F
Formic AI

Formic AI is a Toronto responsible AI company, Formic AI Ltd., founded in 2021 and led by Ontario lawyer Daniel Escott, that sells the Formic Engine to law firms, federal regulators and other teams working with sensitive records. The Engine answers questions only from the files a customer grants it, with no model memory. Its Boreal governance layer, designed by technology chief Varun Ranganathan, breaks documents into a knowledge graph through a method Formic calls Recursive Decomposition.

A Compiler assembles each answer and a Matcher checks every claim against the source before delivery, so each sentence opens the exact page and line behind it. When the files hold no answer, the Engine says so, and audit logs record how it interpreted the question and built the response.

A five month law firm pilot at AmLaw 100 scale ran more than 150,000 questions with no hallucinations recorded, and ICI Innovations embedded the Engine in RIVAS, its major projects system, in April 2026.

The cloud subscription isolates each customer's data and serves accounts within their own jurisdictions, and an OnSite container runs offline on the customer's infrastructure. Order forms charge per enabled seat or per monthly active user, in Canadian or United States dollars.

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.

BB on AI CentralityThe models are the engine of a core capability, layered on a product that would still function without them as a document or workflow system.

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 Formic Engine is sold as AI that answers from a customer's own files. Boreal turns a question into retrieval over a knowledge graph built from those files, the Compiler assembles the response, the Matcher verifies each claim and language models supply the wording. Remove the language models and Boreal's graph retrieval and source matching would still return passages, but the composed, verified answers customers pay for would stop, which makes the models the engine of the core capability rather than the whole product.

Source: Vendor Published
BB on Citation Accuracy and Hallucination DisclosureGrounding is real and documented, with linked primary sources and a described retrieval method, short of published accuracy figures an outsider can test.

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.

Each sentence in an answer opens the exact page and line it rests on, the Matcher checks generated claims against the source material before delivery, and the Engine says when it cannot find the requested information. The architecture is billed as unable to fabricate, and a five month pilot at AmLaw 100 scale ran more than 150,000 questions with no hallucinations recorded, but the firm, review method and test set behind that figure stay unidentified, so no outsider can reproduce it.

Source: Vendor Published
AA on Autonomy and Oversight ModelWhat the system runs alone, what constrains it, and how a lawyer checks it are all published: modes, thresholds, review surfaces, and the route a matter takes back to human judgment. A categorical limit on a named mode or tier, stating what its output may not be used for, meets the threshold limb without a number.

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 Engine answers only from files the customer authorizes, with no model memory, and tells the user when those files hold no answer. Each supported sentence opens its source page and line, and the audit log shows how the question was interpreted, which passages it drew on and how the answer was built. The contract positions the product as decision support and requires direct review by a licensed professional where professional use calls for it, so judgment returns to the lawyer at every answer.

Source: Vendor Published
BB on Operational and Outcome EvidenceReal deployment evidence with substance, short of full attribution or measurement: a named customer without figures, or figures without the named customer.

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.

ICI Innovations put the Engine into production in its RIVAS Major Projects Information Management System in April 2026, where regulators and proponents work with project submissions and supporting reports. An unnamed law firm pilot ran five months with more than 150,000 questions, no hallucinations recorded, more than ten client referrals and answers in under 90 seconds. The named deployment carries no measured result, and the measured pilot carries no name.

Source: Vendor Published
BB on Privilege and Confidentiality PostureSubstantive published commitments on confidentiality and training use, short of the full picture: commonly silence on segregation between users or matters, or on what the underlying model provider may retain.

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.

Customers keep ownership of their files, and the license lets Formic process them solely to deliver the service. The privacy policy separately bars using files, prompts and outputs to train or test models. User, group and organization filters at the database level govern access, and OnSite keeps all processing inside the customer's network. Nothing ties those filters to matter walls already set in a firm's document system.

Source: Vendor Published
BB on UPL and Professional Responsibility PostureA real position is published on advice versus tooling, short of full treatment: commonly a disclaimer without the supervision and competence dimension, or silence on jurisdiction limits.

