I
Ironclad

Ironclad is a contract lifecycle management platform for in house legal, procurement, sales, finance and IT teams, covering intake, drafting, no code workflows, approvals, negotiation, eSignature, the repository and analytics. Its AI layer runs through Ironclad Agent, a conversational front door that finds contracts by meaning and answers questions about the portfolio. It calls specialized agents for intake, contract review, renewals, analytics, contract families and workflow design.

Contract Review Agent, formerly Jurist, redlines against a company's playbooks and past negotiated agreements, flags risk and drafts, in Ironclad or in Microsoft Word. MCP connectors bring repository answers into Claude, ChatGPT and Slackbot. The agents draw on models from Google Gemini, OpenAI and Anthropic, and Ironclad trains its own models only on data from customers who opt in. Founded in San Francisco in 2014, it reports more than 2,000 customers and over 2 billion contracts processed.

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.

Intake, the no code Workflow Designer, approvals, signature, the repository and analytics run without AI, and Ironclad sells that system of record as the base of the platform. Ironclad AI sits on top of it. Ironclad Agent takes a plain language request and returns the answer, the contract or the workflow, or hands the task to a specialized agent. Contract Review Agent redlines third party paper against playbooks and precedent, Intake Agent pulls terms into launch forms, Renewal Agent briefs upcoming renewals, and Analytics Agent charts contracting data.

The agents act on live workflow state, such as where a contract sits in approval, as well as on documents. The contract platform predates generative AI and would still work without the models.

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.

Ironclad's AI Systems team published how its extraction is grounded in May 2026. The model returns each value with a description of where it sits in the contract. A second pass ties that value to the exact OCR tokens, so the interface highlights the passage with its page and position. If the model returns no match, the system tries the raw text and then the normalized value before giving up on a location. Contract Review Agent bases redlines on the customer's playbooks, templates and precedent agreements, and redlines can show their rationale.

Answers through the MCP connectors link back to the contract in Ironclad. No accuracy figure, error rate or test set is published, and the AI Addendum says output may be incorrect or inaccurate.

Source: Vendor Published
BB on Autonomy and Oversight ModelA written commitment that the models work alongside a supervising lawyer, with real review surfaces, short of the full control structure: commonly the threshold at which the system stops or what happens after it is wrong.

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.

Ironclad Agent routes each request to the right agent or workflow, and the agents work inside existing contracting workflows under the user's permissions. Contract Review Agent proposes redlines, flags risk and applies positions in Ironclad or directly in Microsoft Word, with the final decision left to the reviewer. Playbooks set preferred positions and fallbacks, and answers through the MCP connectors are limited to contracts the user can already open.

The AI Addendum makes the customer responsible for reviewing and validating output before using it, and says the AI products are no substitute for human oversight. Ironclad does not publish when an agent stops or hands work back, or a limit on what an agent may complete without a person.

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.

Named customers report figures. NEXT Insurance says Contract Review Agent gave legal operations half its time back, and AES reports first pass redlines that took 30 minutes to two hours now reach a first draft in minutes. Signifyd puts an MNDA review or custom order form clause at minutes rather than an hour to a day, and Hormel's senior counsel reports turnaround cut from 12 weeks to 3. A Forrester study drawn from five customer organizations puts the three year return at 314 percent. Apart from that study, the figures come from the customers themselves, undated and with no method stated.

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.

Ironclad's AI Addendum lets it train its own models on customer data, input and output only after a customer enables AI Training Settings in the admin console. An opt out stops new training from that date. That data is first de identified, anonymized and aggregated, and output for other customers may not contain it. Whether or not a customer opts in, Ironclad may use input and output to evaluate the performance and accuracy of the service.

External model providers may not train on customer data and run under zero data retention where available. The Enterprise Services Agreement treats customer data as confidential information, and SOC 2 covers the privacy category. The terms say nothing about attorney client privilege or work product, and Ironclad documents access by role rather than separation between matters.

Source: Vendor Published
CC on UPL and Professional Responsibility PostureA boilerplate disclaimer sits in the terms while the marketing describes the product in advice terms, or the intended audience is left ambiguous.

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.

