Docket Alarm
Litigation intelligence platform built on United States court records, holding over a billion dockets and documents drawn from every federal court, more than 43 state court systems, the USPTO, the Patent Trial and Appeal Board, the Trademark Trial and Appeal Board, the International Trade Commission and a growing set of international jurisdictions, all in one searchable index. The core of the product is search, alerting and retrieval: full-text search across docket text, documents, parties, attorneys and judges; alerts on any case, party, attorney or judge; and direct PACER retrieval with government fees passed through at the government's own rate. Layered on that are litigation and motion analytics sold as add-ons, judge and opposing-counsel profiling, a developer API that returns a single normalised JSON structure across every jurisdiction, and an AI layer of two parts: Talk to the Docket answers a plain-English question about any case and links each statement to the exact docket entry or document behind it, and AI Summaries condense an individual filing on opening. Buyers are litigators, IP counsel, in-house and corporate legal teams, law students and legal-tech developers, with the corporate use case extending to screening counterparties, vendors and prospective hires for litigation history. Filings that one subscriber retrieves enter a shared library available to every other subscriber, which is why no court fee applies to material already held. Pricing is published in full across two self-serve tiers with every unit priced, alongside a free weekly allowance of searches and document views. Docket Alarm, Inc. remains the contracting entity: the company joined Fastcase in 2018, Fastcase combined with vLex in 2023, and Clio acquired vLex in 2025, so the product is now sold under its own name as part of Clio.
Capability grades
All 15 axes, graded from public sources on the date shown. 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 machine learning is real and the product it sits on predates it and stands without it. What the AI does: Talk to the Docket answers plain-English questions about a case, AI Summaries condense a filing on opening, and extraction turns docket entries and underlying filings into the normalised party, counsel and per-entry structure the API returns and the motion and judge analytics run on. Strip all of that out and what remains is the business Docket Alarm sold for a decade and still leads with: full-text search across a billion filings, alerts on any case, party, attorney or judge, docket tracking, direct PACER retrieval and bulk document download. The pricing page settles it more clearly than the marketing does. Both published plans are priced in units of the non-AI product, at 4 dollars a document, 3 a docket update, 2 a day for tracking and 1 a search, and the analytics products are sold separately as Federal Analytics, Easy Analytics and the Analytics Workbench add-ons. That is a feature layer on a data platform whose value stands without it. Graded one band below UniCourt, read the same day, where the normalised dataset is the whole product rather than one route into it.
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 documented and demonstrated; measurement is absent. The AI Assistant page states that the system draws only on the docket being viewed, reads the actual record from the complaint through the most recent filing including the documents behind the entries rather than the entry text alone, and links every statement to the docket entry or document supporting it so a reader can open the source in one click. The page carries a worked example: an answer on a named 2026 Northern District of California declaratory-judgment case, with a Sources list identifying the complaint and its filing date. What is missing is any number. No accuracy rate, no test set, no evaluation and no error analysis was located on 31 Aug 2026, while the same page asserts as an absolute that output carries no hallucinated facts or fabricated citations. That claim sits against the vendor's own agreement, which states that machine-generated output is probabilistic and should be evaluated for accuracy including by expert human review. The marketing promises what the contract declines to warrant, and a buyer should read both. One limb of this axis does not apply rather than failing: a citator or good-law check is not in scope for a system that retrieves case records instead of asserting propositions of law, and it is neither credited nor penalised here.
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.
A real review surface and a real stopping behaviour are published, and what happens after an error is not. The review surface is specific: each statement in an answer links to the entry or document that supports it, and the vendor frames the point of that link as letting the reader decide for themselves whether the answer is correct. The stopping behaviour is stated too, that where a fact is not in the record the system says so rather than filling the gap. The agreement adds the supervision obligation in writing, telling the customer that machine-generated output should be evaluated for accuracy including by employing expert human review. Against that, checked on the home page, the AI Assistant page, the API page and the terms on 31 Aug 2026: no confidence threshold or exception behaviour is described, no route exists to report or correct a wrong answer, and the case-briefing function, which the vendor says reads hundreds of docket entries and traces the procedural history unattended, has no described checkpoint between the request and the finished brief.
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.