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 Engine is research support built on the files a customer supplies, and the terms reserve legal advice and professional judgment to the licensed professional and require review of output where applicable. Sentence level citations give the reviewer the underlying material. The terms speak to professional use in general, with no jurisdiction limits or guidance specific to supervision inside a law firm.

Source: Vendor Published
BB on AI Governance and Bias DisclosureA published governance framework with real substance, short of testing results or a named owner.

AI Governance and Bias Disclosure

Published governance over model behavior: who owns it inside the vendor, what is tested before release, and what is disclosed about disparate output across matter types, parties, or populations.

Boreal acts as a deterministic governance layer that checks every output against the customer's data before delivery, tags every fact with lineage parameters and logs how the question was understood, which data it drew on and how the response was built. Those logs give a reviewer a record to audit for any single answer. No named owner, release testing or comparison of results across matter types, languages or users accompanies the mechanism.

Source: Vendor Published
BB on AI Safety and Data StewardshipSubstantive published policy covering most of the ground, short of the full set: commonly no named subprocessor list or no stated incident practice.

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.

Cloud workspaces are separated by database level access filters and network segmentation, and stored files, indexes and compiled data are encrypted. On termination the customer has 30 days to request an export, after which files, knowledge graphs and indexes are deleted from production systems and backups, and enterprise officers can ask for a signed certificate of deletion. The privacy policy commits to notice of material changes in data handling. No complete processor list or breach notice deadline is given.

Source: Vendor Published
BB on AI Liability and RecourseA real published position on liability, short of the full picture: commonly a stated indemnity without scope or caps.

AI Liability and Recourse

What the vendor stands behind contractually when its output is wrong. Indemnities, caps, carve outs, insurance, and whether any of it is published or only reachable through a negotiated agreement.

The license warrants that the software meets its specifications, with repair or modification as the remedy for a verified failure. Direct damages are generally capped at the fees paid in the prior 12 months, with separate exceptions for breach of an active mutual confidentiality agreement and certain misconduct, a C$2 million cap for qualifying security losses and a patent and copyright indemnity for the core services.

The output clauses narrow that warranty for AI answers, which require professional review, so a wrong answer carries no remedy of its own.

Source: Vendor Published
BB on Practice Systems Integration DepthReal integrations exist and are documented, short of depth: named connections without a description of what they actually move.

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.

ICI Innovations embeds the Engine in RIVAS so regulatory teams can query their existing reports and check the source, and the contract licenses programmatic use with terms for embedded third party portals. A law firm testimonial mentions iManage, but no connector, setup steps, permissions model or description of what moves between iManage and the Engine is offered. A firm wanting the Engine inside its document system arranges it through the order form.

Source: Vendor Published
BB on Deployment Model and Data ResidencyDeployment model is stated clearly with partial residency detail, or residency is offered without the processing location being addressed, or the tenancy model is stated on its own with no residency detail published.

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.

The default subscription is a shared cloud service with logically separated tenant data, and Formic serves cloud accounts within their respective legal jurisdictions. Customers can instead run the Engine on a local network or on premises, and OnSite delivers an offline container on customer maintained infrastructure that Formic cannot reach, with the customer's compliance team picking the deployment. No list of cloud regions or of where model operations run comes with the cloud offer.

Source: Vendor Published
DD on Security Certifications and Trust CenterNo independent security attestation located.

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.

Stored files, indexes and compiled data are encrypted, and traffic and account data are separated by access controls and network segmentation, while OnSite moves the whole environment under the customer's control. Formic holds no SOC 2, ISO 27001 or other independent attestation for the Engine, Boreal or its model operations, and there is no trust center; only the payment processors carry certifications of their own.

Source: Vendor Published
BB on Model Supply Chain DisclosureThe supply chain is partly disclosed: providers named without change notification, or architecture described without the providers.

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.