Ironclad markets its AI to legal teams and to sales, procurement and operations users, and describes the platform as built for everyone, lawyers or not. Its AI Addendum states that Ironclad is not the customer's legal advisor, that using the AI products creates no attorney client relationship, and that the customer should consult its own counsel. A separate AI Disclaimer says output is not legal advice and should be verified with a licensed professional before anyone acts on it. Neither document addresses competence or supervision duties, or which jurisdictions the product suits.

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.

Ironclad says every AI feature passes through its Evaluation Framework, which measures accuracy, consistency and reliability against curated benchmarks and real contract data. Subject matter experts validate model output during development and after release, with feedback loops meant to catch bias early, and training data is chosen to cover varied contract types and jurisdictions. Dimitry Dukhovny, the chief information security officer, leads security, compliance and AI governance.

Customers can review, override and keep governing agent behavior across teams and contract types. No evaluation results are published, and nothing is disclosed about where output quality differs by contract type, jurisdiction or party.

Source: Vendor Published
AA on AI Safety and Data StewardshipRetention, deletion, access control, subprocessors and incident practice are all published, current, and specific enough to hold the vendor to.

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.

Ironclad's Data Processing Addendum, version 3.10 from 5 March 2026, limits storage of customer personal data to what the services need. It destroys all copies, archives included, within 90 days of the agreement ending, returns data within 30 days on request and issues a deletion certificate within 30 days. A security incident is reported to the customer within 48 hours of Ironclad becoming aware of it. The subprocessor list names each provider, its activity and its location for the United States platform, the EU datacenter and Clickwrap.

A new subprocessor comes with 30 days' email notice and a right to object. External model providers run under zero data retention and may not train on customer data. Data is encrypted with TLS 1.2 or higher in transit and AES 256 at rest, with per tenant keys and bring or hold your own key options, and access follows administrator set roles.

Source: Vendor Published
AA on AI Liability and RecourseWhat the vendor stands behind when its output is wrong is published and specific: indemnity scope, caps, carve outs, and any insurance or warranty a buyer can actually invoke.

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.

Version 2.2 of the Enterprise Services Agreement, in force since 1 July 2026, is published in Ironclad's legal center. Ironclad defends and indemnifies customers against third party claims that the platform infringes intellectual property, and against claims from unauthorized disclosure of customer data caused by its breach of the data and security terms. It warrants that the services materially conform to the agreement and comply with applicable law; if a nonconformity is not fixed within 30 days, the customer may terminate for a pro rata refund.

Liability is capped at the fees paid in the prior twelve months, rising to three times that amount for data and confidentiality breaches and the data indemnity. Gross negligence, intentional misconduct and the intellectual property indemnity sit outside the cap, and consequential loss is excluded. The service level schedule targets 99.7 percent uptime, with service credits as the remedy. The customer bears sole responsibility for results from AI output, which Ironclad does not warrant as accurate, and the agreement carries no insurance commitment.

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.

Ironclad offers native integrations, iPaaS connections to more than 8,000 applications, and a documented public API, with Tray, Zapier and MuleSoft in its directory. The directory names Salesforce, which is AppExchange approved, Coupa, SAP, NetSuite, Zip and Ramp, Microsoft Word and Teams, Slack, DocuSign and Adobe Acrobat Sign, and Box, Google Drive, OneDrive, Dropbox and Egnyte. MCP connectors answer questions about contracts and workflow status in Claude, ChatGPT and Slackbot, scoped to the user's permissions and linked back to Ironclad.

Ironclad publishes no connector to iManage or NetDocuments, and its directory entries do not say what data moves or in which direction.

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.

Ironclad runs as a cloud service on Google Cloud Platform, in multiple zones in the United States, with a separate EU datacenter. For EU hosted customers the subprocessor list places Anthropic and OpenAI, along with Elastic Cloud, Datadog and the product database, in the EU, so AI processing stays in that region too. Clickwrap runs separately on Amazon Web Services in the United States. Customers can bring or hold their own encryption keys. Ironclad does not describe a tenancy model or what changes between plans.

Source: Vendor Published
BB on Security Certifications and Trust CenterCertification is real and stated, short of accessible evidence: a named standard without scope, date, or a way to obtain the report.