No production evidence of any kind was located. Searched the home page, the full use-cases page covering all five buyer segments, the about page, the AI Assistant page, the developer API page, the coverage page and the pricing page on 31 Aug 2026. There is no named customer, no attributed testimonial, no case study, no customer count, no logo strip and no figure for what changed at any organisation. What stands in its place is unattributed: a claim that thousands of legal professionals rely on the platform, a signup modal saying the vendor has relationships with many large firms, and a claim of direct partnerships with law schools that names no school. The five example workflows on the use-cases page are explicitly hypothetical illustrations rather than accounts of real deployments, and the one time figure published, that a case brief replaces two to four hours of paralegal work, is a product claim with no customer or method attached. For a platform this long established, in market since well before its 2018 acquisition, the complete absence of a named reference is the finding.
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.
Two live documents on the same property give a buyer opposite answers, and that is the finding. The privacy policy makes a specific and well-aimed commitment: it recognises that litigation activity involves viewing potentially adverse litigant material, and states that personal search history is kept strictly confidential to the user's organisation and is not sold to third-party marketers. That addresses the real exposure here, which is that the cases a firm looks up map onto its live docket. The terms of service, which govern, run the other way. They reserve the right to monitor and record customer usage including content interactions, search history and IP addresses, and to track activity to gather further information about the customer and its authorised users. They grant the vendor an irrevocable, perpetual and sublicensable licence over user content to improve its products. They disclose that a limited number of staff and third-party foundational AI models may technically access user content. And the confidentiality clause excludes information that is publicly available, which describes nearly everything in the corpus. Nothing reconciles the two documents. Beyond them, no training position, no retention period, no access model within a subscribing firm and no privilege or work product treatment was located on 31 Aug 2026. Graded on the documented position rather than the better half of it.
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.
A substantive position, published in the agreement and nowhere else. The terms carry a clause headed to the effect that no legal counsel is provided: content is for practical and informative purposes, should not be construed as professional advice, does not create an attorney-client relationship, and users are told to seek qualified professional counsel before acting on it. A separate clause covers AI output specifically, describing it as machine-generated probabilistic prediction to be evaluated for accuracy including by expert human review, which is the competence and supervision dimension most vendors omit. The audience is defined rather than left ambiguous: the services are stated to be for professional use, and consumers and non-professionals are expressly prohibited from accessing them, so there is no consumer-facing surface to disclose on. Two things hold this short of the top band. No bar or ethics guidance is named anywhere, including ABA Formal Opinion 512. And no practice-jurisdiction limit is stated; the only geographic limit located is a licence restriction on using the service outside the country in the subscription agreement, which is a commercial term rather than a statement about where the product's answers can be relied on. Checked the home page, AI Assistant page, use-cases page, about page and footer on 31 Aug 2026: none of this appears on any product surface, only in the terms.
AI Governance and Bias Disclosure
Published governance over model behaviour: who owns it inside the vendor, what is tested before release, and what is disclosed about disparate output across matter types, parties, or populations.
A governance surface exists at parent level and nothing connects it to this product or to the exposure this product creates. The Clio trust centre, publicly reachable at trust.clio.com and read on 31 Aug 2026, carries an AI section with three named topics, AI Risk Management, Third-Party AI Diligence and Employee AI Usage, and a featured document titled AI Overview. Their contents were not read, so nothing is credited here beyond their existence. On Docket Alarm's own property there is nothing at all: searched the home page, AI Assistant page, use-cases page, about page, coverage page, API page, privacy policy and terms and located no responsible AI statement, no accountable owner, no pre-release testing regime and no fairness or uneven-output evaluation, and no link from any Docket Alarm page to the trust centre. The gap has a concrete edge on this product. The published corporate use case invites customers to search every court for an individual's complete litigation record before hiring an executive or onboarding a vendor, and to judge whether a person is litigious and how their cases tend to end. Screening identified individuals on litigation history is the question this axis exists to ask, and no published position addresses it.
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.
