Ontra vs PERSUIT: how they compare in 2026

The models are the engine of the data products and a layer on the service products. Insight abstracts clauses and covenants from fund and credit agreements into structured obligation data, Accord produces AI markup against a digital playbook, and DDQ generates responses from a precedent library — all model-driven. But Contract Automation, the flagship, states on the pricing page that the contract resource is a human: qualified contract professionals on Standard, qualified lawyers on Advanced, lawyers with top-tier firm experience on Premier. Remove the models and an outsourced negotiation service and an entity directory remain saleable. That places this at B rather than A: a core capability is genuinely model-driven, sitting on a workflow and document platform that would still function without it. Pages read 1 September 2026.

Persi is described by the vendor as the intelligence layer built into PERSUIT, and the pricing position is explicit: it is not bought separately and arrives with the platform. PERSUIT ran as a competitive RFP, AFA and e-billing platform for about a decade before Persi existed, and the underlying capabilities — request management, panel governance, invoice review, spend reporting — are workflow and data functions that stand without models. What the models add is real and central to how the product is now sold: matter triage, fee benchmarking, firm shortlisting, proposal evaluation and anomaly detection across six named workflow stages. That is the B band exactly, a model-driven core capability layered on a product that would still function without it. Pages read 1 September 2026.

Accuracy is asserted and never measured. The AI page promises “reliable AI” and “expert-validated, actionable intelligence”, and the platform markets AI Search that answers questions “informed by the full scope of your agreements”, but no accuracy figure, no test set, no evaluation and no named failure mode was located on any page read on 1 September 2026. Nothing published establishes that an answer links back to an openable source passage a reader can verify, which is what the B band requires alongside a described retrieval method. The subject-matter-expert review process that catches hallucinations is real and is graded on Autonomy and Oversight, where it answers the question the band actually asks; it is not spent again here. Two limbs of the higher bands do not apply, since the corpus is the customer’s own fund documents rather than primary law, so grounding to primary authority and citation-status checking are outside what this product claims.

The corpus is described with unusual precision and the output is never measured. PERSUIT states that Persi draws on more than $22 billion in proposal data across 4,800-plus firms, and the proof page dates that dataset to platform transactions as of February 2026, which is a stronger provenance statement than most vendors give. But no accuracy figure, no evaluation, no test set and no named failure mode was located for Persi’s own outputs on any page read on 1 September 2026. Persi’s illustrated answers cite counts of comparable matters and percentages against benchmark, and nothing published establishes that a user can open the underlying comparables to check them. The primary-authority and citator limbs of the higher bands do not apply, since the product reasons over commercial pricing data rather than case law. C: accuracy asserted without measurement.

A written human-review commitment with a described mechanism. The security page publishes a Human-in-the-Loop Model under which subject-matter experts continuously review text outputs for accuracy, flag potential hallucinations and enforce consistency across repeated questions, then troubleshoot root causes and attribute them to retrieval-augmented generation prompting, query understanding or embedding models. That last detail is unusual: it names the layers of its own pipeline where failures are diagnosed. The pricing page adds a second published control by tiering who performs the work, from contract professionals to lawyers with top-tier firm experience. What is missing for A is the rest of the control structure — no thresholds, no statement of what runs unattended versus what a customer must approve, and no described in-product review surface or escalation route. B on the strength of the review commitment, held there by the absent thresholds.

A clear written division of labour with a visible approval point. The Persi page states that Persi does not replace the customer’s judgement and that the customer brings the judgement while Persi brings the infrastructure, and every one of the six illustrated workflows ends at a confirmation rather than an action — requests are drafted ready to review and send, a breaching invoice is put on hold pending review, exceptions are flagged for the customer. Constraint is customer-configured: engagement rules such as minimum competitive bids, spend thresholds, panel requirements and diversity obligations are applied to every matter, with a worked example of a $100,000 threshold triggering a competitive request to at least three panel firms. Held at B rather than A because no autonomy modes are published, nothing states what Persi may do without asking, and the control structure is shown through product illustrations rather than documentation a buyer could hold the vendor to.