Recursive Decomposition turns text into a knowledge graph, Boreal replaces transformer self attention with graph based symbolic processing for retrieval, the Compiler traverses the graph to assemble the response, the Matcher verifies claims, and language models only handle fluent wording, with no more than 32,000 tokens per question. That is a described architecture. The language models themselves, their providers and any notice before a change stay unidentified, and OnSite runs the same components inside the customer's network.

Source: Vendor Published
BB on Commercial TransparencyReal pricing is published for part of the range, with enterprise tiers withheld, or the unit and structure are stated without the figure.

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.

Pricing lives in the order form, which charges either a fixed price per enabled user or a monthly meter of active users and can also license programmatic use. Terms run monthly or annually, billed in advance in Canadian or United States dollars and renewing by default, with 30 days' written notice to end them. Pilots and OnSite deployments are contracted separately. No rate, tier or sample figure appears before the order form.

Source: Vendor Published
BB on Firm and Practice CoverageSegment and practice coverage is described with substance, short of the boundaries: what is supported is clear, what is not is left open.

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.

Law firms use the Engine to find and verify facts in their own matter files, and Formic describes production at AmLaw 100 scale. Federal regulators, defense bodies and major project review teams use it for submissions and supporting records, the RIVAS deployment among them, and the Canada School of Public Service partnered on a demonstration. Because it answers only from the customer's collections, coverage follows the documents a team loads rather than any jurisdiction or practice area.

Source: Vendor Published
Sources on file

4 public documents

The public pages on file for Formic 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.

Pricing

No published figure

  • The subscription price is set in the order form.
  • Formic can charge for each enabled seat or for users who are active during the month.
  • Cloud and OnSite deployments are contracted separately.
  • Pilot and programmatic use fees depend on the agreement.

Enabled users carry a fixed seat price and active users are metered each month. The order form sets a monthly or annual term and whether charges are in Canadian or United States dollars, and the firm can also contract for an OnSite container or programmatic use.

Implementation: OnSite delivery and any work to configure it are scoped in the order form.

Note: Fees are generally billed in advance and nonrefundable. Subscriptions renew monthly or annually unless the order form says otherwise, with 30 days' written notice for termination.

Legal Signals

What each signal means

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.

Confidentiality and Privilege

Client Data in Training

Can material a lawyer puts into this product be used to train a model?

Purpose limited, in the contract

The customer agreement or data processing addendum contractually limits use of Customer Data to providing the contracted service, and no surface names training either way. The limit is bound, which a policy page is not, but it is not an express training prohibition. If any surface names training in either direction, one of the other values is true and this one is not.

The license limits processing to delivering the service, and the privacy policy separately bars using files, prompts and outputs to train or test models.

Source: Vendor Publishedsolely to the extent required to deliver the Services to CustomerAs of Nov 21, 2026Evidence

Prompt and Output Retention

How long does the product keep what a lawyer typed, and can that be set to zero?

Disclosed fixed window

A specific retention period is published and the customer cannot change it.

Files, indexes and graphs are kept for a 30 day export window after termination and then deleted from production and backups, with a certificate of deletion on request.

Source: Vendor Publishedstrict safety window of thirty (30) daysAs of Nov 21, 2026Evidence

Ethical Walls and Matter Segregation

Does retrieval respect the firm’s ethical walls, or can the model read across them?

Own model, documented

The product maintains its own permission model, documented, requiring the firm to keep it aligned.

Database level access filters and isolated tenant workspaces separate customers, users and groups, and the Engine answers only from the files each one is authorized to see.

Source: Vendor Publishedaccess control filters at the database levelAs of Nov 21, 2026Evidence

Third Party Request and Subpoena Notice

If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?

Not addressed

No located term or policy addresses third party requests for customer data.

Cloud accounts are served within their own jurisdictions. No procedure covers notice to the customer before compelled disclosure of its files.

