Q
Quilia
Quilia is a client-facing case app for personal injury and workers' compensation firms, from Record System, Inc. of Las Vegas, Nevada. The firm invites its client from its case management system; the client gets a text, installs the app, and from then on records treatment visits, providers, prescriptions, pain scores, symptoms and photographs, uploads police reports, medical records and bills, signs documents through the built-in e-signature tool, and messages the firm.
Quilia's AI reads what arrives: it extracts dates, locations, parties, providers, treatment details and injury descriptions from documents and images, classifies and files them chronologically, asks the client to confirm what it found before anything lands on the case, and tracks treatment for gaps. An AI case assistant called ℚ, described by the vendor as rolling out, reads a client their own file in plain language and records what they answer, under published limits: it does not discuss fault, liability, insurance, bills, liens or what a case is worth, and it never tells a client their case has ended.
Everything syncs two ways with the firm's case management system, with named integrations for Clio, Filevine, MyCase, CasePeer, SmartAdvocate, Litify, Neos, Smokeball, Salesforce and others, plus Zapier, an API and a Model Context Protocol connection for the firm's own AI tools. Quilia states HIPAA compliance, US servers and no training of AI models on customer data; it publishes no independent security attestation of its own. Fees are set per case tier in an order form and are not published.
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 models are the engine of the thing the firm is actually buying, on a product that would still stand without them. Strip the AI out and Quilia is a competent client app: a mobile file where an injured client logs visits and pain scores, uploads records and photographs, signs documents and messages the firm, syncing both ways with the case management system. What the AI adds is the part the firm cannot staff. It reads each upload, extracts the provider, dates, parties, location, treatment details and injury descriptions, classifies images, files everything chronologically, and asks the client to confirm what it found before it lands on the case.
It tracks treatment against what is on file and surfaces gaps. An AI case assistant called ℚ, which reads a client their own file and records their spoken answers, is described by the vendor as rolling out rather than generally available. Read on the artificial intelligence, document management and HIPAA pages and in the privacy policy of 6 September 2026. Verified 20 September 2026.
Citation Accuracy and Hallucination Disclosure
Whether the vendor publishes measured accuracy on citations and assertions, grounds output to primary sources, and says plainly what its system does when it does not know. Legal has a documented public record of fabricated citations reaching filed briefs, so an untested claim of accuracy is not evidence.
No accuracy figure is published, and the checking mechanism is unusually concrete. Nothing the AI extracts reaches the case file on its own: it presents what it found to the client, the client confirms it, and the confirmation carries a receipt that can be undone for ten minutes. The source of every structured item is therefore the document the client uploaded and the client's own confirmation of it, which is as close as this shape gets to grounding a reader can follow.
The vendor also describes a pre-release test: every change to how the assistant writes is replayed against the exact inputs real cases produced, each message is graded on a fixed rubric, including whether it is specific to that client's file and whether it crossed any of the published limits, and the grader is run against itself to measure its own noise. That is a method described without its results. No error rate, no sample, nothing on what a wrong extraction costs if the client confirms it anyway. Verified 20 September 2026.
Autonomy and Oversight Model
What the system decides on its own, what a lawyer must approve, and whether the vendor documents where the review point sits. A tool that drafts under review and a tool that files without one are different products and different risks.
The controls are published in more detail than most, and the feature carrying the strongest of them is not yet generally available. On the live product the structure is clear: extracted data is proposed, not written, the client confirms each item, the confirmation can be undone, and everything lands in the firm's case management system where the legal team works. The AI case assistant ℚ adds categorical limits, stated as product constraints rather than model instructions: it does not discuss fault, liability, insurance, bills or liens; it never says what the firm is doing or intends to do; it never puts a date on anything not already in the record; the firm's phase label is withheld from the model entirely so it cannot leak one; and it never tells a client their case has ended, in any wording.
Asked what a case is worth it declines, explains why no tool can answer, and gives the client the firm's number. Those would carry this row higher, but the vendor describes ℚ as rolling out to firms rather than in general use, and an announced control is not an operating one. Worth re-reading when ℚ is generally available. Verified 20 September 2026.