Attribution is strong and the figures sit on the wrong side of it. Named customers carry named officers with titles: Vincent Taurassi, General Counsel and COO at Sentinel Capital Partners; Lizette Perez-Deisboeck, General Counsel and CCO at Battery Ventures; Jonathan Romick, General Counsel at Linden Capital Partners, describing an SEC production; Ed Zelaski, Director of Compliance at Blue Point. Blackstone and Bain Capital Credit carry their own case studies. Separately the vendor publishes portfolio figures — 1,000-plus customers, 2 million-plus documents processed, 9 of the top 10 PEI-ranked firms, and 96% customer retention with a genuine methodology footnote stating it is the average across customers using a platform solution for at least a year as of December 2025. That footnote is the kind of dated, assessable method the A band asks for, but it describes the vendor’s book rather than what changed at a named firm. B: named customers without figures, and figures without the named customer.

All four elements the band asks for are present. Named legal departments with named officers: Gabriel Harnier, General Counsel at SAP; Anne Robinson, Chief Legal Officer at IBM; Sabine Chalmers, General Counsel at BT Group; Leanne Geale, Group General Counsel at Nestlé; Ernst van de Weert, General Counsel at Heineken. A figure for what changed at a named customer: SAP reached its three-year savings target in nine months. Dated platform figures with the basis stated — 26% average savings per matter, 30% with competitive sourcing, 48% of the time the lowest bid wins, 79% of value awarded under an AFA — each with its denominator explained, footnoted as sourced from PERSUIT platform transactions as of February 2026, and carrying an explicit variance caveat that individual results differ by matter type, industry and process design. Publishing the limitation alongside the number is what makes the method assessable rather than decorative. The softness is that the per-customer testimonials are quotes without their own figures.

Substantive commitments on the two questions that decide most of this axis, and silence on the third. The security page states plainly that Ontra never uses customer data to train models, and that data passed to third-party LLM providers is processed only to deliver the service and deleted afterwards — a position on what the model provider retains, which most vendors in this pull do not give. Terms of Service clause 10.2 commits Ontra to a reasonable degree of care over User Data. Against that, clause 6.2 reserves the right to collect usage data on an aggregated and anonymised basis to improve products, except as limited by a Customer Agreement, so the no-training promise lives on a marketing page while the published agreement reserves an improvement use. Segregation is the gap: nothing documents tenant separation, and clause 10.2 warns that other Users with access rights may reach User Data. Privilege and work product are not addressed. B, held off A by segregation and by the Customer Agreement being unpublished.

Confidentiality is asserted in general terms and the substance sits behind an NDA. The privacy policy commits to reasonable steps against misuse and unauthorised access, states that employees are prohibited from viewing customer data unless necessary to diagnose a problem, and describes SSL/TLS in transit and sign-in logging; the security page adds SSO and role-based access controls. Nothing addresses training on customer content, which is a live question here because Persi is marketed as trained on the aggregate proposal dataset that customers themselves generate, and no published term says whether a given customer’s proposals feed it. No retention or deletion position, no matter or tenant segregation, and no privilege or work product treatment was located. The documents that would answer these — subprocessor list, data flow diagrams, SOC 2 report — are enumerated on the security page as available in the Trust Center under NDA, so a buyer cannot read the commitments before entering an evaluation. C on the band’s second limb.

A real, product-scoped position rather than a website disclaimer. Terms of Service clause 12.2 is headed NO LEGAL OR OTHER ADVICE and states that nothing in the Services, the software or the documentation constitutes legal advice, that Ontra is not a law firm and does not provide legal services, advice or representation, and that no attorney-client relationship is formed. It also places responsibility for decisions taken in reliance on output with the user. That clause reaches the Platform, not just the marketing site, which is what separates this from most of the lane. It is worth reading against the business: Ontra operates the Ontra Legal Network and the pricing page sells access to qualified lawyers who negotiate the customer’s contracts, so the no-legal-services clause and the commercial offer sit in visible tension. B rather than A because nothing addresses the buyer’s own competence and supervision duties and no jurisdiction limits are named; ABA Formal Opinion 512 is not mentioned.

