Filevine is case management software for law firms and legal teams, built around the life of a matter from intake to settlement. Its AI layer, LOIS, includes MedChron for medical chronologies, DemandsAI for demand letters, Depo CoPilot for live deposition help, and drafting inside Microsoft Word. Filevine says its customers draft about 100,000 demand letters a month with it. The buyer base is wide: personal injury and other plaintiff firms, defense and family practices, prosecutors and public defenders, government agencies and in house teams, from solo to big law.
What sets it apart is the contract. Third party model providers are barred from training on or keeping customer data, and users must check citations before output reaches a client or a court.
Side by side on the same 15 capability axes and 12 legal signals: where the two part company, and what neither one publishes.
Vendors in the same categories as Filevine whose graded profiles look most alike, with a line on what each product does.
Similar is not the same, so check that each one does the job you need. No vendor pays for placement.
Ordered by how closely the graded profiles match across the 15 capability axes, inside the same categories. See how each axis is graded in the methodology.
All 15 axes, graded from public sources on the date shown. Hover a grade to see what the letter means on that axis.
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.
Filevine names more than a dozen AI products and sells them on top of a complete practice management system. They include MedChron for medical chronologies, DemandsAI for demand letters, Depo CoPilot, Depo Summaries, AIFields, AI Doc Review, AI Data Mapping, ValidationAI, ImmigrationAI, Leads AI, Ask LOIS, LOIS for Word and the LOIS Console. The medical chronology and the demand letter are core work in a personal injury practice, and the AI drives both.
Underneath sits the platform Filevine sold before the AI: matters, documents, intake, billing, electronic signature, analytics and deadlines. Without the models, a firm keeps running on that platform. The AI is sold as packages and extensions on top of it, with its own free tier.
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 claimed across products and accuracy is not measured. Ask LOIS returns grounded, traceable answers, LOIS for Word gives source linked recommendations, Depo Summaries carry page and line references, and the free LOIS Explore tier cites every result to source. A legal research opinion reader lets a user open the authority. The agreement names the failure modes: output may be incomplete, contain factual or legal errors, reflect bias, or differ for similar inputs.
What is missing is any number. No accuracy figure, test set, evaluation or error rate is published, though AIFields marketing invites buyers to stop struggling with hallucinations. MedChron's output points to sources only when applicable.
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 review duty is in the contract, and it names the act required. Section 3.2 of the subscription agreement makes the customer responsible for reviewing and verifying output, and for confirming citations and legal conclusions before relying on it or sharing it with a client or a court. An AI Acceptable Use Policy is part of the agreement, AI drafting is described as human verified, and outputs link back to the underlying material.
The agreement also warns that similar inputs may produce different output. What is missing: no abstention behavior, no confidence indicator, no threshold at which a feature declines, and no route to report a wrong output.
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.
Volume claims, not customer outcomes, are what Filevine puts forward. It says customers draft about 100,000 demand letters a month, and claims the largest and most accurate personal injury case management data repository in the world. One case study headline names a shareholder who brought 125 active cases under control, without naming the firm. The agreement lets Filevine use a customer's name and logo, with a right to withdraw. No measured outcome for a named organization is published. Filevine has a customers and case studies library.
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.
Confidentiality and training are both covered by commitments in Filevine's agreement. Third party AI providers are barred in writing from training on customer data and from keeping it beyond the request, except briefly for abuse monitoring. The customer owns its data, including output, and customer data is confidential information for seven years, with trade secrets protected indefinitely. Filevine cannot reduce its confidentiality or security commitments mid term without consent.
Another customer's queries and output are not the customer's data, which separates tenants. What is missing is separation inside a firm: nothing describes a wall between two matters, which is live for a plaintiff firm with adverse cases.
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 clearly in the agreement. Filevine is not a law firm and does not give legal advice, and using output creates no attorney client relationship. The customer must confirm citations and legal conclusions before relying on output or sharing it with a client or a court, which names the duty and the audience. An AI Acceptable Use Policy governs permitted uses. What is missing: no bar or ethics authority is named, including ABA Formal Opinion 512, and no jurisdiction limit is stated for a platform used in immigration, criminal defense and prosecution work across every state.
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.
Fairness, reliability and safety, privacy and security, inclusiveness, transparency and accountability make up Filevine's six AI principles. The fairness principle commits to fighting bias in its legal AI tools. The agreement repeats the point where it binds, stating that output may reflect biases in training data. Governance has some structure: an AI Acceptable Use Policy is part of the agreement, Filevine keeps a list of permitted models, and it cannot reduce its confidentiality or security commitments mid term without consent.