Operational and Outcome Evidence
Named, dated evidence that the product works in production at real firms or legal departments. Case studies with figures and identified customers count. Unattributed testimonials and launch announcements do not.
Named firms and a survey figure, never joined to each other. Five law firms are named with the people who spoke for them: Brian Riley and Liz Lasslett of Kurtz Riley Law Group, Brandon Hewitt of Michigan Auto Law, Monica of Bay Injury Law, and Dean Tingey of Tingey Injury Law Firm, whose account is the most specific, that the product found treatment gaps the firm had been leaving money on. Separately the vendor publishes that 82% of clients would recommend their attorney because of Quilia, from 568 clients surveyed, and says plainly that this is a survey rather than a selected quote.
The sample size is given; the date, the question wording, who was asked and how they were selected are not. None of the named firms is attached to a figure for what changed, and no dated case study exists on the estate. Client testimonials on the same page are user satisfaction, not deployment evidence. Verified 20 September 2026.
Privilege and Confidentiality Posture
How client confidences are handled: attorney client privilege and work product treatment, segregation of one client matter from another, whether client data trains any model, and what the vendor commits to in writing rather than in marketing.
Strong written commitments, and one published sentence that sits badly against the product. The Terms of 22 May 2026 carry mutual confidentiality surviving three years, with trade secrets protected for as long as they qualify, and a compelled-disclosure clause requiring prior written notice and assistance in seeking a protective order. The customer owns its data. Three separate documents say customer data is not used to train AI models.
Row-level security confines a client to their own case and a firm to its own data, a Business Associate Agreement is available on request, and the servers are in the United States. Against that: the AI disclosure of 27 December 2025 tells users they are responsible for ensuring that what they put into the app does not contain confidential, sensitive or privileged information, on a product whose purpose is collecting medical records; privilege and work product are named only in the clause disclaiming responsibility for third-party AI assistants a firm connects; and the Terms assign all rights in Resultant Data to the vendor. Verified 20 September 2026.
UPL and Professional Responsibility Posture
Whether the vendor is clear that it supplies a tool rather than legal advice, who its audience is, and how it addresses unauthorized practice of law, competence and supervision duties, and jurisdiction limits. ABA Formal Opinion 512 is the reference point. Where the advice line is not the duty a product raises, the axis is read through the nearest professional duty it does raise: judicial conduct rules and the reviewing duty for products sold only to courts, and the duty to bill for time actually spent for products that draft time entries.
The advice line is drawn where it matters most, on the surface a non-lawyer actually sees. The End User License Agreement of 10 March 2026, which the injured client accepts before using the app, states that the user shall not rely on any part of the app or its output as legal advice and shall not represent to anyone that it is, that the app does not provide legal advice and creates no attorney-client relationship, and that the user should consult a qualified attorney.
The same agreement says the app is prone to error and that output should be reviewed before use. The product draws the line again in its own behaviour: the case assistant does not discuss fault, liability, insurance or liens, never states a case value or a range, and hands the client the firm's phone number instead. The Terms put professional responsibility obligations and the rules of professional conduct on the firm where it connects an outside AI assistant.
What is missing is jurisdiction: the product is sold across the United States and nothing names a state limit or addresses the firm's supervision duty over what the assistant says to its clients. Verified 20 September 2026.
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 real testing regime is published, with no owner and no findings attached to it. The vendor describes what happens before a change to how its case assistant writes can ship: the exact inputs real cases produced are replayed against the new wording, every resulting message is graded on a fixed rubric covering whether it is specific to that client's file, whether each sentence stands on its own read cold, and whether it crossed any of the published limits, and the grader itself is run against itself so its own noise is measured and a small gain is not mistaken for a real one.
Published behavioural limits sit alongside it. That is more than a principles page and it is auditable in shape. What is absent is who inside the company owns it, what the replays have found, and anything at all on whether output differs across client populations, languages or injury types, on a product whose users are often in pain, unfamiliar with the process, and served in more than one language through automatic translation. Verified 20 September 2026.
AI Safety and Data Stewardship
Retention, deletion, access control, and what happens to prompts and documents after they are processed. Whether the vendor states its subprocessors and its incident practice, or leaves the buyer to assume.
