Agiloft
Agiloft is an enterprise contract lifecycle management platform, sold as the Agreement Platform, that runs contract creation, negotiation, approval, signature, repository and obligation management for legal and procurement teams across industries including financial services, healthcare, pharma and biotech, manufacturing, technology, energy and business services. Its AI layer, sold as AI Your Way and Astra, includes AI Trainer, a no-code tool for training contract review models on a customer's own documents, ConvoAI Document Q&A, which answers questions about a contract and links each answer to the highlighted source text, a GenAI Prompt Lab for building saved multi-step prompts, generative redlining that proposes insertions and deletions against approved clause language, and Screens; an Integration Hub built on Workato connects contract data to other business systems. The company is Agiloft, Inc. of Redwood City, California, and the platform is offered on either a single-tenant or multi-tenant basis. Its published Services Agreement and Supplemental Terms state that customer data is not used to train AI models or to improve the services without the customer's explicit prior written consent, name Microsoft Azure OpenAI Service as an AI sub-processor and disclose its temporary abuse-monitoring storage, incorporate the Microsoft, OpenAI and AWS acceptable-use policies, commit to prompt notice of compelled disclosure, delete customer data within thirty days of termination, and set a twelve-month liability cap with a three-times cap for security and confidentiality breaches alongside indemnities for intellectual property infringement and for unauthorised disclosure of customer data; independent security audit reports are available through a trust portal and a sub-processor list is published with emailed notice of changes.
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 models are the engine of a core capability layered on a product that would still function without them. AI Trainer, ConvoAI Document Q&A, the GenAI Prompt Lab, generative redlining and Screens are sold as AI Applications on top of a contract lifecycle management platform whose workflows, repository, approvals and integrations predate them; the Services Agreement defines AI Applications as functionality made available as part of the subscription and specified in the order, and the Supplemental Terms govern them as an add-on class. Remove the models and the CLM stands. Home page, news releases, Services Agreement and Supplemental Terms read 6 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.
Grounding is real and documented with links to source, short of a testable accuracy figure. ConvoAI Document Q&A ties each answer to the passage in the contract that supports it and lets the user jump to the highlighted text, which the vendor calls white-box AI; Services Agreement section 9.3(c) states that AI features exhibit varying degrees of accuracy and the Supplemental Terms require the customer to check output independently. No test set, benchmark or evaluation is published. The primary-authority limbs do not apply to a tool that reads the customer's own contracts. February 2025 release, Services Agreement and Supplemental Terms read 6 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.
A written commitment that the models work alongside a reviewing person, with real review surfaces, short of the full control structure. Generative redlining proposes insertions and deletions against approved clause language for a negotiator to accept, the vendor's own product lead describing it as best used on substantive differences and not on minor edits; AI Trainer lets a firm decide what the review model looks for; ConvoAI answers link to source for verification; Services Agreement section 2.3(c) makes the customer solely responsible for evaluating output. What is not published is a threshold at which any AI Application acts without a person or a stated route back after a wrong output beyond the customer's own review. January 2024 launch coverage, Services Agreement and Supplemental Terms read 6 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.
Scale claims stand in for named deployments on the surfaces read. The vendor states a 99.6 per cent implementation success rate without a method and describes customers as some of the largest companies; a customer stories page and Pacesetter awards exist in the navigation and were not opened, and are the rebuttal route. News releases and navigation read 6 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.
Four of the five limbs are in the published agreement and the privilege limb is absent. No training: Services Agreement section 2.3 and Supplemental Terms section 1 provide that customer data is not used to train AI Applications or AI Models without the customer's explicit prior written consent, and section 4.3 extends that to developing or improving the services at all. Segregation at the level an in-house buyer requires: the platform is offered single-tenant or multi-tenant by order, and section 2.2 confines Agiloft's use of customer data to providing and maintaining the services. Retention and deletion: section 14.3 deletes customer data within thirty days of termination and section 4.2 allows export at any time. Third-party model providers: Microsoft Azure OpenAI is named as sub-processor with its temporary abuse-monitoring storage of input and output disclosed, and AI service providers are defined as those on the published sub-processor list. Nothing addresses privilege or work product. Services Agreement and Supplemental Terms read 6 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.
A real position on advice versus tooling, short of a supervision dimension framed for lawyers and of jurisdiction limits. Services Agreement section 9.3(b) states that Agiloft is not an attorney or law firm or a substitute for one and does not provide advice, explanation, opinion or recommendation about legal rights, remedies, defences, options, selection of forms or strategies; section 9.3(c) and the Supplemental Terms state that AI output may be inaccurate and must be checked independently. The buyer is legal and procurement teams, and nothing describes how the product supports a supervising lawyer's duties or names a jurisdiction limit beyond export control. Services Agreement and Supplemental Terms read 6 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.
