C
Clausehound
Clausehound is a legal knowledge engine from Clausehound Inc. of Toronto, founded in 2016 by legal technologist Rajah Lehal with CTO and coinventor Joshua Koudys, and protected by United States Patent 12,670,200 on its clause taxonomy. It turns statutes, regulations and contracts into a versioned clause model that sits over the untouched originals, so a lawyer can ask what a provision said on any date. The corpus is deepest for Canada, with about 4,000 federal and provincial statutes and regulations and municipal bylaws.
AI Insights reviews contracts and regulatory documents in two modes. Guided mode applies one prompt at a time and waits for approval of each change, while autonomous mode runs a playbook across a document set and returns findings for review. A citation check flags any answer whose cited section does not exist in the version in force on the date asked, and every AI finding carries a review state recording who verified it and when.
Firms run it behind their own firewall, in their own cloud or hosted by Clausehound, bring any AI model they have approved, and connect through GraphQL, MCP tools, SharePoint and S3. Cobalt, a separate law firm, runs its diligence services on it. Licensing is arranged with the team.
Alternatives to Clausehound
Other products that do the same kind of work as Clausehound, and what each one does.
- OOne LEGALEOne LEGALE is Wolters Kluwer Italia's online legal research platform for Italian lawyers, built on legislation, commented case law, doctrine, journals and practice tools from the group's…
- AAranzadi AllegraAranzadi Allegra is the generative AI legal assistant of the Spanish legal publisher Aranzadi LA LEY, working from the publisher's own collection of close to 10 million documents of…
- LLegitQuestLegitQuest combines an Indian legal research database with tools that retrieve authorities, break down judgments and display later case treatment. iSearch finds relevant material, iDRAF…
- DDoctrineDoctrine is a Paris legal research and AI platform published by Forseti SAS, used by 27,000 legal professionals across France, Italy, Germany, Spain and Luxembourg.
- SSimpliciterSimpliciter is an AI legal research and drafting assistant for Italian legal professionals, built on a database of Italian and European law.
- LLexroomLexroom is an AI research and drafting platform for legal professionals in continental Europe, sold in Italy, Spain and Germany and working entirely in the local language.
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 versioned clause model, statute version comparison and point in time lookup are a structured knowledge base that works without language models, and its Rust core answers clause lookups in milliseconds. AI Insights adds the review work firms buy it for: guided and autonomous contract and regulatory review against playbooks, AI proposed tags and findings, and answers grounded in the versioned corpus. Machine learning on clauses dates from 2020 work with Seneca College, and large language models were built in from 2023.
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 answer resolves to a specific clause, version, section path and source, and a check flags any answer whose cited section does not exist in the version in force on the date asked. Autonomous runs test every suggestion against its source paragraph before export. Originals stay unchanged under the model, and every AI finding carries a verification state. No accuracy rate, test set or error analysis accompanies these checks.
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.
Two modes set how far the AI goes. Guided review runs one prompt at a time through report, prompts, analyze and apply steps, writes to the documents and changes nothing until each change is approved. Autonomous analysis runs every prompt against every document unattended, stays read only, checks each suggestion against its source paragraph and returns a Word report for review and export. AI proposed tags and findings stay in a review state until a person verifies them, and the app records who verified what and when. MCP access uses revocable tokens limited to specific read and fill permissions.
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.
The one named user is Cobalt, a separate law firm led by Clausehound's founder, whose lawyers verify every finding while running its Diligence Monster, DealPrep, Policysaurus and records retention services on the platform. The product has been built since 2016, joined the Seneca HELIX incubator in 2018 and filed its patent in 2022. Corpus figures such as 4,000 statutes and regulations describe the data, not a customer result, and no independent customer or measured outcome is reported.
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.
Firms can run Clausehound behind their own firewall, in their own cloud or on their own servers with an AI model they have already approved, in which case Clausehound adds no new party holding their documents. A hosted option run by Clausehound also exists, with no shared multi customer cloud. PDF parsing and OCR run in an isolated service on a separate network. No customer terms or privacy policy are posted, so commitments on training, privilege and retention, including for the hosted option, are left to the license negotiation.
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.
Clausehound Inc. states that it does not provide legal services or advice and describes Cobalt as a separate, technology focused law firm whose lawyers can be engaged on their own. The principle that AI proposes and people verify is built into the workflow for lawyers and legal teams. Supervision duties and the professional rules of any jurisdiction are not addressed.
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.
AI proposed tags and findings stay unverified until a person reviews them, the app keeps a record of who verified what and when, and citations are tested against the law in force on the relevant date. Versions are immutable and chained, with clause by clause differences between any two dates. Those are product controls, with no AI policy, release testing, accountable owner or review of uneven output across document types behind them.
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.
Deployment in the customer's own environment, an isolated parsing service, single sign on through Azure AD or OpenID Connect and revocable, permission scoped access tokens describe the controls. No privacy policy, retention or deletion terms, subprocessor list or incident practice is posted for either the self hosted or the hosted option.