A short privacy policy that is specific where it speaks and silent across most of the ground. What it does publish is real: Google App Engine named as the hosting environment for all non-payment data, a PCI-certified payment processor with card details encrypted with AES-256 and no employee access to them, a commitment to seek consent before using collected information for a new purpose, and prior versions of the policy available on request. The removal route is better than most: a free suppression request with a stated five-business-day decision window, published criteria in both directions including automatic suppression of cases mentioning minors, an explicit refusal to suppress strong-public-interest cases or requests from legal professionals wanting to be disassociated, and an honest statement that search engine snippets may persist for weeks afterwards because the vendor does not control them. What is absent, checked on the same surfaces plus the terms on 31 Aug 2026: no retention period for anything, no deletion route for customer account data as distinct from court records, no access control model, no encryption statement covering anything other than payment details, no subprocessor list, and no incident or breach notification practice. The policy also carries no date, so its currency cannot be established. The parent trust centre publishes incident response, breach notification and backup material, but is not linked from this property and its scope over Docket Alarm was not established.
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.
The allocation of loss is published in full, readable before signing, and almost entirely one-directional. The terms cap total liability, whatever the cause or theory of recovery, at one hundred dollars, and exclude indirect, special, incidental and consequential damages including lost profits, revenue, data and opportunities. Carve-outs are named: fraud, wilful misconduct, violations of applicable law, infringement or misappropriation of intellectual property and breach of confidentiality obligations sit outside the limitation. Warranties are disclaimed expressly, on an as-is and as-available basis, including merchantability, fitness for a particular purpose and non-infringement, with specific disclaimers that results will be accurate, reliable or error-free. A separate clause addresses AI output directly and declines to warrant it. The contracting entity for Docket Alarm subscriptions, the governing law and the forum are all named. What a buyer does not get is anything running the other way: the only indemnity in the agreement runs from the customer to the vendor, no warranty attaches to output, and no insurance position appears in the agreement, though the parent trust centre lists a liability insurance document that was not read. What lifts this above the vendors in this pull that publish nothing is that the risk is allocated precisely and a buyer can read the number before signing. What the number is, is one hundred dollars.
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.
A documented API and no documented connector into the systems legal work actually lives in. The developer API page names its endpoints with methods and describes what each returns: Search and Search PACER for querying by party, attorney, court, date range, case type or full text; Get Docket for the full docket sheet including every entry, parties, counsel, related cases and status; Get Document for the PDF behind an entry; Track a Case as a POST subscribing to a case or saved search; and Ask Docket, exposing the AI assistant's grounded answers programmatically. A sample JSON response is published showing the normalised structure, and a full API reference sits at a public documentation path. The terms name three native integrations, Microsoft OneDrive, Dropbox and PACER ECF accounts. What is not there is a practice system. The API page invites a firm to pipe dockets into its matter management, DMS or BI stack but names no such system and offers no pre-built connector, and nothing was located for iManage, NetDocuments, Word, Outlook, e-billing or filing on 31 Aug 2026. Data moves one way, out of Docket Alarm, with no write-back or two-way sync described.
Deployment Model and Data Residency
Where the software runs and where the data sits. Multi tenant cloud, single tenant, private deployment, on premises, and whether region of residence is a published option or an enterprise conversation.
One hosting fact is published and everything a buyer would ask after it is not. The privacy policy states that all data other than payment information is stored on servers operated by Google through the Google App Engine hosting service, and points to Google's own privacy material for how that is protected. That is more than most of this lane publishes and it is where the disclosure stops. Searched the home page, API page, coverage page, terms and privacy policy on 31 Aug 2026 and located no region, no residency option, no tenancy model, and no distinction between where data is processed and where it is stored. There is also an unreconciled discrepancy across the ownership layers: Docket Alarm's own privacy policy names Google App Engine, while the parent trust centre's infrastructure section names Amazon Web Services and publishes separate status pages for United States, Canada, European Union and Australia. Which applies to this product is not stated on either property.
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.