Access, infrastructure and incident practice are covered in detail; the end of the data's life is not. Published: dual-layer encryption, AES-256 at the database and AES-256-GCM at the application layer for integration credentials and session data, TLS 1.3 in transit, enforced multi-factor authentication, row-level security at the database, role-based permissions, audit trails of every upload, view and download, automated daily backups with point-in-time recovery, tested restores and stated recovery objectives, and an incident response procedure run by a named team composition of security, legal and technical staff, with supervisory notification inside 72 hours and notification of affected users.
Infrastructure subprocessors are named: Supabase for database and authentication, Vercel for the portal and API, Expo for the mobile app, Stripe for payments, PostHog and Sentry in the product. The gaps are retention and deletion: no period is stated for case content, deletion follows an account termination request with exceptions, and the AI providers are named nowhere. Verified 20 September 2026.
AI Liability and Recourse
What the vendor stands behind contractually when its output is wrong. Indemnities, caps, carve outs, insurance, and whether any of it is published or only reachable through a negotiated agreement.
Liability is handled by limitation and disclaimer, and the exposure the product creates is pushed outward. The Terms of 22 May 2026 cap the vendor's total liability at the fees paid in the single month before the event for customers paying monthly, or twelve months for annual payers, exclude consequential and indirect damages both ways, apply the caps to negligence and even where remedies fail of their essential purpose, and provide the services as is with all implied warranties disclaimed.
No indemnity runs to the customer. The client-facing agreement asks the injured user to agree not to hold the vendor liable for any loss arising from reliance on app output. Where a firm connects an outside AI assistant, the Terms put errors, omissions and hallucinations on the firm. What does exist is a service level: 95% availability measured yearly, with credits as the sole remedy, capped at 5% of fees and claimable only within 24 hours. That answers downtime, not a wrong extraction on a case file. Verified 20 September 2026.
Practice Systems Integration Depth
How deeply the product reaches into the systems legal work already lives in: document management such as iManage and NetDocuments, Word and Outlook, contract lifecycle management, matter management, e-billing, and court filing systems.
This is the strongest part of the record, and it is documented rather than listed. Named integrations with the systems personal injury work already lives in: Clio, Filevine, MyCase, CasePeer, SmartAdvocate, Litify, Neos, Smokeball, Salesforce, CasePacer, FileMaker and Neostella, each with its own page. The direction is stated: two-way sync, with documents, treatment updates, confirmed extractions and client messages flowing into the firm's case file and matter data flowing back, and the firm invites a client with one click from inside its own system.
What moves is scoped: the vendor describes API-only integration in which it stores no protected health information from the external system and syncs case metadata, with integration credentials encrypted before storage. Configuration is described too, including who does it and how long it takes, with guides published and custom routes through Zapier and an API. A Model Context Protocol connection lets a firm point its own AI tools at its Quilia case data. Verified 20 September 2026.
Deployment Model and Data Residency
Where the software runs and where the data sits. Multi tenant cloud, single tenant, private deployment, on premises, and whether region of residence is a published option or an enterprise conversation.
Both questions are answered and neither is a choice. Tenancy: a shared platform with separation enforced in the database itself through row-level security, so a client sees only their own case, a firm only its own data, and administrators only what their role allows. Residency: the privacy policy of 6 September 2026 states that the servers are in the United States, and the vendor names the providers that hold the data, Supabase for the database, Vercel for the portal and API functions, Expo for the mobile app, with daily backups held in geographically distributed storage for redundancy.
What is not published is any option or any detail beyond that. No single-tenant or private deployment, no region selection, nothing distinguishing where processing happens from where data rests, and no statement of where the AI providers process what is sent to them, since those providers are not named. Verified 20 September 2026.
Security Certifications and Trust Center
Independent attestation a buyer can pull without a sales call: SOC 2, ISO 27001, penetration test summaries, a trust center with current reports and named scope rather than a badge image.
No independent attestation of this vendor was located, and the vendor does not claim one. What it publishes, accurately, is that its infrastructure providers hold SOC 2 certification: Supabase, Vercel and Expo are named as SOC 2 certified cloud providers. That is its hosts' attestation, not its own, and the distinction is the vendor's own wording rather than a reading imposed on it. HIPAA compliance is asserted directly and in detail, with a Business Associate Agreement offered on request, but HIPAA compliance is a legal obligation the vendor states of itself rather than an audit anyone else performed.
