The company describes a long term focus on legal data infrastructure without saying how the data is obtained.
Sources are named at dataset level and listed court by court, with no license basis stated. The Harvard Caselaw Access Project is named as a source. The coverage page (data as of 18 July 2026, read 11 September 2026) lists 492 courts and tribunals and 10,969,112 opinions across federal, state, bankruptcy, specialized and agency, and tribal courts, plus statutes, regulations, constitutions, session laws and court rules, with an opinion count and date range for every court and source availability shown where applicable.
Not stated: the license or public domain basis for any source, or which source feeds which court. The content is AI summaries of opinions rather than the opinions alone, so provenance raises a second question the vendor does not address: what the summarization step did to the source.
Good Law Verification
Does the product tell you when the authority it just cited has been overruled?
CaseMine publishes its own citator rather than licensing one. The FAQ describes the CaseMine Citator, which tracks how a case has been treated in later decisions, flags negative treatment and overruling, and shows where and how often a judgment is cited, so a user can judge whether a precedent is still good law. It is included, not an add-on: the pricing page lists checking for negatively treated and overruled judgments with an advanced citator in the entry tier, alongside an importance matrix and citation search.
Nothing says how treatment is decided, whether by editors or automatically, what categories exist beyond negative treatment and overruling, or how fully each of the three jurisdictions is covered.
The vendor has its own treatment signal. Cytator launched June 2025 in the Legal Research Toolkit, showing treatment flags in the familiar positive, negative, cautionary and neutral convention, with analysis by issue rather than a single verdict per case. It shows forward treatment with binding and nonbinding citing cases, and a backward citator showing how an opinion treated each authority it relied on. Treatment review is also offered to outside assistants through Descrybe Legal Engine.
It is the vendor's own AI citator, not a licensed one such as Shepard's or KeyCite, and no accuracy, coverage or agreement measure against an established citator is published, which a practitioner would want before relying on a negative flag.
Refusal and Uncertainty Behavior
What does the product do when the answer is not in the corpus?
Retrieval before generation is the stated defense, with nothing on what happens when retrieval comes back empty. Nothing describes what the platform does when it cannot ground an answer, and no confidence, relevance or grounding indicator is described as shown to the user. The architectural answer offered instead is that retrieval happens before generation, so that responses are grounded in real legal sources rather than sounding convincing, and that citations link back to source authority the user can open.
That is a design claim about how hallucination is reduced rather than an account of what the system does at the point where retrieval returns nothing useful. Given the product drafts motions and petitions with judgment-based citations included, what happens when authority is thin is a live question the published material does not reach.
Documented. The vendor says DescrybeLM tells users when it needs more information and flags uncertainty rather than filling gaps with guesses, and that the engine clarifies jurisdiction, posture and key facts before answering an underspecified question. That is an explicit no answer path described as product behavior. The terms add that outputs are not reviewed or validated for accuracy. No published evaluation, transcript or example shows the behavior working, and no threshold or trigger for the uncertainty flag is described.
Fabricated Citation Record
Does a public court record exist addressing fabricated or hallucinated legal citations in output from this product?
Searched the AI Hallucination Cases database maintained by Damien Charlotin, and reporting drawing on it, on 2 September 2026 on the product name CaseMine, the assistant name AMICUS and the corporate name Gauge Data Solutions. No court order, opinion or disciplinary record naming the product was located. This is a statement about the public record rather than a finding about the product. Two things bound it. The database is heavily weighted to United States filings and its Indian coverage is thin, while India is CaseMine's home market and the market where fabricated-citation incidents have recently reached the High Courts and the Supreme Court.
And Indian legal commentary published in July 2026 discusses AMICUS alongside other Indian retrieval-grounded tools as reducing rather than eliminating hallucination, which is analysis rather than a court finding and is recorded here only because a reader searching this ground will meet it.
None located. General web searches combining the vendor and product names with court, opinion, sanction, hallucination and fabricated citation terms returned nothing on 29 Aug 2026. No docket database or court record tracker was searched. This records what the search found and is not a clearance. The vendor's terms disclose that it does not validate outputs, so a defective citation is an acknowledged possibility rather than one the vendor denies.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
Sold into the United States, India and the United Kingdom, the product engages no professional guidance from any of them. No published material engages professional or ethics guidance in any of the three markets served: no ABA Formal Opinion 512 or state bar opinion for the United States, no Bar Council of India or Solicitors Regulation Authority material, and no guidance for UK barristers or solicitors. This matters because the platform is sold into three regulatory systems and drafts filings in all of them, and because the vendor runs a legal technology certificate program with an Indian law school, engaging the profession's institutions for training but not their guidance in the product.