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.
No terms of service, license agreement or limitation of liability is posted, and no warranty or remedy covers a wrong finding, tag or point in time answer. Licensing and engagements are agreed after a kickoff call.
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.
MCP tools such as find_statutes, list_statute_versions, search_statute_text, compare_statute_versions and fill_schedule expose the corpus over JSON RPC to Claude as a custom connector and to other clients, and a GraphQL API serves developers. SharePoint and Amazon S3 supply documents, Kafka and RabbitMQ carry events, Docker and Kubernetes package deployments, and findings export to Word with tracked changes, CSV, JSON or Excel. Document management and matter systems are not named.
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 data layer runs where the customer needs it, hosted by Clausehound, in the customer's cloud or on the customer's own servers, with no shared multi customer cloud. When it is self hosted, storage and processing sit wherever the customer runs it and the customer chooses the AI model, and parsing and OCR run in a separate isolated service. The region and provider behind the Clausehound hosted option are not given.
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.
Clausehound is described as SOC 2 audited, without the report type, auditor, period or a way to request the report, and there is no trust center. Single sign on runs through Azure AD or OpenID Connect.
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.
Customers bring an AI provider they have already approved, and Claude and other MCP clients connect through the MCP tools, so the model and its provider are the customer's choice. No default model ships with the platform, and the architecture around the model is described, from structured clause lookup that cuts token use to WebAssembly builds. The model used for the hosted option is not named.
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 price, unit or plan is posted. Buyers can license the platform, connect it to their own stack, or take an engagement in which Clausehound engineers set up and run it for the legal team, each arranged after a kickoff call with the founder.
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.
Lawyers and legal teams use it for contract review, due diligence and regulatory research, and developers build legal AI products on its API. Canadian law is covered most deeply, at the federal level, in Ontario, Québec, British Columbia, Alberta, Saskatchewan, Manitoba and New Brunswick and in municipal bylaws, across subjects from employment, health and safety and privacy to tax, securities, building codes, pensions and anti money laundering. The pipeline can take on the EU, the UK or other jurisdictions, though none beyond Canada is listed as live.
1 public document
The public pages on file for Clausehound, 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.
-
Fabricated Citation Record
Read Nov 21, 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?
No customer agreement, terms of service or equivalent contract is published on any surface located, and no policy page states a position on training. Nothing is granted and nothing is withheld, so a client has no term to hold the firm to. Where a policy page does state a position, the row takes the matching policy value instead and the summary records that no agreement exists.
No terms, license agreement or privacy policy is posted. Customers can deploy in their own environment with their own approved AI model.
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.
No retention or deletion period is stated for documents, prompts or findings, including in the self hosted deployment.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
Segregation is asserted in public materials with no published detail on how it is enforced.
Access tokens are limited to specific read and fill permissions and can be revoked, with single sign on for users. Walls between matters or clients are not described.
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.
With no terms or privacy policy posted, nothing covers legal demands for customer data.
Primary Law Corpus Provenance
Where does the law in this product come from, and does the vendor have the right to use it?
Coverage is described by jurisdiction with no identification of the underlying corpus.
The corpus covers about 4,000 Canadian federal and provincial statutes and regulations and municipal bylaws, drawn from official law and kept with every consolidation as a version, including 84 versions of Ontario's Employment Standards Act and 71 of its Insurance Act. The publishers it draws on and its update schedule are not named.
Good Law Verification
Does the product tell you when the authority it just cited has been overruled?
The vendor computes and surfaces subsequent history itself, with the method described.
Clausehound tracks every version of a statute, compares any two dates clause by clause and flags an answer whose cited section does not exist in the version in force on the date asked. Case law treatment is not covered.
Refusal and Uncertainty Behavior
What does the product do when the answer is not in the corpus?
The vendor describes refusal or abstention behavior in public materials.
An answer citing a section that is not in force on the date asked is flagged rather than presented as sound, autonomous runs test each suggestion against its source paragraph, and AI findings stay unverified until a person checks them.
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.
The AI Hallucination Cases database that Damien Charlotin maintains names neither Clausehound nor Cobalt in any case.
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.
No Law Society or other ethics guidance on AI use is referenced.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
The product sits inside a lawyer to client fee relationship and no located public material addresses billing, fee or disclosure treatment, with no savings claim published either.
Law firms use the platform for client review and diligence, and nothing covers how AI assisted review is recorded or billed. No time savings are claimed.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
No located public material supports a client side disclosure obligation.
No subprocessor list or client disclosure material is offered. A self hosted deployment adds no new party holding documents, since the customer chooses the AI model.
Court Disclosure Support
If a judge’s standing order requires an AI disclosure, can the product produce one?
Some elements of the record are available, short of a document level export.
The app records who verified each AI finding and when, citations name the statute version relied on, and findings export with tracked changes or as a Word report. The record does not name the model used for each finding.