No SOC 2 or ISO report for Quilia, no penetration test summary, no trust centre, no audit period or scope, and nothing obtainable by a buyer without asking. Checked the HIPAA and security page, the detailed HIPAA documentation link, the terms, the privacy policy, the AI disclosure and the full footer on 20 September 2026. Verified 20 September 2026.
Model Supply Chain Disclosure
Which models sit underneath, whose they are, where they run, and whether the vendor commits to telling customers when that changes. A legal buyer inherits every dependency it cannot see.
The commitments about AI providers are specific and the providers themselves are never named. The AI disclosure of 27 December 2025 says the vendor may work with various AI providers and models, selected against its security, privacy and performance standards, that any provider must meet its requirements, and that specific retention periods and handling practices vary by provider and will be disclosed. The privacy policy adds that data is not shared with AI providers for their own purposes and that AI processing is scoped to the case.
None of that is checkable, because no model, version or provider appears anywhere on the estate, nothing says where inference runs, and the promise to disclose per-provider practices is written in the future tense. By contrast the vendor names its infrastructure providers plainly, which shows the omission is a choice rather than an oversight. Nothing commits to telling customers when a model or provider changes. Verified 20 September 2026.
Commercial Transparency
Whether a buyer can learn what this costs without entering a sales process: published rates, the unit being charged, what sits behind an enterprise tier, and what implementation adds.
The shape of the bill is published in the agreement; the numbers are not. The Terms of 22 May 2026 set out the structure clearly: fees are stated in an order form, capacity is sold as a case tier with a limit on the number of cases, and exceeding it triggers case overage fees and excess user fees, with the vendor entitled to refuse the overage instead. Payment runs monthly or annually, the distinction matters enough that the liability cap differs between them, fees can be changed on 30 days' notice at the end of a term, late balances carry 1.5% a month, cancellation takes effect 30 days before the next renewal with access continuing to the end of the period, and unused subscription time is refundable pro rata.
The Terms also record that the fee may be billed on to the client as a hard cost recoverable at settlement. What no buyer can learn without a demo is any rate, any tier size, or what implementation adds. Verified 20 September 2026.
Firm and Practice Coverage
Who the product is actually built for. AmLaw, midlaw, small firm and solo, in house departments, government and courts, and which practice areas are supported rather than merely claimed.
The practice boundary is stated plainly, which is rare, and the firm boundary is not stated at all. The vendor says it is built for personal injury and workers' compensation first, that the underlying approach of collecting client information and structuring it into the case file applies to other case types, that it expands on firm demand, and that a firm in another practice area should ask where it fits today. That is a limit a buyer can act on rather than a claim of universal coverage.
Within personal injury the coverage is specific: motor vehicle, treatment-heavy claims with multiple providers, builder and provider documentation, e-signature, intake checklists and multi-case clients. What is absent is any statement of firm size or type. The named reference firms are small and mid-size plaintiff practices, there is nothing on high-volume or mass tort inventories beyond an integration with systems that serve them, and in-house and government use is not addressed, which is consistent with the product but left unsaid. Verified 20 September 2026.
7 public documents
The public pages on file for Quilia, with the recorded signals each one supports and the date it was last read. Open any of them and check the reading against the record.
-
Primary Law Corpus Provenance, Good Law Verification, Refusal and Uncertainty Behaviour
Read Sep 20, 2026
-
quilia.com/privacy-policy2 signals
Outside Counsel Guideline Readiness, Client Data in Training
Read Sep 20, 2026
-
quilia.com/terms-service2 signals
Third Party Request and Subpoena Notice, Billing and Fee Posture
Read Sep 20, 2026
-
quilia.com/ai-disclosure1 signal
Prompt and Output Retention
Read Sep 20, 2026
-
quilia.com/eula1 signal
Bar Guidance Alignment
Read Sep 20, 2026
-
Court Disclosure Support
Read Sep 20, 2026
-
quilia.com/how-it-works/hipaa1 signal
Ethical Walls and Matter Segregation
Read Sep 20, 2026
Not published; quoted per case tier in an order formUSD, as published, never converted
- Quilia does not publish a price. What it does publish, in its Terms of Service, is how the pricing works.