Something real was located and none of it connects to this product, so it is described here rather than credited. The Clio trust centre, read on 31 Aug 2026, is a SafeBase portal reachable without a sales conversation. It names GDPR, HIPAA, SOC 1 Type 2, SOC 2 Type 2 and TX-RAMP, publishes a dated update stating that the 2025 SOC 2 Type II report covers 1 June 2024 to 31 May 2025, and offers a penetration test report summary, both SOC reports, a liability insurance document and an AI overview behind a self-serve access request. What is missing is any connection to Docket Alarm. The portal tab reached states that it applies to the Vincent and vLex products, Docket Alarm is named nowhere on any page read, and no page on docketalarm.com links to the portal at all: the footer carries privacy, terms and case removal and nothing else. On the product's own property there is no security page, no named certification, no auditor, no scope statement, no coverage period and no trust portal, checked across the home page, AI Assistant page, use-cases page, about page, API page, coverage page and both legal documents. A buyer cannot establish that the thing they are buying sits inside any audited boundary, which is what the scope section of an attestation exists to answer. One honest qualification about the band rather than the vendor: neither published band fits this case cleanly, because C describes badges on the site with no scope or report and there are no badges here, while D describes no attestation located and one was located at the corporate family level. A band amendment for the corporate family case is logged open.
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 product acknowledges that third-party models sit underneath and never says whose. The terms disclose that a limited number of staff and third-party foundational AI models used in rendering the vendor's intelligent technology services may technically access user content, which tells a buyer that customer material reaches an external model without identifying it. Beyond that the property is silent: nothing on the AI Assistant page, the developer API page or any other Docket Alarm surface names a model, a provider, a version or a processing location, and no commitment to notify customers of a change to the model or its training data was located on 31 Aug 2026. That silence is what this grade rests on. The corporate parent does publish a subprocessor list naming Google Cloud, OpenAI, Anthropic and Amazon Web Services, and it is recorded here for completeness rather than credited: the portal states coverage of the Vincent and vLex products, does not name Docket Alarm, and is linked from nowhere on this property, so nothing connects those providers to Talk to the Docket or AI Summaries. On what a buyer of this product can establish about the system answering their questions, the answer is that it is a foundation model belonging to someone else.
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 price this product completely without speaking to anyone. Two self-serve plans are published with figures: Flat-Fee at 99 dollars a month covering unlimited document access, case alerts, docket tracking and search plus motion filtering, and Pay-As-You-Go at 39.99 dollars a month with every unit priced at 4 dollars a document, 3 a docket update, 3 a day for new case alerts, 2 a day for docket tracking and 1 a search in federal courts. Add-ons are named as Federal Analytics, Easy Analytics and the Analytics Workbench. The pass-through costs are handled with unusual candour, which is where most of this market becomes vague: government fees are stated to be passed on without markup at the government's own rate, the PACER schedule is reproduced at 10 cents a page with a 30-page or 3 dollar cap and no cap on transcripts, nothing is charged where the document is already held, and for-pay state courts, mostly in California, are always presented as an accept-or-decline choice at the point of access. A free tier is published as a stated weekly allowance of searches and document views for material already in the system, excluding anything bought from a government system, and a three-day no-commitment trial is offered. The one gap is API pricing, which is sales-led and carries no published figure or unit.
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.
Both halves of this axis are documented, and the boundaries are stated rather than left open. Five buyer segments each get their own treatment on the use-cases page, with the question each brings and the features answering it: litigators on docket tracking, opposing counsel and judge analytics; IP counsel on USPTO, PTAB, TTAB and ITC Section 337 work; law students on real filings and law school programmes; in-house counsel on portfolio litigation risk and agency enforcement; and corporate teams on counterparty and vendor screening. The API page adds four more, and the signup form enumerates nine organisation types including government, school and bar association. Coverage is published court by court on a dedicated page: all 94 federal district courts, all 13 circuits, every bankruptcy court and the Supreme Court, stated as complete PACER coverage; per-practice-area agency tables naming the USPTO, PTAB, ITC, FDA Orange Book, TTAB, US Tax Court, FERC, FTC and NLRB with a coverage status against each; state courts at 43-plus with a court-level table; and international coverage by country. The limits are explicit rather than implied: the state map legend carries a not-yet-covered category alongside full and partial, 43-plus is not 50, and a request-a-court route is published for the gaps. One retrieval limit to record: the interactive state and international tables render client-side and did not return values on 31 Aug 2026, so the structure and legend were read but not the per-state detail.
Legal Signals
What each signal meansA signal records what public sources say on the date shown. It is not a grade and it is not a recommendation. Where a signal reads Not addressed, it means the index did not locate the material in public sources on that date, which is a statement about disclosure rather than about the product.
Client Data in Training
Can material a lawyer puts into this product be used to train a model?