Transparency claims and incorporated third-party policies without a governance framework, testing regime or accountable owner. The vendor markets white-box AI that shows its reasoning by linking answers to source, the Supplemental Terms bind customers to the Microsoft AI Code of Conduct, OpenAI usage policies and AWS responsible AI policy, and section 3 warrants that usage-data insights are anonymised; no responsible AI framework of Agiloft's own, ISO 42001 or equivalent, pre-release testing description or statement about uneven output is published on the surfaces read. Supplemental Terms and news releases read 6 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.
Substantive published policy covering most of the ground. Retention and deletion: Services Agreement section 14.3 deletes customer data within thirty days of termination unless law requires longer, and section 4.2 gives export at any time. Sub-processors: a sub-processor list is published at a stated URL, AI service providers are defined by reference to it, and section 16.2 commits to emailed notice of updates to it; Microsoft's temporary storage of AI content for abuse monitoring is disclosed. Access control: section 5.1 commits to organisational, physical and technical precautions against access by employees and subcontractors, with detail in a security page and trust portal not opened. Incident practice: section 5.3 gives the customer a reporting route; a notification commitment sits in the DPA, which was not opened. Services Agreement and Supplemental Terms read 6 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.
What the vendor stands behind is published and specific, and it goes further than most: Services Agreement section 10.1 gives an indemnity for intellectual property infringement and a second indemnity for unauthorised disclosure or misuse of customer data resulting from a breach of the or security obligations, with exclusions in 10.2 and mitigation remedies including a pro-rata refund; section 11 caps each party at twelve months' fees, raises the cap to three times for breaches of the security, confidentiality and compliance sections, and excludes gross negligence, wilful misconduct, fraud and the indemnities from any cap; section 9.2 warrants material conformity with a repair-or-refund remedy; and the Supplemental Terms state that AI output is as-is and that Agiloft has no liability for any damages arising from use of output. No insurance is stated. Services Agreement version 1.2 of 22 July 2026 and Supplemental Terms read in full 6 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.
Real integrations, documented, with depth described in general terms. The Integration Hub is an embedded integration platform powered by Workato that the vendor states connects Agiloft to hundreds of other systems through connectors and recipes, with public help documentation at the Agiloft wiki; Third-Party Services are addressed in the Services Agreement. What a firm must configure and which systems sync in which direction are not described on the surfaces read; the integrations page and wiki were not opened. Supplemental Terms, Services Agreement and navigation read 6 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.
The tenancy model is stated in the agreement and residency is partial. The Services Agreement defines the core platform as available on a single-tenant basis, a dedicated cloud environment for one customer, or a multi-tenant basis, and the order selects between them; an EU Data Act addendum exists for EU customers. No region list, hosting provider or processing location for the AI providers is stated on the surfaces read; the security page and trust portal are the rebuttal route. Services Agreement read 6 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.
Independent audits are stated in the agreement with a named route to the reports, short of a report reachable without asking. Services Agreement section 5.2 states that Agiloft has completed audits by an independent auditor of the design and effectiveness of its security controls and that customers may obtain the reports through the trust portal at trustportal.agiloft.com, where they are confidential information. The audit standard, auditor and coverage period are not named in the text read, the portal was not opened, and whether a prospect can obtain a report self-serve is not established, so the lower access tier is graded. Services Agreement read 6 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.
Providers and change notice are published and the models are not named. The Supplemental Terms name Microsoft Azure OpenAI Service as the provider behind certain AI Applications and disclose that Microsoft temporarily stores input and output for abuse monitoring with human review of flagged content; the incorporated acceptable-use policies are Microsoft's, OpenAI's and AWS's, which identifies the provider set; AI service providers are defined as those on the published sub-processor list and section 16.2 commits to emailed notice of changes to that list. No model is named and no inference location is stated. Supplemental Terms and Services Agreement read 6 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.
The unit and structure are stated in the agreement without a figure on the surfaces read. The Services Agreement prices by order with use limits by type and number of users, prorated mid-term increases, add-on services priced separately, thirty-day payment terms and a choice of single-tenant or multi-tenant platform; a pricing page exists in the navigation and was not opened on 6 September 2026, and is the rebuttal route in either direction. Services Agreement read 6 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.
Segment and coverage are described with substance and one limit is stated. Solution pages address legal and procurement functions and seven industries, the buyer is enterprise, and the platform is offered single-tenant or multi-tenant with an EU addendum; the Services Agreement bars use for high-risk activities, which is a stated boundary. No contract type or practice area is named as unsupported and no law firm use is described. Navigation and Services Agreement read 6 September 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?
Training occurs only where the customer has affirmatively enabled it.
Training happens only if the customer affirmatively enables it, and the switch is in the agreement. Services Agreement section 2.3 and Supplemental Terms section 1 provide that customer data is not used to train AI Applications or AI Models without the customer's explicit prior written consent, and section 4.3 provides that Agiloft does not use customer data to develop or improve the services without the same consent; usage data, defined to exclude the contents of customer data, may be used to improve the AI Applications. The default is no training and the mechanism is written consent. Surfaces checked 6 September 2026.
Prompt and Output Retention
How long does the product keep what a lawyer typed, and can that be set to zero?
A specific retention period is published and the customer cannot change it.