Nothing published on the advice line. No terms of service, customer agreement or disclaimer was located on 1 September 2026: the footer link labelled Privacy & Terms resolves to a privacy policy alone, and the home page, platform, Persi, security, integrations and proof pages carry no disclaimer of any kind. This matters less here than for a drafting product, because PERSUIT produces sourcing and pricing recommendations rather than legal work product, and the note records that as the reason the axis bites softly. It is not nothing, though: Persi is described as advising whether a matter should be insourced or sent out, recommending which firms to instruct, and evaluating which proposal represents best value, and those recommendations are consumed by in-house lawyers and by law firms. No position on the line between tooling and professional judgement is published anywhere, and ABA Formal Opinion 512 is not mentioned.

Ontra holds ISO 42001, the international management-system standard for AI, stated on the security page and carried as a badge on the home page. A certified AI management system is a published governance framework with real substance rather than a set of principles, and it is the second instance of this certification in the pull after Corlytics. What is absent is everything the A band adds: no individual or committee inside Ontra is named as accountable, no pre-release testing regime is described, and no result is disclosed about uneven output across document types, jurisdictions or counterparties. Security certifications are excluded from this axis by the band and the SOC 2 and ISO 27001 attestations are graded on Security Certifications instead, so ISO 42001 is the only certification spent here. B. Checked 1 September 2026.

No governance position published for a system that allocates work between law firms. The home page, Persi page, platform pages, security page, integrations page, proof page and privacy policy were read on 1 September 2026; there is no responsible AI page, no named accountable owner, no pre-release testing regime, and no ISO 42001 or equivalent. The gap is sharper here than the grade alone conveys. Persi shortlists firms, scores firm performance with numeric value scores, and decides which firms are invited to compete for mandates worth six and seven figures, and the vendor itself lists diversity obligations among the engagement rules the system enforces. Nothing published addresses whether those recommendations have been tested for uneven treatment across firms, jurisdictions or matter types. Security certifications are a different subject under this band and are graded on Security Certifications instead.

Most of the ground is covered and the gap is incident practice. The security page publishes AES-256 with rotating keys at rest and TLS 1.2 or above in transit, states that third-party LLM providers encrypt in both states, and commits to deleting data passed to those providers once processing is complete. It names a dedicated security team and the tooling classes it runs, listing SAST, DAST, CDR, EDR, continuous GRC, WAF and MDM. A subprocessor list is published on the trust centre at trust.ontra.ai, which also offers email notification of subprocessor changes — a live disclosure mechanism rather than a static page. Not located: any incident response or breach notification practice, and any statement of how long documents and generated outputs are retained inside the platform itself, as distinct from the zero-retention commitment at the model layer. B on the band’s own named exception for an unstated incident practice.

A generic privacy policy plus a security summary, with the operative detail withheld. What is published: SSL/TLS encryption in transit, employee access restricted to necessary diagnosis, sign-in logging, regular external security audits by contracted firms, business continuity and disaster recovery with tested contingency planning, role-based access control and full audit logging across platform activity. What is absent: any retention period, any deletion practice beyond GDPR erasure rights, and any incident or breach notification practice. The subprocessor list exists but the security page places it in the Trust Center under NDA, so under the three-tier rule it earns no credit. The privacy policy is also stale in a way a buyer should notice — it is dated October 2021 and still relies on the EU-US Privacy Shield Framework, invalidated by Schrems II in July 2020. C: the policy covers the product without addressing what happens to matter data, proposals and prompts after processing.