A Head Legal Futurist and a Legal Futurist are named roles. What is missing is testing: no evaluation before release, no published result, and no bias audit or fairness measure.
Retention, deletion, access control, and what happens to prompts and documents after they are processed. Whether the vendor states its subprocessors and its incident practice, or leaves the buyer to assume.
The Data Protection Agreement (updated 14 Jan 2026) covers retention, return, access, subprocessors and breaches. Personal information is kept for the life of the service, then destroyed or returned on the customer's written direction. If law forces Filevine to keep something, it must tell the customer what, why and when it will be destroyed, and certify destruction within thirty days. Access is limited to staff who need it, with training and background checks where lawful.
Subprocessors sign terms at least as protective, the list sits in Annex C, customers can subscribe to change notices and object within ten business days, and Filevine is liable for its subprocessors. Breaches are notified promptly, with the data affected, the number of people, the steps taken and a named contact, and Filevine bears reasonable costs. What is missing is a fixed notice deadline in hours.
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.
Filevine publishes a full liability position on four fronts. It defends and indemnifies against third party claims that the service infringes a US patent or copyright or misappropriates a trade secret, with remedies of a license, a fix, or termination and refund. Total liability is capped at the greater of twelve months of fees or two hundred thousand dollars, so smaller customers get a floor. The cap reaches claims from a security breach of customer data, gross negligence and intentional misconduct sit outside it where law requires, and the infringement indemnity is uncapped.
Filevine commits to insurance from carriers rated A minus, VI or better by A.M. Best. It also warrants that the service conforms to its documentation and complies with law, with 99.0 percent uptime and service credits. What is not covered: output carries no warranty and is excluded from the indemnity.
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.
Named integrations connect the platform to other tools. LOIS for Word puts drafting and redlining inside Microsoft Word with source linked recommendations. Microsoft 365 integration is reported, DataBridge gives real time access to Filevine data, Vinesign has a public API, and native and partner integrations are sold as an extension. The agreement grants API access. Deposition scheduling books court reporters and creates Zoom links.
What is missing is depth: nothing says what moves, in which direction or with what setup, and nothing is published for iManage, NetDocuments or ebilling. Filevine has an integrations and API page.
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.
Amazon Web Services hosts the platform in the cloud, named on Filevine's case management page and pricing FAQ, with separate US and Canadian login domains. The Data Protection Agreement names Filevine, Inc. in Salt Lake City, Utah as the data importer and names a Data Protection Officer. Transfers from Europe, the UK and Switzerland run on Standard Contractual Clauses and the UK addendum, with supervisory authorities listed.
Government access requests are notified to the customer only. What is missing is a residency choice: no region list, no selectable location, no tenancy model, and no statement of where AI inference runs, which matters because third party AI providers process customer data.
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.
Filevine claims SOC 2 Type II, SOC 3, HIPAA, CJIS, HITECH and GDPR on a structured security page. Its Data Protection Agreement adds the route behind the claims. An independent third party audits data processing and security controls at least once a year. Audit reports are available on the customer's written request, under confidentiality, and issues are handled through a corrective action plan. If no audit happened in the past year, or Filevine declines to share reports, the customer gets a direct audit right on 21 days' notice.
Filevine also answers a security questionnaire once a year within six weeks. What is missing: no auditor, report date or coverage period is published, and the certifications carry no scope.
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 agreement confirms third party models are used and does not say whose. Section 3.1 says AI features use models from third party AI providers, and section 4.3 commits Filevine to keep a list of permitted models. No model, provider or version is published. The AI principles page calls Filevine's AI proprietary, and nothing reconciles that with the agreement. The providers are bound not to train on customer data or keep it beyond a request, except briefly for abuse monitoring. Material changes need thirty days' notice, though nothing is specific to models. Filevine links a Subprocessors page.
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 full shape of the pricing is public, with no figure attached. Filevine's pricing page lists the LOIS AI tier with seven components, four platform packages (Matters, Intake, Depositions and Signatures) with their features, and five Matters extensions. A free tier, LOIS Explore, needs no credit card and covers document questions, AI drafting, legal research and the opinion reader, but does not connect to case management.