Source: Vendor Publishedwithin their respective legal jurisdictionsAs of Nov 21, 2026Evidence
Accuracy and Authority

Primary Law Corpus Provenance

Where does the law in this product come from, and does the vendor have the right to use it?

Not addressed

No located public material identifies the corpus behind the product’s answers.

Answers come only from the documents a customer authorizes, with no model memory. Formic supplies no legal research collection of its own.

Source: Vendor Publishedbased on the files you've granted it access toAs of Nov 21, 2026Evidence

Good Law Verification

Does the product tell you when the authority it just cited has been overruled?

Prompts the user to verify

The product instructs the reader to check the citation without performing the check.

Each sentence opens its source page and line for the lawyer to read. The Engine has no treatment or current status check for any authority it cites.

Source: Vendor Publishedopens to the exact page and lineAs of Nov 21, 2026Evidence

Refusal and Uncertainty Behavior

What does the product do when the answer is not in the corpus?

Documented

The vendor describes refusal or abstention behavior in public materials.

When the authorized files hold no answer, the Engine says so rather than composing one, and every answer it does give links to its source page and line.

Source: Vendor Publishedthe Engine tells you soAs of Nov 21, 2026Evidence

Fabricated Citation Record

Does a public court record exist addressing fabricated or hallucinated legal citations in output from this product?

None located

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 Engine ties every answer to a source file and passage, and no court decision in the AI Hallucination Cases database names Formic.

Source: Bar Guidance or Court RecordAs of Nov 21, 2026Evidence
Professional Responsibility

Bar Guidance Alignment

Has the vendor engaged in public with the ethics opinions its buyers are bound by?

Generic reference

Public materials refer to professional responsibility in general terms without naming guidance.

The license requires review by a licensed professional where applicable and treats output as research support. No bar opinion or law society guidance is mapped to the product.

Source: Vendor Publishedreview by a licensed professional where applicableAs of Nov 21, 2026Evidence

Billing and Fee Posture

Does the vendor address what happens to the bill when the work takes an hour instead of six?

Outside the fee relationship

The product does not touch a fee between a lawyer and a client. It operates before an engagement exists, or it is bought by a team that bills no client for the work. Savings claims aimed at the buyer’s own cost are recorded in the summary and do not make the row a savings claim, because no client bill is in the loop.

The order form bills by enabled seat or by monthly active user. The Engine's logs record research use and source lineage, not fees charged to a client.

Source: Vendor PublishedActive User basis (utility pricing metered monthly)As of Nov 21, 2026Evidence

Outside Counsel Guideline Readiness

Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?

On request only

The material exists behind a sales conversation or an executed agreement.

Enterprise officers can request a signed certificate that files and indexes were deleted, and OnSite keeps records in house. No subprocessor list or security pack is offered up front.

Source: Vendor PublishedCertificate of DeletionAs of Nov 21, 2026Evidence

Court Disclosure Support

If a judge’s standing order requires an AI disclosure, can the product produce one?

Partial record

Some elements of the record are available, short of a document level export.

Audit logs trace each statement to a source passage and record how the answer was assembled. Any statement of AI use to a court is left to counsel.

Source: Vendor PublishedAudit Logs and Sentence-Level ProvenanceAs of Nov 21, 2026Evidence
Contact

Correct a record, or ask how something was graded

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.

AI Legal Index

The AI Legal Index is an independent index that tracks changes to AI vendors in legal. It holds 670 vendors across 9 categories, each graded on the same 15 capability axes and recorded against 12 legal signals, from public sources. No vendor pays for inclusion, placement, or rating.

Index Status
Last index update
November 21, 2026
The AI Legal Index is an editorial reference. It is not a regulatory body, not a law firm, and nothing published here is legal advice or a recommendation to retain or avoid a vendor. Records are verified against published sources, bar guidance and public court records. Where a record reads not addressed, the material was not located in public sources on the date shown. See the Methodology page for evaluation standards and limitations.
© 2026 AI Legal Index
3801 N Capital of Texas Hwy, Ste E240 · Austin, TX 78746