No located term or policy addresses the question either way.
No located term or policy states whether customer content is used to train or refine any model, either way. What the terms do carry is a licence pointing in that direction without naming it: the customer grants the vendor an irrevocable, perpetual, transferable, sublicensable, royalty-free and worldwide licence over user content to maintain and provide the services, to improve its products and services, and to create aggregated and de-identified information. The asymmetry is worth a buyer's attention, because the same agreement expressly prohibits the customer from using the services or their content to train any AI models. Read on 31 Aug 2026 alongside the privacy policy, which addresses search history and marketing but not model training.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
Retention is acknowledged in public materials with no stated period.
Retention is acknowledged in two documents and quantified in neither. The privacy policy states that a record of each search is kept so the user can find it later in their search history, and that search history is also used for aggregated statistics and support. The terms assert a right to monitor and record customer usage of the service, naming content interactions, search history and IP addresses, and to track activity to gather further information about the customer and its authorised users. Searched both documents plus the home page, AI Assistant page and API page on 31 Aug 2026: no retention period is stated for search history, for questions put to the AI assistant or for its answers, no deletion route is offered, and no customer-configurable or zero-retention setting exists.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
Segregation is asserted in public materials with no published detail on how it is enforced.
One separation claim is published and nothing describes how it is enforced. The privacy policy states that personal search history is kept strictly confidential to the user's organisation, which is a tenant-level assertion aimed at the right risk, since the cases a firm looks up reveal its docket. No mechanism accompanies it: nothing states whether one user's searches, saved work or tracked cases are visible to colleagues inside a subscribing firm on plans carrying multiple users, and no matter-level concept exists anywhere in the product. Two features cut the other way and are disclosed by the vendor: filings retrieved by any user enter a shared library available to all other users, and the terms reserve a right to monitor and record content interactions and search history that the privacy policy's confidentiality claim does not mention. Checked 31 Aug 2026.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
No located term or policy addresses third party requests for customer data.
Disclosure to authorities is contemplated and notice to the customer is not. The terms list compliance with lawful requests from authorities among the grounds on which staff and third-party foundational AI models may access user content, and the confidentiality clause permits release of confidential information where necessary to comply with legal requirements. Searched the terms, the privacy policy, the home page and the footer on 31 Aug 2026 and located no commitment to notify the customer of a subpoena or government demand, no undertaking to seek a waiver where notice is prohibited, and no transparency report. Whether a firm would learn that its search history had been demanded rests entirely with the vendor.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
The vendor names its primary law sources and the licence or public domain basis for each, with an update cadence.
Sources are named to the level of the individual court and the rights basis is stated, which together is the strongest provenance position located in this pull. The coverage page enumerates the corpus rather than describing it: all 94 federal district courts, all 13 circuits, every bankruptcy court and the Supreme Court through a stated direct PACER connection, state courts across 43-plus states, and named agencies including the USPTO, PTAB, TTAB, ITC, FDA Orange Book, US Tax Court, FERC, FTC and NLRB, each with a coverage status against it. The rights basis is public record, stated on the removal page and reflected in the terms, which classify content sourced from government publications and public records as publicly sourced material and distinguish it from third-party licensed content. Cadence is stated as real-time updates on new filings. Two limits: the basis is stated for the corpus as a whole rather than per source, and the international courts are listed without a stated basis or supplier.
Good Law Verification
Does the product tell you when the authority it just cited has been overruled?
No located public material addresses whether authority is checked for subsequent history.
Searched the home page, AI Assistant page, API page, coverage page and the terms on 31 Aug 2026. No citator, treatment signal or subsequent-history check is claimed, and the question does not bite on this product: Docket Alarm delivers dockets, filings and court documents rather than propositions of law requiring validation, and its AI assistant answers questions about a case record instead of citing authority for a legal proposition. Recorded as not addressed because that is the honest value, with the note that this is a question the product class does not raise rather than one the vendor has neglected. The adjacent statement the vendor does publish is a general one in the terms, that content from courts and other official sources is provided as is with no warranty as to accuracy or completeness.
Refusal and Uncertainty Behaviour
What does the product do when the answer is not in the corpus?
The vendor describes refusal or abstention behaviour in public materials.