A specific period is published and the customer cannot change it. Services Agreement section 14.3 deletes all customer data, which includes input and output, within thirty days of termination unless law requires longer, and section 4.2 allows export at any time during the term; the Supplemental Terms disclose that Microsoft temporarily stores AI input and output for debugging and abuse monitoring. Nothing states a shorter or configurable window for prompts and outputs during the term. Surfaces checked 6 September 2026.
Ethical Walls and Matter Segregation
Does retrieval respect the firm’s ethical walls, or can the model read across them?
The product maintains its own permission model, documented, requiring the firm to keep it aligned.
The product's separation model is documented in the agreement at tenant level. The Services Agreement defines single-tenant as a dedicated cloud environment for one customer and multi-tenant as an environment that may be shared, with the choice made in the order; section 2.2 confines Agiloft's use of customer data to providing the services. Nothing describes matter-level or department-level walls within a customer's instance or how the AI Applications respect user permissions; the security page was not opened. Surfaces checked 6 September 2026.
Third Party Request and Subpoena Notice
If someone subpoenas the vendor for a firm’s data, does the firm hear about it first?
Terms commit to notice where lawfully permitted. No transparency report located.
The published agreement commits to notice. Services Agreement section 7.3 permits disclosure of confidential information, which includes customer data, to the extent required by law and requires the receiving party, if legally permitted, to give the disclosing party prompt notice of the compelled disclosure and reasonable assistance should it wish to contest it; section 5.1 references the same clause for compelled access to customer data. No transparency report is published. Surfaces checked 6 September 2026.
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.
No located public material identifies a legal corpus behind the product's answers, and the product is not built on one: ConvoAI answers questions about the customer's own contracts with links to the source passage, AI Trainer learns from the customer's own documents, and generative redlining works from the customer's approved clause language. No primary law source, licence or update cadence is published. News releases and terms checked 6 September 2026.
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.
No located public material addresses whether authority is checked for subsequent history, and the product does not retrieve or cite primary law; its output is contract answers, redlines and extracted data. Recorded as the honest value for a product without a citator function. Surfaces checked 6 September 2026.
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 product documents how a user can check an answer rather than what the model does when it cannot answer. ConvoAI Document Q&A links each answer to the supporting passage so the user can verify it in the source, which the vendor markets as white-box AI; nothing states an abstention path or confidence signal when no supporting passage exists, and the terms require independent checking of every output. Recorded as documented on the strength of the source-linking design, with the abstention gap noted. February 2025 release and terms checked 6 September 2026.
Fabricated Citation Record
Does a public court record exist involving output from this product?
No court order, opinion or disciplinary record naming this product has been located as of the date shown. This is a statement about the public record, not a finding about the product.
No court order, opinion or disciplinary record naming Agiloft, ConvoAI or Astra was located as of 6 September 2026. The AI Hallucination Cases database maintained by Damien Charlotin was searched on the company and product names together with a general search for court findings; results returned sanctions involving general-purpose chatbots and, in one Fifth Circuit matter, other named legal research tools, none of which is this product. This is a statement about the public record, not a finding about the product; a contract platform that cites no authority carries a remote exposure on this signal.
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 located public material names an ethics opinion, bar rule or professional responsibility framework. The Services Agreement addresses the advice line at section 9.3(b) and the Supplemental Terms incorporate the AI providers' acceptable-use policies, but no bar or regulator guidance on lawyers' use of AI is named on any surface read. Terms and news releases checked 6 September 2026.
Billing and Fee Posture
Does the vendor address what happens to the bill when the work takes an hour instead of six?
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 buyer is an in-house legal or procurement function that bills no client, so the product sits outside a lawyer-to-client fee relationship. The published savings framing is about cycle time and negotiation loops rather than fees, and nothing addresses how AI-assisted work is recorded or disclosed on any bill; no law firm is a named buyer segment. Surfaces checked 6 September 2026.
Outside Counsel Guideline Readiness
Can a firm get this vendor through a client’s AI clause without a bespoke negotiation?
A current subprocessor or model provider list is published.
A current sub-processor list is published with the AI providers inside it, and client-facing material exists in the published terms. The Services Agreement defines AI service providers as the third-party AI providers listed on the sub-processor list at a stated URL and commits to emailed notice of updates; the Supplemental Terms name Microsoft Azure OpenAI Service and disclose its content handling, and incorporate the Microsoft, OpenAI and AWS policies; the DPA is published. The sub-processor list page itself was not opened on 6 September 2026, so whether it identifies every model provider is not established, which is why the row sits at this value rather than the top one; the list is the rebuttal route.
Court Disclosure Support
If a judge’s standing order requires an AI disclosure, can the product produce one?
No located public material addresses court disclosure or verification certification.
No located public material addresses court disclosure or verification certification of AI-assisted work. ConvoAI's source-linked answers give a user a per-answer verification path inside the product, but nothing states that a record of the model used, sources and human verification can be exported for a court, and the product produces no court-facing work product. News releases and terms checked 6 September 2026.