F
Foundation AI
Foundation AI runs the document operations of high-volume law firms: everything that arrives by post, e-fax, portal, shared mailbox or individual inbox is captured, read, matched to a matter, named to the firm's convention, filed, routed and tasked, with every step tracked. The platform combines fine-tuned language models, retrieval, vector search, jurisdiction-specific taxonomies and deterministic guardrails, evaluating each document from several angles and reconciling the results into a calibrated confidence score for every field, so confident results pass straight through and uncertain ones are escalated to a reviewer rather than guessed.
Extracted data drives downstream work such as calendaring and updating treatment and damages records, and dashboards and BI feeds show where documents are. Integrations cover CASEpeer, Clio, Filevine, Litify, SmartAdvocate and Smokeball. The customer base is concentrated in personal injury, workers' compensation, SSDI, VA and insurance defence practices, with a separate Claims Operations product for insurers. Foundation Inc., trading as Foundation AI, is a Delaware corporation based in Irvine, California.
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.
Take the models out and what is left is a mailroom. The platform's published architecture is the product: fine-tuned language models, retrieval, vector search, business logic, jurisdiction-specific taxonomies and deterministic guardrails evaluating each document from several angles, with the results reconciled and calibrated into a confidence score for every field. Classification, matter matching and extraction are all model work, and the vendor argues the point directly, saying general-purpose models look impressive in demos and fail silently in production because categories overlap, taxonomies are hierarchical and a wrong model sounds as confident as a right one. 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.
For an extraction product the question is whether a value traces back to the document and whether the system knows when it is unsure, and both are addressed with real method. Every classification, match and extraction is evaluated against calibrated confidence thresholds, results from several analytical routes are reconciled, and the confidence attaches per field rather than per document. Matching is described the same way, using names, dates, addresses and claim and policy numbers through custom models, retrieval and guardrails.
What is not published is a number: no accuracy, precision or error rate from the vendor, no test set and no evaluation an outsider could check. The only figures are customer-reported, such as the 98 per cent accuracy quoted in the Stockwell Harris story. 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.
What runs alone, what stops it and how a person checks it are all published together. Confident results move straight through; anything below the calibrated threshold is held and surfaced in the review interface with the uncertainty highlighted, and on matter matching the vendor states plainly that when confidence is low it does not guess but offers the most likely matches for a person to confirm. The escalation route is described at the organisational level too, with most firms centralising review in an operations team rather than spreading it across case managers.
Execution is bounded by firm-defined rules for routing, tasking, deadlines and escalation, and every action is tracked. The vendor's own summary is that the platform does not remove humans from the process, it removes the wrong work from humans. 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, named people, described deployments and figures, at a stated scale. Seven customer stories run on the homepage with the customer, the individual and their title attached to each outcome: Nyman Turkish at more than 12 times the document processing productivity, Miller Dawson Sigal & Ward down from five or six hours a day to fifteen minutes, Acumen Law from about 120 hours a week to roughly five, Stockwell Harris at five times faster with 8,000 dollars a week saved and 400,000 dollars of temporary staff spend avoided.
The Floyd Skeren study, read in full, sets out the firm's prior process across ten offices, the system it integrates with, the staffing change and the results, 350 per cent more document processing efficiency and 81 per cent fewer non-billable hours, with a downloadable version. Around forty firm logos appear across the platform and integration pages, and the vendor states seven years of operations, hundreds of firms and millions of documents a month. What no story gives is the measurement method behind the percentages. 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.
This product takes in every document that reaches a firm, medical records, liens, pleadings, demand letters and client correspondence, and nothing published says what happens to them. There is no customer agreement on the estate: the Terms and Conditions govern the website and the Privacy Policy covers website visitors and their contact details by its own terms. No statement was located on whether firm documents train or improve the models, how long they are held, whether they are segregated between firms, or what happens to them when a customer leaves.
The one adjacent claim, made in a feature block on a case study page, is that the company is SOC 2 Type 2 certified and HIPAA compliant. Checked the homepage, the platform page, the Filevine integration page, a full case study, the terms and the privacy policy on 20 September 2026. 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.
Nothing published addresses the professional dimension of the work. The product makes decisions that bear directly on a lawyer's duties, classifying a pleading, matching a document to a matter, deciding who is alerted to a time-sensitive notice, and the estate frames the risk in operational terms, missed deadlines, misfiled documents and missed liens, without ever addressing supervision, competence or what the firm remains responsible for.