An explicit no-answer path is described in the vendor's own material, which is uncommon. The AI Assistant page states that the system draws only on the docket being viewed, that it does not fill gaps with plausible-sounding but fabricated information, and that where a fact is not in the record it says so. The design rationale is published alongside it, contrasting general-purpose tools trained to sound confident with a system built so that every answer is traceable to the record. What is not published is any demonstration: no worked example of a refusal, no evaluation, no measurement of how often the behaviour fires or fails, and no confidence or grounding score exposed to the user. Documented rather than demonstrable on that basis, read 31 Aug 2026.
Fabricated Citation Record
Does a public court record exist involving output from this product?
No court order, opinion or disciplinary record naming this product has been located as of the date shown. This is a statement about the public record, not a finding about the product.
No court order, opinion or disciplinary record naming this product has been located as of 31 Aug 2026. Instrument searched: the AI Hallucination Cases database maintained by Damien Charlotin, which tracks decisions worldwide where a court addressed hallucinated AI content and records the tool implicated where known, searched on the product name and on the company name, alongside 2026 sanctions coverage and trade press. This is a statement about the public record on the date shown rather than a clearance, and it is bounded by what that database covers. One appearance is worth distinguishing from a hit: a parent-company blog post uses Docket Alarm's own corpus as the denominator for a hallucination base rate, putting the tracked cases against roughly 40 million US cases filed since January 2023. That is the product as the measuring instrument, not as the tool implicated.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
No located public material engages with bar or ethics guidance.
Searched the home page, AI Assistant page, use-cases page, about page, coverage page, API page, privacy policy and terms on 31 Aug 2026. No bar or ethics authority is engaged with anywhere, including ABA Formal Opinion 512 and any state bar guidance, and nothing maps the product to a professional conduct obligation. The terms do carry a clear no-legal-advice position and an instruction that AI output should receive expert human review, but both are drafted to protect the vendor's position rather than to help a lawyer evidence compliance with their own duties, and neither cites the guidance those duties come from. Recorded as not addressed on that distinction.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
Public materials claim time savings without addressing billing or disclosure.
Time savings are claimed and the client's side of the equation is not addressed. The AI Assistant page states that a full case brief is generated in seconds against work that takes a traditional paralegal two to four hours, and the use-cases page describes research finishing in far less time. Searched the pricing page, the AI Assistant page, the use-cases page and the terms on 31 Aug 2026 and located no per matter record of AI-assisted work intended for fee purposes and no guidance on billing, fee or client disclosure treatment where an AI answer informs advice. One adjacent strength belongs to the vendor's own charging rather than to this signal: per-document and per-search unit pricing, with court fees passed through at the government's rate without markup, gives a firm a clean disbursement to attribute to a matter.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
No located public material supports a client side disclosure obligation.
No material supporting a client-side disclosure obligation was located for this product. Searched the home page, AI Assistant page, use-cases page, about page, developer API page, coverage page, privacy policy and terms of service on 31 Aug 2026: no subprocessor list, no model provider named, no data processing agreement, no trust portal, no named certification and no client-facing consent or notification material. A firm asked by a client which vendors see its research activity would have nothing to forward. One thing was located and is described rather than credited, because it belongs to the corporate parent and nothing connects it to this product: the Clio trust centre, publicly reachable without an agreement, names Google Cloud, OpenAI, Anthropic and Amazon Web Services as subprocessors and publishes a data processing agreement alongside them. Docket Alarm is not named on any page of that portal, the tab reached states it covers the Vincent and vLex products, and docketalarm.com references the portal nowhere, so a firm would have to find it unaided and then assume it applies to what it bought.
Court Disclosure Support
If a judge’s standing order requires an AI disclosure, can the product produce one?
Some elements of the record are available, short of a document level export.
The sources behind an answer are recorded and the AI provenance is not. Every statement the assistant produces links to the docket entry or document supporting it, the worked example on the vendor's page displays a Sources list identifying the filing and its date, and the underlying material is the court record itself, so a lawyer can produce the primary documents an answer rests on. What cannot be established is the other half a standing order asks for: no model is identified or versioned anywhere on the property, so which system generated a given answer is unknowable, nothing records who verified it, and no export designed for an AI-use disclosure or certification was located on 31 Aug 2026.