- Capacity is sold as a case tier. The agreement sets a limit on the number of cases, and going over it triggers case overage fees and excess user fees, with the vendor entitled to refuse the overage instead of billing for it.
- Fees are set in an order form. Firms can pay monthly or annually, and the choice matters beyond cash flow: the liability cap is one month of fees for monthly payers and twelve months for annual ones.
- Fees can change on 30 days' notice at the end of a term. Unpaid balances carry 1.5 per cent a month.
- Cancellation takes effect 30 days before the next renewal, access continues to the end of the paid period, and unused subscription time is refundable pro rata.
- The agreement also records that the fee may be billed to the client as a hard cost, recoverable at settlement.
- What a firm cannot learn without a demo: any rate, how big a case tier is, or what an overage costs.
Unit and structure published in the agreement, figure withheld everywhere. **Charged against a case tier**: the Terms of 22 May 2026 define a Case Tier Limit, and if the total number of cases exceeds it the vendor charges Case Overage Fees and Excess User Fees, while reserving the right not to permit the overage at all. **Fees are stated in an order form**, with a separate Service Allocation for authorised users, usage and data storage, and excess fees where a firm exceeds it. **Monthly or annual payment**, a distinction the agreement uses elsewhere: the liability cap is one month of fees for monthly payers and twelve for annual. **Fee changes on 30 days' notice** at the end of the initial or renewal term. **Late balances carry 1.5% per month.** **Cancellation effective 30 days before the next renewal**, with access to the end of the current billing period and a **pro rata refund of the unused portion**. **The fee may be billed to the client as a hard cost, recoverable at settlement.** A 95% annual availability commitment carries service credits as the sole remedy, capped at 5% of the fees for the period and claimable only within 24 hours.
No rate, tier size, seat minimum or implementation figure is published, and no free or self-serve tier exists for the firm.
Implementation: None published as a separate charge, and none referred to in the Terms. Onboarding is described on the integration pages rather than in the agreement: most case management integrations are set up in under 30 minutes with the vendor handling the technical configuration and the firm connecting its account, custom integrations through Zapier or the API may take longer, and step-by-step guides are published. Client onboarding is described as a single click from inside the firm's existing system, which sends the client a text with the app links. No migration, configuration or training fee appears anywhere on the estate.
Confidentiality and data terms: A Business Associate Agreement is offered on request and is not tied to a tier or a price. The HIPAA page states that the vendor can provide a BAA where an organisation requires one, that its standard BAA carries the required HIPAA provisions, and that it can be customised, with requests going to its compliance team. No charge is mentioned and no plan is named as a prerequisite. This matters more here than on most records in the index: the product's core purpose is collecting medical records, bills, treatment appointments, pain scores and injury photographs from an injured client, so protected health information is the ordinary content of the service rather than an edge case. The Terms state HIPAA compliance directly, and a separate consumer health data policy covers the state regimes that treat provider-visit data as health information.
Note: No figure is published anywhere on the estate: there is no pricing page, no pricing link in the footer, and the question is answered by a demo request. entryPriceUsd is null rather than zero, since there is no free tier for the firm; the client's app is free to the client because the firm is the customer. What puts Commercial Transparency at B rather than C is that the structure and the unit are published, though not on a marketing page: they are set out in the Terms of Service of 22 May 2026, which a buyer can read before any sales contact. One third-party listing carries a monthly figure in euros for a differently described product and is visibly mixed with another vendor's entry; it is not used. A named customer describes the product as affordable for what the firm gets, which is not a figure and is recorded on the evidence row rather than here.
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?
A public policy or trust page states no training on customer content, with no matching term located in the published agreement.
The policy pages say never; the contract says something narrower. Three published surfaces carry the flat statement: the AI disclosure of 27 December 2025 says customer data is not used to train AI models and is not exposed to public models or shared with third parties for training; the privacy policy of 6 September 2026 repeats it and adds that AI processing is initiated by the legal team and scoped to the case; the AI product page answers the same question the same way.