Not addressed. No named ethics opinion, no ABA Formal Opinion 512, no state bar guidance and no engagement with professional conduct rules was located, despite the cofounder holding an ABA Woman of Legal Tech recognition and the company positioning itself around access to justice. The absence is most visible where the product is offered to non lawyers alongside lawyers, since that is the configuration bar guidance speaks to most directly. Checked the home page, the FAQ, the user guide, the terms of service and the product pages on 29 Aug 2026.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
AI credits are counted per user, but no record of AI assisted work feeds a client bill. Nothing addresses how AI assisted work is recorded, billed or disclosed to a client, and no per matter record of it was found. There is also no quantified savings claim: the vendor's language about speed, clarity and working smarter gives no time or cost figure. The product counts AI credits per user per month and organizes work into cases, so the raw material for a per matter record exists, but nothing describes it as reportable or exportable, and credits are presented as an allowance, not a record.
Not addressed. Nothing published covers billing for AI assisted time or a record a practitioner could show a client of what was machine generated. The vendor's own prices, $25 and $50 a month with all features included, are clear, but that is its charge to the subscriber, not a position on the practitioner's billing. No time record, audit record or fee guidance was found. Checked the home page, the FAQ and the terms of service on 29 Aug 2026.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
A client asking which entities see its matter documents would get no named subprocessor, model provider or processing agreement. No subprocessor list exists in any form. The privacy policy names categories only, listing data storage and infrastructure hosting, database and account management, web analytics and payment processing, without identifying a single entity behind any of them. No model provider is named as a legal entity, the assistant being described only as powered by GPT.
No data processing agreement is published, no forwardable client-facing disclosure material was located, and there is no trust center or review pack. A firm asked by a client which entities may see its matter documents could not answer from anything CaseMine publishes. The two ISO certifications are claimed without any accompanying scope or report, so they do not fill the gap either.
A subprocessor and model provider list is published, each entry linked to that provider's own attestation. The security page (updated 30 July 2026, read 11 September 2026) names OpenAI's API as the primary model provider, DigitalOcean for hosting and storage in its New York region, Clerk for authentication and Stripe for payments, with links to each provider's trust portal or security documentation and the certifications each holds; the privacy policy (updated 29 September 2025) adds Google Analytics for site analysis.
That is the list a firm forwards to a client under outside counsel guidelines, and it exists without a sales conversation. What is not published is client facing disclosure material of Descrybe's own, and Descrybe holds no attestation of its own to add to the pack, which it states plainly. Corrected from not addressed on 11 September 2026.
Court Disclosure Support
If a judge’s standing order requires an AI disclosure, can the product produce one?
Linked citations, search history and case folders are kept, but none is a record of model use. Nothing addresses court standing orders, disclosure of AI use, or certification that a person checked the citations. The platform does produce records a professional could keep: citations linked to sources, unlimited search history on every tier, task management and case folders. None is described as a record of model use, and nothing shows which model produced which passage or who reviewed it.
The product drafts motions and petitions with citations in three jurisdictions where courts have begun asking this question, and its citator is the nearest thing it offers to a verification trail.
Partial record, strong on sources. DescrybeLM answers carry a table of authorities with verified citations linked back to the primary law, the reasoning is stated to be visible at every step, quote verification and citation resolution are available as discrete tools, research sessions are saved and revisitable, and work product exports to Word, so a practitioner can produce what was retrieved and relied on. In the engine configuration the user can additionally see which tools an assistant called and read what each returned, a trail at the tool level.
What is missing is the other two elements a standing order asks for: nothing records which model produced a given output, and no human verification record is captured or exportable. The vendor's disclosure that it does not validate outputs makes the human verification gap more consequential, since the checking is entirely the practitioner's and nothing in the product evidences that it happened.
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.
- AI Governance and Bias Disclosure
- Bar Guidance Alignment
- Billing and Fee Posture
Which one fits
Choose CaseMine if
- Your work crosses more than one jurisdiction. CaseMine publishes coverage court by court in three: the Supreme Court of India with all High Courts and tribunals, the United Kingdom Supreme Court, Court of Appeal, High Court and major tribunals, and the United States across all fifty states and the federal system, with the limit stated rather than glossed, since appellate and supreme court coverage is complete while trial court rulings are described as select states.
- You want the grounding described rather than asserted. CaseMine states that its assistant runs on a retrieval augmented generation framework that pulls relevant case law, statutes and legal principles from its own database before generating any response, describes CaseIQ as extractive rather than generative so a whole brief or judgment becomes the query, and publishes its own citator flagging negative treatment and showing where and how often a judgment has been cited.
- You want to try the research before paying for the AI. Keyword search, citation search, statute search and full judgment texts are usable without an account, with a separate free student registration, and the paid tiers are published at 99 US dollars a month for 250 AI credits with CaseIQ and 50MB uploads, and 149.99 for 500 credits, 100MB uploads, web search and unlimited cases, with an annual rate of 1,499.99.
Choose Descrybe if
- You do not want the price to depend on which features you need. Descrybe publishes 25 US dollars a month for its Legal Engine and 50 for the full platform with the Engine included, states that every feature is included at either tier, and explains the basis, saying it is self funded, carries no enterprise sales overhead and prices for the work rather than the market.