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.

Ironclad holds SOC 1 and SOC 2 Type II reports from routine third party audits covering security, availability, confidentiality and privacy, along with ISO 27001, 27701, 27017 and 27018. It runs a GDPR program with standard contractual clauses, complies with HIPAA and holds Cloud Security Alliance Trusted Cloud Provider status, with a public CAIQ on the alliance's site. Penetration tests run each year and vulnerability tests each quarter, and disaster recovery is tested twice a year.

Certification documents are requested through a security portal, and the website gives no auditor, audit period or report date.

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.

Contract Review Agent uses models from Google Gemini, OpenAI and Anthropic, chosen for quality, latency and performance, and Ironclad says it evaluates and swaps in models as they change. Its own models, which learn from aggregated, anonymized data of customers who opt in, sit alongside them. The subprocessor list names Anthropic, OpenAI and Extend for AI and Google Cloud Platform for cloud and AI infrastructure, each in the United States or the EU.

A new subprocessor brings 30 days' email notice and a right to object, and AI subprocessors run under zero data retention where available and may not train on customer data. Ironclad does not say which model serves which agent, and a change of model within an existing provider triggers no notice.

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.

Ironclad prices by custom quote and publishes no figure. Its pricing page sets out the structure. A customer picks one or more of the CLM, AI assistant and eSignature products, and a deployment led by its own team, Ironclad's legal engineers or a partner. Add ons cover the API and integrations, extra instances and success plans. Security costs nothing extra, some integrations are included and others are paid add ons, and the academy, community and in app guides are free.

Its fair use policy for AI sells agent use as AI Credits set in the order form and bills overage at the then current list price. Fair use of the embedded AI features is judged against seat count. Contract Review Agent comes within Ironclad CLM, with a seven day trial arranged through sales.

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.

Ironclad sells to enterprise and growing companies, with dedicated pages for general counsel, legal operations, procurement, sales and IT. It describes the platform as handling every contract type, from legal and sales to finance, HR, marketing and procurement, from intake through renewal. Named customers include Mastercard, L'Oreal, Cisco, Docker, Gusto and Rippling. The product covers contracting only, with no litigation or legal research function. Ironclad does not say which organization sizes or contract types it does not serve.

Source: Vendor Published
Sources on file

6 public documents

The public pages on file for Ironclad, 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.

What changed

Tracked changes at Ironclad

Oct 8, 2026Product / capabilityVerified

Ironclad Assistant became Ironclad Agent on 8 Oct, a single entry point that routes requests to Intake, Contract Review, Archive, Renewals and Analytics agents grounded in a Contract Knowledge Graph of clauses, obligations and negotiated decisions. The Search, Analytics, Workflow Designer and Contract Family agents and policy based access control, checked on every request, are generally available, while Playbook Agent is in early access. Where enabled, the Agent homepage becomes the default view, and switching the Agent off in Company Settings also disables AI Search and MCP access.

Bears on: AI CentralitySource
Sep 17, 2026Integration / interoperabilityPartially Verified

Ironclad released an Ironclad Contracts plugin for ChatGPT. Users can search their Ironclad contract repository and workflows in plain language from inside ChatGPT.

Bears on: Practice Systems Integration DepthSource

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?

Opt in

Training occurs only where the customer has affirmatively enabled it.

Section 1.1 of the AI Addendum, version 2.1 from 20 April 2026, lets Ironclad train its own models on customer data, input and output once the customer turns on AI Training Settings. The data is de identified, anonymized and aggregated first, an opt out stops new training but not training already under way, and external model providers may never train on it. Ironclad's product FAQ says the default setting depends on the customer's agreement and data location.

Source: Vendor Publishedthe rights set forth in this Section 1.1 if Customer has Opted InAs of Oct 14, 2026Evidence

Prompt and Output Retention

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

Disclosed without a period

Retention is acknowledged in public materials with no stated period.

During the term, the Data Processing Addendum keeps customer personal data only as long as the services need it, and the AI Addendum sets no separate period for prompts and output. All copies are destroyed within 90 days after the agreement ends, with a 28 day window to export the repository. External model providers run under zero data retention.