The Terms of 22 May 2026 commit to less. Section 2.8 permits the vendor to process customer data with machine learning and AI for the benefit of all of its customers, and its promise is that it does not train public models or inappropriately disclose customer data for such purposes, which leaves a model of its own untouched by that sentence. Section 16.1 also assigns all rights in Resultant Data to the vendor. A firm that needs never rather than not publicly should ask for it in the order form.
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 addressed and never given a period. The AI disclosure says data processed by the AI is governed by the privacy policy and that personal information is kept while the account is active; the privacy policy says information is kept as long as necessary for the stated purposes, and that when there is no ongoing need it is deleted or anonymised, or isolated from further processing where it sits in backup archives.
Deletion follows a request to terminate an account, with material retained where needed for fraud, investigations, enforcement or law. For the AI specifically, what is said is that processed data is stored under the same access controls as other case data. What a firm cannot learn is how long the text sent to an AI provider is held by that provider, since no provider is named, whether extraction inputs and outputs are kept separately from the case file, or what happens to a client's data when the matter closes rather than when the account does. No customer-set window exists.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
The product maintains its own permission model, documented, requiring the firm to keep it aligned.
Separation is enforced where it is hardest to bypass, and the vendor says where. Row-level security operates at the database itself: a client reaches only their own case, a firm only its own data, and administrators only what their role permits. Role-based permissions govern which members of a legal team can open which documents, audit trails record every upload, view and download, and integration credentials for the firm's case management system are encrypted before storage.
Two details go further than most: the client app is built around the client's own file rather than the firm's, so a client never holds a view across matters, and case delegates, the family members or helpers a client can add, are a named and bounded role rather than a shared login. What is not described is any wall between matters inside one firm, which matters where a firm acts for two clients in the same collision.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
Terms commit to notice where lawfully permitted. No transparency report located.
The commitment is in the agreement and it goes further than notice. Section 15.4 of the Terms of 22 May 2026 requires the party compelled by law to disclose confidential information to notify the other in writing promptly and before disclosing, so that the other can seek a protective order or waive its rights, to give reasonable assistance in opposing the disclosure, and to disclose only the portion legally required.
It is mutual and it is not qualified by cost beyond who pays for the opposition. Two limits a firm should hold. The clause attaches to confidential information under Section 15, not expressly to every item of case data, and the privacy policy describes disclosure to comply with a court order, subpoena or law enforcement request without attaching any notice to it. The Terms also allow suspension of the service on receipt of a governmental demand. No transparency report is published.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
No located public material identifies the corpus behind the product’s answers.
There is no legal corpus here to have provenance. The material this product works on is the client's own: the police report, the medical records and bills, the photographs, the appointments and pain scores they enter. No case law, statute or secondary source is drawn on, licensed or named, and the estate's legal glossary and state rules pages are consumer explainers rather than anything the product reads. What stands in the place of provenance is per-item: the AI proposes what it extracted, the client confirms it before it reaches the case file, and the confirmation carries a receipt that can be undone for ten minutes, so each structured fact traces to a document the client supplied and an act of confirmation.
That is recorded as the accuracy mechanism on the capability row rather than counted twice here. Nothing addresses what the underlying models were trained on.
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.
Nothing in this product cites legal authority, so there is nothing for a treatment signal to check. The output is structured case facts, treatment timelines and messages to a client, not propositions of law with citations behind them. The vendor draws that line itself in what its case assistant will not do: it does not discuss fault, liability, insurance, bills or liens, and it never puts a date on anything not already in the record.
The estate does publish a legal glossary and state-by-state rules pages, which are written for injured clients reading about their own situation rather than for a lawyer relying on them, and no currency or verification practice is described for them.
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.
The refusals are written down, in detail, as product limits rather than model instructions. The case assistant does not discuss fault, liability, insurance, bills or liens; it never states what the firm is doing or intends to do; it never puts a date on anything that is not already in the record; the firm's internal phase label is held out of everything the model is shown, so it cannot repeat one it was never given; and labels meaning a case has ended are blocked outright.