There is no statement that the output is not legal advice, no description of what a reviewer is expected to check, and no guidance on the firm's responsibility when a document is routed to the wrong person. Checked the homepage, the platform page, the Filevine integration page, a full case study, the terms and the privacy policy on 20 September 2026. 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.
The approach is published in detail and the governance around it is not. The platform page explains how model behaviour is controlled, several analytical routes reconciled against each other, deterministic guardrails, jurisdiction-specific taxonomies, calibrated confidence thresholds and human validation of low-confidence results, which is more architectural transparency than most vendors offer. Nobody is named as accountable for model behaviour, nothing describes what is tested before a model or taxonomy change ships, no evaluation results are published, and nothing addresses whether accuracy falls unevenly across document types, practice areas, languages or handwriting quality. 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.
Nothing published covers the handling of the documents the platform processes. No retention period, no deletion commitment, no encryption statement, no access control description, no subprocessor list and no incident practice was located; there is no security or trust page on the estate, and the privacy policy addresses website visitors, saying only that reasonable administrative, physical and technical controls are adopted and that the company is GDPR compliant.
The certification claim on a case study page is the closest thing to a statement. Checked the homepage, the platform page, the Filevine integration page, a full case study, the terms and the privacy policy on 20 September 2026. 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.
Nothing published says who bears the loss when a document is misfiled or an alert never fires. The Terms and Conditions cover the website: they disclaim warranties, exclude consequential damages, cap liability at whatever the user paid to access the site, and require the user to indemnify Foundation AI. No customer agreement, service level commitment, accuracy warranty or indemnity for the product was located, which leaves the risk the vendor's own marketing names, missed deadlines and missed liens, allocated nowhere on the published record.
Checked the terms, the privacy policy, the homepage, the platform page and a full case study on 20 September 2026. 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.
Integration is the point of the product and it is documented per system. CASEpeer, Clio, Filevine, Litify, SmartAdvocate and Smokeball each have their own page; the Filevine page describes what actually moves, matching documents against parties, case numbers, claim numbers and dates held in Filevine, filing into matter folders, alerting the people the firm nominates, and writing extracted fields such as parties, senders, treating providers and dates back into Filevine to drive calendaring.
Elsewhere the platform tags the originating message in Outlook or Gmail, matches Salesforce objects and Filevine collections, and feeds Domo and Power BI. What is missing is the engineering detail: no API or developer reference, nothing on authentication, sync direction or failure handling, and no named certification in any partner marketplace. 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.
Nothing published states how or where the platform runs. No hosting provider, region, tenancy model or residency option appears anywhere, and the only deployment language found is a claim on a case study page that the software integrates with existing hardware and core systems. For a platform that ingests a firm's entire inbound document stream, including medical records, a buyer has nothing to evaluate. Checked the homepage, the platform page, the Filevine integration page, a full case study, the terms and the privacy policy on 20 September 2026. 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.
The certification is claimed, and only just. A footer badge links to the AICPA's general SOC page on every page of the site, and a feature block on a case study page states that Foundation AI is SOC 2 Type 2 certified and HIPAA compliant. There is no security page, no trust centre, no auditor named, no report period or scope, and no route to request the report; the HIPAA claim, which matters for a product handling medical records in injury and workers' compensation matters, appears only in that one block.
Checked the homepage, the platform page, the integration page, a full case study, the terms and the privacy policy 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 architecture is described and the models are anonymous. The platform page names the components, fine-tuned large language models, retrieval, vector search, custom models, deterministic guardrails and proprietary methods, which tells a buyer more about the shape of the system than most vendors do, while identifying no model, version or provider and saying nothing about where inference runs. The vendor's argument that general-purpose models fail in production implies models of its own, and no commitment to notify customers when they change was located. 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.
No pricing information is published at any level, including the unit of charge, which matters here because the natural units, documents, pages or mailboxes, would tell a firm a great deal. Checked the homepage, the platform page, the six integration pages linked from it, the customer stories index, a full case study, the about page, the terms and the privacy policy on 20 September 2026: there is no pricing page, no tier, no rate and no trial, and every route ends at a demo booking or a call. 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 coverage is stated concretely by practice and by document type. The vendor publishes models, taxonomies and workflows for personal injury, workers' compensation, SSDI, VA, property claims, bankruptcy and immigration, and names the document classes it is trained to recognise, correspondence, discovery demands, pleadings, medical reports and event notices, with a separate Claims Operations product for insurers. The customer roster matches the claim, dominated by injury, workers' compensation and insurance defence firms from solo practices to Morgan & Morgan.