Source: Vendor Publishednot store or retain any Customer Personal Data except as necessaryAs of Oct 14, 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.

Ironclad keeps its own role based permissions, set by administrators, rather than inheriting them from a document system. Agents and MCP answers respect the same permissions, so a user sees only the contracts already open to them. Nothing describes walls between matters, since the product is built for in house teams rather than firms.

Source: Vendor PublishedGive users visibility into contract data based on pre-set, admin-controlled permissions.As of Oct 14, 2026Evidence

Third Party Request and Subpoena Notice

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

Notice committed

Terms commit to notice where lawfully permitted. No transparency report located.

The Data Processing Addendum commits Ironclad to tell the customer promptly of any government or law enforcement request for its personal data, unless the law forbids it, and to help the customer contest it. The Enterprise Services Agreement allows disclosure of confidential information under a court or government order after reasonable notice where the law permits. Ironclad publishes no transparency report.

Source: Vendor Publishedshall promptly notify Customer of any request by government agency or law enforcement authorityAs of Oct 14, 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.

Ironclad's AI works over the customer's own contracts, playbooks and workflow data rather than a body of law. Its own models learn from aggregated, anonymized patterns across customers who opt in, and Ironclad cites more than 2 billion contracts processed. No source, scale, license or refresh cycle is given for that training data.

Source: Operator VerifiedAs of Oct 14, 2026Evidence

Good Law Verification

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

Not addressed

No located public material addresses whether authority is checked for subsequent history.

Ironclad works on contracts and has no case law research function, so its output rarely cites authority. No citator, treatment signal or check of later history is described.

Source: Operator VerifiedAs of Oct 14, 2026Evidence

Refusal and Uncertainty Behavior

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

Not addressed

No located public material addresses what the product does when it cannot ground an answer.

Contract Review Agent flags missing clauses and departures from a company's standards, and extraction highlights the passage each value came from. When the model cannot place a value, the grounding system retries on the raw and normalized text rather than returning nothing. No path for declining to answer, and no confidence signal for the user, is described.

Source: Operator VerifiedAs of Oct 14, 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.

Ironclad has no entry in the AI Hallucination Cases database, Damien Charlotin's record of court decisions involving hallucinated AI content. Ironclad is a contracting platform, and its output is seldom filed with a court as cited authority.

Source: Operator VerifiedAs of Oct 14, 2026Evidence
Professional Responsibility

Bar Guidance Alignment

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

Not addressed

No located public material engages with bar or ethics guidance.

Ironclad's AI Addendum and AI Disclaimer say output is not legal advice and should be verified with a licensed professional. No ethics opinion or bar guidance is named, ABA Formal Opinion 512 included.

Source: Operator VerifiedAs of Oct 14, 2026Evidence

Billing and Fee Posture

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

Savings claims only

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.

Ironclad's savings claims come from named customers: NEXT Insurance recovering half of legal operations time, and AES and Signifyd cutting first redlines and NDA reviews to minutes. Its buyers are in house teams that do not bill clients by the hour, and nothing records AI assisted work for fee purposes.

Source: Vendor PublishedAs of Oct 14, 2026Evidence

Outside Counsel Guideline Readiness

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

Subprocessors listed

A current subprocessor or model provider list is published.

Ironclad's subprocessor list names Anthropic, OpenAI and Extend for AI and Google Cloud Platform for cloud and AI infrastructure, with locations for the United States platform and the EU datacenter. The Enterprise Services Agreement, Data Processing Addendum and AI Addendum are public in its legal center. A new subprocessor brings 30 days' email notice and a right to object. No notice pack written for a client's outside counsel guidelines is published.

Source: Vendor PublishedIronclad will notify Customer via email and allow Customer thirty (30) days to objectAs of Oct 14, 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.

Ironclad's workflows record routing, assignment and approvals for each contract, and AI review is described as auditable, with humans able to review and override it. Extracted values highlight their source passage and redlines can show their rationale. No export records which model produced an output or who verified it. Ironclad's output is contract work rather than court filings.

Source: Vendor Publishedtransparent, auditable AI behavior you can review, override, and continuously governAs of Oct 14, 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 430 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
October 14, 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