Asked what a case is worth it does not deflect to ask your firm: it says no tool can put a fair number on a case because too much of what decides it is not on the file yet, names treatment, coverage and what the attorney finds, and gives the client the office number, never a range or an example figure. What holds this short of demonstrable is that the behaviour is described rather than shown, and the assistant carrying it is described by the vendor as rolling out.
Fabricated Citation Record
Does a public court record exist addressing fabricated or hallucinated legal citations in output from this product?
No court order, opinion or disciplinary record addressing fabricated or hallucinated legal citations produced by this product has been located as of the date shown. This is a statement about the public record on that one subject, not a finding about the product, and this signal is not a litigation history.
No record was located of this product's output being found fabricated or inaccurate in a proceeding, a regulatory action or a published account. Searches on 20 September 2026 across the vendor's estate, press and directory profiles returned nothing of the kind. The shape of the risk here is different from a drafting or research tool: the product asserts no law and cites no authority, and the failure a firm should watch for is a mis-extracted provider, date or diagnosis reaching a case file after a client in pain confirmed it without reading closely. Nothing published describes such an error, and no account of one was found.
Bar Guidance Alignment
Has the vendor engaged in public with the ethics opinions its buyers are bound by?
Public materials refer to professional responsibility in general terms without naming guidance.
Professional responsibility is engaged in general terms, and no rule, opinion or bar is named. The client-facing licence agreement of 10 March 2026 states that the app does not provide legal advice and creates no attorney-client relationship, that the user must not rely on its output as legal advice or represent to anyone that it is, and that they should consult a qualified attorney. The Terms of 22 May 2026 go further in one place: where a firm connects an outside AI assistant, the firm is made responsible for deciding whether that operator's practices satisfy attorney-client privilege, the work-product doctrine, HIPAA and the applicable rules of professional conduct, and for ensuring its use complies with professional responsibility obligations.
That is the duty referred to as such, without a rule of professional conduct, an ethics opinion or any bar guidance behind it, and nothing engages the guidance on generative AI, supervision or communication with clients.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
The vendor’s charge reaches the client as a disbursement or case expense, typically advanced by a contingency firm and recovered from a settlement, so the fee question is disclosure of a pass through cost rather than compressed hours. The summary records whether the vendor addresses that disclosure.
The vendor's own agreement contemplates the client paying for this. Section 12.3 of the Terms of 22 May 2026 states that the fees for the services may be billed to the client as a hard cost, recoverable at settlement, which in a contingency practice means the cost of the software can come out of the injured person's recovery rather than the firm's overhead. Nothing published addresses what follows from that: no guidance on disclosing the charge in the fee agreement, nothing on how a per-case charge is allocated where a client has several matters or a matter has several clients, and no position on whether a cost recovered at settlement should be disclosed to the client at the point it is incurred.
The fee itself is not published at any level. The firm-facing case for the product is that better treatment documentation raises case value, which is an argument about the client's recovery rather than about the fee.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
A current subprocessor or model provider list is published.
A buyer can name most of the third parties holding the data from published pages, and not the ones processing it with AI. Named: Supabase for the database and authentication, Vercel for the web portal and API functions, Expo for the mobile app, Stripe for payment data, PostHog for product analytics in the app, portal and browser extension, Sentry for error monitoring, and the analytics and advertising tags used on the marketing site.
A Business Associate Agreement is offered on request, a data processing agreement is promised where the law requires one, breach notification runs to supervisory authorities within 72 hours, and the incident response procedure is described. The gap is the one that matters most on this signal: the AI disclosure says only that the vendor may work with various AI providers and that per-provider retention and handling will be disclosed, in the future tense, so the providers seeing case content cannot be named from anything published.
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.
A record of what the machine did exists inside the product and is never described as something a firm could produce. Each structured fact on a case file was proposed by the AI, shown to the client, confirmed by them and receipted, with a ten-minute window to undo, so per item there is a trail of a machine proposal and a human confirmation. Audit trails record every upload, view and download, and confirmed data flows into the firm's case management system by the same path as everything else.
What is absent is any export, format or guidance addressed to disclosure: nothing says the confirmation trail can be produced, nothing distinguishes an AI-extracted field from one the client typed once it has landed in the case file, and nothing addresses a court or an opponent asking how a treatment timeline was assembled. On a product built for litigation, that is the gap worth naming.