The agreement adds billing terms: interest at up to 1.5 percent a month after ten days overdue, suspension at ten days, a ten day window to dispute a charge, retroactive overage billing from usage telemetry, and no refunds outside warranty or breach. Implementation, configuration and data migration are done by separate certified partners for their own fees. What is missing is any number, at any tier. Every route to a price is a demo or a custom quote.
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 runs across seventeen segment pages. By practice: personal injury, family law, mass torts, immigration, insurance defense, criminal defense and estate planning. By government tier: prosecutors, public defenders, states and municipalities, attorneys general and federal agencies, backed by agreement clauses on sovereign immunity and FAR and DFARS designations. By corporate use: in house counsel. By size: solo, small, mid size and big law.
ImmigrationAI handles documents in 170 languages, and Timely generates court deadlines for any jurisdiction in the country. What is missing is any limit: no unsupported practice, matter type, jurisdiction or scale is named, and the deadline claim has no coverage table.
The public pages on file for Filevine, 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.
Client Data in Training, Prompt and Output Retention, Third Party Request and Subpoena Notice and 3 more
Read Aug 31, 2026
Fabricated Citation Record
Read Aug 31, 2026
Billing and Fee Posture
Read Aug 31, 2026
A full structure is published on the pricing page and in the subscription agreement dated 12 August 2026, with no figure at any level. All packages are custom built, and buyers are sent to sales for a quote. The LOIS AI tier has seven components: LOIS for Word, Ask LOIS, AI Drafting, Depo Copilot, Depo Summaries, Depo Library and MedChron. Four platform packages follow: Matters (matter and document management, document assembly, payments), Intake (Lead Docket, Lead Docket Automation, Leads AI), Depositions (scheduling and hosting with court reporter booking, Depo Copilot, Depo Library, Deposition Summaries AI) and Signatures (Vinesign).
Five Matters extensions are named: Periscope analytics, Timely deadlines, a Client Portal, DataBridge and Integrations. The free LOIS Explore tier needs no credit card and covers document upload and questions, AI drafting cited to source, legal research, the opinion reader and the LOIS Console with Skills, but not case management. From the agreement: fees are set in a Sales Order. Amounts more than ten days overdue accrue interest at the lesser of 1.5 percent a month or the legal maximum, and may trigger suspension and acceleration.
A charge must be disputed within ten days of invoice. Overages found by telemetry may be billed retroactively at list rates. Fees are nonrefundable except for a warranty failure or Filevine's uncured breach. Implementation, configuration and data migration are done by separate certified Implementation Providers under their own agreements and fees, and not engaging one does not pause subscription fees.
Note: No figure is published at any tier, so the entry price is left empty. It is not shown as zero, because LOIS Explore is a free tier of a paid product and the lowest paid rate is not published. The free tier is described in the pricing basis and summary instead.
LOIS Console can now build and edit PDFs through chat. Users can assemble demand packets and exhibit packets, apply Bates numbers and edit PDF text without leaving the console.
LOIS for Word now lets users mark any playbook check as a Deal breaker for terms where any change should stop the deal. The flag sits alongside the High, Medium and Low ratings LOIS assigns, rather than replacing them.
LOIS for Word gained a built in viewer for Excel and CSV files, with sheet tabs, search and zoom. When LOIS cites a spreadsheet, the rows it relied on are highlighted, so the citation can be checked without opening the file in another app.
A 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.
Can material a lawyer puts into this product be used to train a model?
The published terms prohibit training on customer content. Not a policy page, the agreement.
The subscription agreement (updated 12 August 2026) bars Filevine's third party AI providers, in writing, from using customer data to train or improve their models, and from keeping it beyond the request except briefly for abuse monitoring or trust and safety. Filevine keeps a list of permitted models consistent with that. A separate clause lets Filevine use customer data in deidentified, aggregated form to improve its services and other technologies, with a commitment not to attempt reidentification, so the two clauses sit side by side.
How long does the product keep what a lawyer typed, and can that be set to zero?
A specific retention period is published and the customer cannot change it.
Filevine's Data Protection Agreement (updated 14 Jan 2026) sets retention for the life of the service. On termination it destroys personal information or returns it on the customer's written direction. If law forces it to keep something, it must say what, on what basis, and when it will be destroyed, and it certifies destruction in writing within thirty days. These terms cover personal information, not prompts and AI output as a class, and nothing says whether AI conversation threads are kept separately or for how long. The subscription agreement sends retention questions to the Security Addendum.
Does retrieval respect the firm’s ethical walls, or can the model read across them?
No located public material addresses walls or matter level segregation.