The published allocation is specific and it runs one way. Terms of Service clause 13.2 caps Ontra’s aggregate liability at $100, disclaims all consequential, incidental and special damages including loss or corruption of data, and carves out only fraud and willful misconduct. Clause 12.1 provides the Services as-is with all warranties disclaimed, and clause 13.1 runs the only indemnity from the user to Ontra. There is no warranty on output, no insurance position, and no indemnity flowing to the customer. The instrument that would actually govern a buyer is not published: clause 1 states that where a user has access under a customer’s written Customer Agreement, that agreement supersedes these Terms, and no Customer Agreement, MSA or DPA was located on the site on 1 September 2026. C rather than D because a buyer can read a specific figure and carve-out before signing; C rather than B because what is published disclaims the exposure the product creates rather than standing behind it.

Nothing published on who bears the loss when a recommendation is wrong. No terms of service, master subscription agreement, customer agreement or service level agreement was located on 1 September 2026. The escalation ladder was run: the footer was checked across six rendering pages and its Privacy & Terms link resolves to a privacy policy carrying no liability, warranty or indemnity provision, and no separate terms page surfaced from the navigation. The Trust Center is the only route to contractual documentation and the security page states its contents are available under NDA, which is a gating decision by the vendor rather than a retrieval failure on this side, so it earns no credit. No indemnity, no cap, no carve-out, no warranty on output and no insurance position is readable before a buyer enters an evaluation.

Real named integrations, documented at the level of what they do rather than how to configure them. The Ontra MCP Server gives a customer’s own LLM real-time, read-only access to Atlas entity data, letting a firm query entities and ownership structures from its existing AI tool — direction and permission model both stated. Ontra for Word is named in the Terms of Service as an application plug-in, putting the product in the drafting surface lawyers already use. The Atlas pricing panel includes a Standard API package, the DDQ panel lists flexible exports, and Insight publishes side-letter compliance exports. Contract Automation includes a document management system in all three editions. B rather than A because no integrations page, developer index or configuration documentation was located on 1 September 2026, and no connector to a legal document management system such as iManage or NetDocuments is named.

The most detailed integration disclosure located in this lane. Four connection routes are published — managed integrations, an RFP API, a reporting API and composable workflows — and each is documented at field level. Direction is stated explicitly with arrows: matter management system into PERSUIT for matter ID, reference data, legal team assignments, practice group and jurisdiction; PERSUIT out to the e-billing system on RFP conclusion, carrying PERSUIT ID, firm name, total budget, currency, matter owner, entity, category and department. Sample API fields are printed on the page, including request_id, participation_status, pricing_items, total_fee, average_rate, price_at_start and price_at_end. Configuration burden is addressed directly, with managed integrations stated to need zero custom development, and a public API documentation URL is published. The one weakness, named rather than hidden: the target systems are described by category rather than by product, so no specific matter management or e-billing partner is named, and the API documentation sits behind the application and was not opened.

Cloud delivery is obvious and neither dimension is stated. The product is a web platform signed into at app.ontra.ai, but no page read on 1 September 2026 — home, AI capabilities, security and privacy, pricing, or the Terms of Service — states a tenancy model or offers a data residency region, and no processing location is given as distinct from storage. The one thing published that touches location is that third-party LLM providers encrypt data in transit and at rest, which says nothing about where either sits. The trust centre at trust.ontra.ai is a Vanta portal whose body did not render to automated retrieval and may well carry hosting and region detail; that is a limit on this reading rather than a gap in the vendor’s disclosure, so nothing is graded against Ontra for it. C on the band’s own words, which are true here: cloud delivery is implied and neither the tenancy model nor the region is stated on any readable surface.

The host is named and neither dimension a buyer needs is stated. The security page states the platform runs on Google Cloud with MongoDB and Qlik as data providers, which answers whose infrastructure but not where a given customer’s data sits. No tenancy model is published anywhere. No residency option is offered: the only location material is in the privacy policy, which says personal information may be stored outside Australia with cloud providers in the United States and that European personal data may be transferred to Australia and the US under standard data protection clauses. That is a transfer disclosure about personal information rather than a residency offering for platform data, and it comes from a policy dated October 2021. Processing location is not addressed as distinct from storage. C: cloud delivery is evident and neither the tenancy model nor a customer-selectable region is stated.