- Verification is the job, not a feature. Descrybe answers carry a table of authorities linked back to primary law with the reasoning visible at each step, and the toolkit ships quote verification, citation resolution and case verification as first class tools alongside Cytator, a citator offering issue level treatment analysis with positive, negative, cautionary and neutral flags plus forward and backward citing.
- You would rather research inside the assistant you already use. The Descrybe Legal Engine exposes citation lookup, quote verification, treatment review and source retrieval to external assistants including Claude, ChatGPT and Perplexity over an MCP interface and an Open Connector, with the user able to see which tools were called and read what each returned, and the corpus is searchable and summarized in both English and Spanish with plain language versions.
In summary
CaseMine
CaseMine is an AI legal research platform covering case law in India, the United States and the United Kingdom. It runs four tools over a repository the company has built since 2013. AMICUS is a conversational assistant grounded through retrieval augmented generation over its own database. CaseIQ takes a whole brief or judgment as the query and returns matching precedent. Alongside them sit intent based Parallel Search, citation search and its own citator. The AI Legal Index records three published price points, with credits, upload size and storage stated, above a genuinely free search tier. No accuracy figure, AI governance material or hosting region is published.
Descrybe
Descrybe is a United States primary law research platform built on a corpus of AI summarized judicial opinions. Its July 2026 coverage page lists 10,969,112 of them across 492 federal, state, bankruptcy, tribal and specialized courts and tribunals, each court with its own opinion count and date range. DescrybeLM is a guided reasoning engine returning citation backed answers, Cytator handles issue level treatment analysis, and a connector layer exposes citation lookup, quote verification and source retrieval to external assistants. According to the AI Legal Index, Descrybe publishes 25 and 50 dollars a month with every feature included at either tier, and publishes where its corpus starts and stops court by court. Its security page names OpenAI's API as the model provider and DigitalOcean's New York region for storage, and states that customer prompts, files and history do not train its models. Descrybe holds no security certification of its own and states that SOC 2 is in progress. It publishes no treatment of privilege or work product and no accuracy figure.
Questions buyers ask
CaseMine vs Descrybe: which is better for case law research?
They differ in what the AI is doing. CaseMine puts a retrieval layer over a case law repository it has built since 2013 across three jurisdictions. On Descrybe the corpus is itself model output: 10,969,112 judicial opinions summarized by AI across 492 courts and tribunals on its July 2026 coverage page. Both publish their prices in full. From the AI Legal Index, based on each vendor's own published materials as of September 11, 2026. No vendor pays for placement.
What does each cost?
Both publish figures. Descrybe charges 25 US dollars a month for the Legal Engine and 50 for the full platform with the Engine included, and states that every feature is included at either tier. CaseMine publishes 99 US dollars a month for 250 AI credits with CaseIQ and 50MB uploads, 149.99 for 500 credits with larger uploads and unlimited cases, and an annual rate of 1,499.99, with keyword, citation and statute search free without an account. From the AI Legal Index, based on each vendor's own published materials as of September 11, 2026. No vendor pays for placement.
Does either have a citator?
Both do, and they work differently. CaseMine publishes its own citator flagging whether a judgment has been negatively treated or overruled and showing where and how often it has been cited. Descrybe publishes Cytator, which offers issue level treatment analysis with positive, negative, cautionary and neutral flags plus forward binding citing cases and a backward citator, so the treatment is reported per issue rather than per case. From the AI Legal Index, based on each vendor's own published materials as of September 11, 2026. No vendor pays for placement.
Do either name the models behind the product?
Descrybe does, CaseMine only to the level of a family. Descrybe's security page names OpenAI's API as its primary model provider, states that the exact model varies by feature and may change without user notice unless the change is material, and states that it has opted out of OpenAI training on its API traffic. CaseMine describes its assistant as powered by GPT and its premium tier as carrying the latest models, without naming a provider entity, a model or a version. From the AI Legal Index, based on each vendor's own published materials as of September 11, 2026. No vendor pays for placement.
What do CaseMine and Descrybe both leave unpublished?
Neither publishes an accuracy measurement: no test set, hallucination rate or evaluation appears on either record, which matters on products whose corpus and citator are themselves machine generated. Neither publishes an AI governance position, an accountable owner or any bias evaluation. Neither engages any bar, Bar Council or law society guidance, on platforms sold across three jurisdictions in one case and to non lawyers in the other. On data handling they diverge. Descrybe states where data is stored (DigitalOcean, New York), that customer content does not train its models, and how retention and deletion work. CaseMine states none of that, and holds an ISO certification Descrybe does not. From the AI Legal Index, based on each vendor's own published materials as of September 11, 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.
Both vendors tell readers plainly that their output is not validated, which is worth weighing. Descrybe's terms of service state that it does not review or validate outputs to confirm they are accurate, truthful, relevant or reliable. CaseMine's product page states that its assistant may include inaccuracies relating to cases, titles or citation codes, and advises users to consult qualified professionals and refer to original records. Its home page markets the same assistant as engineered to provide dependable advice. Both halves belong on the record. Neither vendor reviewed this page.