Nothing published describes a wall between matters: no matter level permission model, and nothing saying a user can be restricted to certain cases or that AI features respect such a limit. One clause separates customers: another customer's queries and output are not the customer's data. Users must hold unique credentials. The question is live here, since Filevine targets plaintiff firms with adverse matters and both public defender and prosecutor offices, and nothing says whether one matter's material can surface in another's AI output.
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.
Two documents address this. The subscription agreement allows disclosure of confidential information under law, court order or government order, with advance notice to the customer where legally permitted. The Data Protection Agreement goes further for personal information: Filevine will first tell the customer of the legal requirement and give it a chance to object or challenge, unless the law forbids notice, and will disclose only the minimum needed.
On a government access request it notifies the customer, not the data subject. Two limits: the chance to object is qualified where it would prejudice Filevine or expose it to liability, and the DPA covers personal information rather than every document in a matter. No transparency report is published, and nothing commits Filevine to seek a protective order itself.
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.
The question arises here and nothing answers it. Most of Filevine's AI reads the customer's own case file, but the free LOIS Explore tier offers legal research and an opinion reader, so the platform retrieves primary law from some corpus. Nothing names the source, supplier, jurisdictions, license or update cadence.
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.
Filevine offers legal research and an opinion reader in LOIS Explore, and its drafting tools cite results to source. No citator, treatment signal or check that cited authority is still good law is described. The agreement puts that burden on the user, who must confirm citations and legal conclusions before relying on output. Filevine also runs a webinar on checking cites and reading the opinion.
What does the product do when the answer is not in the corpus?
No located public material addresses what the product does when it cannot ground an answer.
Filevine's agreement states the limits frankly: output may be incomplete, contain factual or legal errors, reflect bias in training data, or differ for similar inputs. It does not describe what the system does when it is unsure: no abstention path, confidence indicator, threshold for declining, or behavior when the case file does not support a question is described.
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.
The AI Hallucination Cases database maintained by Damien Charlotin, which held about 1,994 decisions at the time, and several independent 2026 sanctions trackers and practitioner guides record no court order, opinion or disciplinary record naming Filevine or its products. The exposure is direct, because Filevine drafts demand letters and motions and offers legal research, and the agreement requires users to confirm citations before sharing output with a client or court.
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.
No authority is cited: nothing names ABA Formal Opinion 512, state bar guidance on generative AI, or a court standing order. The agreement does state the substance: Filevine is not a law firm, output creates no attorney client relationship, and the customer must confirm citations and legal conclusions before sharing output with a client or court. Filevine also publishes an AI Acceptable Use Policy and a webinar and blog series on AI risk.
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, and the product sits inside a fee relationship between a lawyer and a client where those savings would change the bill.
Savings come with numbers attached. AIFields is said to cut costs by over 75 percent, demand letters take 15 minutes or less, DemandsAI is sold on per case fees with no charges for complexity or edits, and the AI page says firms can pass savings on to clients. No record of AI assisted work per matter for fee purposes, and no guidance on billing or client disclosure, is published. Billing and time keeping are core modules, so the means to record it exist. The gap is sharpest on the demand letter, which in a contingency practice is priced into the recovery.
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
A subprocessor and model provider list plus client facing disclosure material is published or available without an agreement in place.
What outside counsel guidelines ask is answered in Filevine's Data Protection Agreement. Subprocessors sign terms at least as protective, and the list is published at a URL reached from Annex C. Customers can subscribe to notices, object within ten business days, and terminate the affected service with a refund if Filevine cannot work around an objection. An independent party audits Filevine every year, reports are available on request, and a direct audit right applies if that fails.
Filevine answers a security questionnaire yearly within six weeks. Transfers run on Standard Contractual Clauses and the UK addendum, and a Data Protection Officer is named. A Security Addendum and an AI Acceptable Use Policy are also published.
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 agreement expects output may reach a court, requiring the customer to confirm citations and legal conclusions before sharing output with a judicial body. Product support is partial: drafts cite sources, LOIS for Word links recommendations to sources, deposition summaries carry page and line references, and the opinion reader opens the authority. The record is missing: no model identification or versioning, no log of AI use or of what a user accepted, and no export or template for a court's standing order on AI is described.
Every grade and every signal on this index is drawn from public sources and dated. If a record is wrong, out of date, or missing an artifact the index did not locate, send the source and it will be reviewed and the record redated. Vendors are welcome to submit documentation. Nothing on this index is for sale, including a listing, a placement, or a grade.