The strongest certification stack in the lane, one tier short of full credit. The security page states SOC 2 Type 2 across all five trust services criteria — scope actually specified, which is rare — plus ISO 27001:2022 with the version named and ISO 42001. A trust centre exists at trust.ontra.ai on Vanta, is publicly linked from the security page, and states that it carries subprocessor information and offers subscription to change notifications. Two things hold this at B. No auditor is named and no coverage period or report date appears on any readable surface, and the trust centre body did not render to automated retrieval, so whether reports are downloadable, gated behind a click-through, or gated behind a sales conversation could not be established. Under the standing rule, where the access flow cannot be determined the lower tier is graded and the reason stated: that is why this is B and not A. No request was submitted. Read 1 September 2026.

Real, current and unusually well itemised, one step short of accessible evidence. The security page states ISO/IEC 27001:2022 with the version named and describes it as independently audited rather than self-assessed, SOC 1 Type 2 attested, and CSA STAR certified. Third-party external assessments are published with dates and scores: SecurityScorecard A and UpGuard 865/950, both stated as independently assessed in January 2025. A SafeBase trust centre is named and publicly linked, and the page enumerates exactly what it holds — penetration test report, network diagram, data flow diagrams, SOC reports, subprocessor list, ISO certificate and Statement of Applicability, and OSPAR for financial services buyers. What holds this at B is the access flow and the missing attestation particulars: the SOC 2 report is stated to be available under NDA once access is granted, no auditor is named for any attestation, no coverage period is given, and the SOC 2 scope is not specified. No request was submitted.

The architecture is described in real detail and the providers behind it are never named. Ontra sets out its own pipeline layers by name, describing retrieval-augmented generation, query understanding and embedding, and states a zero-retention commitment at the model layer, so a buyer can see the shape of the system and knows customer content is not held by the model provider. What no readable surface answers is the question a buyer actually asks: whose model sees my content. No model, model family or provider is identified, and no inference location is stated. One further fact was established on 1 September 2026 and it cuts in the vendor's favour: the trust centre at trust.ontra.ai maintains a dedicated subprocessors page carrying a subscription for email notification of changes to the subprocessor list, so a change-notification mechanism exists even though the list itself could not be read. That page is a Vanta-hosted portal whose body renders client-side and returned only the page shell across repeated retrieval attempts, which is a limit on the reading rather than a gap in the disclosure, and nothing is graded against the vendor for it. B on the band's second limb, architecture described without the providers; an operator fetch of that page would settle whether the model providers are named there and, if they are, would move this row.

Persi is described entirely through its training data and never through its models. The vendor states repeatedly that Persi is trained on more than $22 billion of outside counsel management data and contrasts it with generic AI, naming Claude and Copilot only as competitor categories rather than as anything sitting underneath its own product. No model, no model provider, no inference location and no change-notification commitment was located on any page read on 1 September 2026. Google Cloud, MongoDB and Qlik are named on the security page, but hosting and data infrastructure answer where software runs rather than whose model sees client content and cannot be spent on this axis. The subprocessor list that might resolve it is in the Trust Center under NDA. C: the vendor refers to proprietary intelligence without identifying what sits underneath.

A real pricing page carrying tiers, feature splits, the unit of charge and one actual number. Contract Automation publishes three editions with a comparison matrix covering playbook customisation, who performs the work, turnaround and document management, and states a monthly minimum fee of $2,000 on both Advanced and Premier, with no monthly minimum on Standard. Insight for Funds publishes three tiers and Insight for Credit two, each with a populated feature matrix. Atlas, Accord, DDQ and KYC publish their licence inclusions and route the figure to Request Pricing. Free trials and a proof-of-value are offered without a sales gate on several products. B rather than A because the per-document rate that actually drives the bill is never published, so a buyer knows the floor but not the price, and six of the seven products carry no figure at all. B rather than C because a published figure exists at all, which C forbids.