What is not published is the boundary: no statement of which practice areas, document types or firm sizes the platform handles poorly, and nothing on jurisdictions outside the United States. Verified 20 September 2026.
4 public documents
The public pages on file for Foundation AI, 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.
-
foundationai.com/platform.html3 signals
Ethical Walls and Matter Segregation, Refusal and Uncertainty Behaviour, Court Disclosure Support
Read Sep 20, 2026
-
Billing and Fee Posture
Read Sep 20, 2026
-
Third Party Request and Subpoena Notice
Read Sep 20, 2026
-
Fabricated Citation Record
Read Sep 20, 2026
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 published agreement or policy exists and none of it addresses the question either way, or the document that would answer it could not be read and the summary names the retrieval limit. The summary states which shape the silence takes: an improvement right granted that never names training, or no improvement right granted at all.
No position either way was located. There is no customer agreement on the estate; the Terms and Conditions govern the website and the Privacy Policy covers website visitors. The platform description says the models are fine-tuned and that the system learns from validation, without saying whose documents that learning uses. Checked the homepage, the platform page, the Filevine integration page, a full case study, the terms and the privacy policy on 20 September 2026.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
No located public material states how long prompts and outputs are retained.
Checked the same pages on 20 September 2026. Nothing states how long the platform keeps the documents it processes, the extracted data or the tracking record of every action taken, and no deletion commitment on termination was located. The privacy policy's retention paragraph addresses personal information collected through the website.
Ethical Walls and Matter Segregation
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.
The platform routes documents by role and matches them to matters, contacts, providers and staff roles, so it plainly holds a view of who should see what, but nothing published describes a permission model, whether it inherits the practice management system's access controls, or how a matter walled off inside a firm is kept out of the wrong queue.
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.
Nothing addresses what happens if a customer's documents are subpoenaed from Foundation AI. The only disclosure clause located sits in the website terms and concerns material submitted through the site, which the vendor may disclose to comply with legal obligations or governmental requests, without notice or attribution; it does not reach the documents the platform processes for a firm.
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.
Checked the homepage, the platform page, the Filevine integration page and a full case study on 20 September 2026. The platform works on the customer's own inbound documents against the vendor's jurisdiction-specific taxonomies; no external legal corpus is involved, and the taxonomies' sourcing is described only as seven years of document operations practice.
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.
Checked the same pages on 20 September 2026. The product classifies and files documents rather than citing legal authority, so no subsequent-history check arises and none is described.
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 abstention path is described plainly and it is the centre of the product's design. Every match, classification and extraction is scored against calibrated confidence thresholds; confident results pass straight through and the rest are held, with the uncertainty highlighted in the review interface for a person to resolve. On matter matching the vendor states that when confidence is low the system does not guess, and instead surfaces the most likely matches for the team to confirm. What would lift this further is a published evaluation or an observable demonstration of that behaviour.
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.
Searched the AI Hallucination Cases database maintained by Damien Charlotin on 20 September 2026 on the name Foundation AI. No court order, opinion or disciplinary record naming the product was located. This is a statement about the public record rather than a finding about the product, which files documents rather than drafting filings.
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.
Checked the homepage, the platform page, the Filevine integration page, a full case study, the terms and the privacy policy on 20 September 2026. Nothing engages a lawyer's professional duties. The risks the vendor names, missed deadlines, misfiled documents and missed liens, are framed as operational and financial, and no rule of professional conduct, ethics opinion or bar guidance is mentioned.
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, and the product sits inside a fee relationship between a lawyer and a client where those savings would change the bill.
The published case is about the firm's own economics rather than the client's bill: an 81 per cent reduction in non-billable hours, staff redeployed to billable work, and a partner quoted on reducing non-billable hours while increasing billable ones. The work the platform replaces is overhead a firm generally cannot bill, and nothing published addresses whether the platform cost is passed through to clients as a case expense.
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.
Checked the homepage, the platform page, the Filevine integration page, a full case study, the terms and the privacy policy on 20 September 2026. No subprocessor list, model provider list or client-facing disclosure material was located, and there is no trust page or request route, so a firm asked by an insurer client what technology reads its claim documents would have nothing published to hand over.
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 real processing record exists, built for firm oversight rather than for a court. The platform tracks each document and every action taken within it, from capture, matching, classification and validation through filing, routing, task creation and delivery, with per-field confidence scores, built-in dashboards, custom reports and feeds into Domo and Power BI. What is not recorded is which model produced a determination, so the record shows what happened and who confirmed it rather than what the AI did.