No pricing information at any level. There is no pricing page: the navigation and footer were checked on 1 September 2026 across the home page, platform, Persi, security, integrations, proof and privacy pages, and none carries a pricing entry. No tier names, no packaging structure, no unit of charge and no figure are published. Every commercial route is a demo request, an industry briefing or a contact form, and the ROI calculator inverts the transaction by asking the buyer to supply its own outside counsel spend rather than disclosing what PERSUIT costs. D rather than C because C requires the shape to be visible with only the number withheld, and here neither the shape nor the number is published. Under the pricing-row rule there is no published structure, so no VendorPricing row was written.

Coverage is described by buyer and by workflow with real precision. The site publishes pages by team — legal, compliance, finance and investor relations — and by firm type, covering investment firms, investment banks and law firms, the last aimed at automating MFN processes. The practice scope is concrete and named: NDAs and other high-volume routine contracts, engagement letters, side letters and MFN elections, credit agreements and covenants, entity management and CTA compliance, reverse KYC, and investor due diligence questionnaires. The private markets framing is itself a genuine boundary and the vendor repeats it everywhere. B rather than A because the limits inside that boundary are not stated: nothing says which document types Contract Automation will decline, government use is not addressed, and the only scope caveat located is a pricing footnote noting that some document types such as commercial agreements or engagement letters run to a 48-hour rather than 24-hour turnaround.

Coverage is documented by role and by sector with real specificity. Five buyer pages are published — General Counsel and CLOs, legal operations, head of litigation, in-house attorney, and law firm partners and business development — which is notable because law firms are a distinct buyer here rather than an afterthought, sitting on the other side of the same transaction. Nine industry pages are published, from financial services and life sciences to energy and real estate. Matter coverage is named: multi-jurisdictional disputes, regulatory investigations, M&A and complex litigation. B rather than A because the limits are not stated and the vendor explicitly declines to state them, answering the objection that some matters are too complex with the claim that every matter type has been run through the platform and that complexity is the use case rather than an exception. Government legal departments are not addressed.

The security page states in its opening line that Ontra protects customer data by never using it to train models, and repeats the point as a headline commitment. No matching term was located in the published Terms of Service, and the Customer Agreement that clause 1 says supersedes those Terms is not published. Terms clause 6.2 runs the other way, reserving the right to collect usage data including on an aggregated and anonymised basis for improving products and services, except as limited by a Customer Agreement. Recorded as policy rather than contractual on that basis.

No located term or policy addresses whether customer content is used to train models. The privacy policy, security page, Persi page, platform pages and proof page were checked on 1 September 2026, and no terms of service or customer agreement is published. The question is squarely raised by the vendor’s own marketing: Persi is described as trained on more than $22 billion of outside counsel management data drawn from platform transactions, which is customer-generated proposal and pricing data, and nothing published states whether or on what basis an individual customer’s proposals contribute to it. The privacy policy is dated October 2021 and predates Persi.

Ontra publishes a Zero Data Retention commitment covering data provided to third-party LLMs through an Ontra application, which is processed only to deliver the service and deleted afterwards. The period is effectively stated as none at that layer and the customer is not offered a setting, which is why this is recorded as a disclosed fixed window rather than customer-configurable. The commitment is scoped to the model layer: no located material states how long documents, extracted obligation data or generated outputs are retained inside the platform itself, which is a repository product by design.

No located public material states how long prompts, Persi outputs, proposals or matter data are retained. The privacy policy of October 2021 contains no retention section and addresses deletion only through GDPR erasure rights, which are exercisable on request rather than a stated retention position. The security page describes access logging and audit logging across platform activity, which implies durable records, but no period is given and no customer control is offered. Checked 1 September 2026.

No located public material addresses ethical walls, matter-level segregation or tenant separation. The security page, pricing page, AI capabilities page, Terms of Service and home page were checked on 1 September 2026. The closest material runs the other way: Terms clause 10.2 states that other Users with appropriate access rights may have access to a customer’s User Data. Atlas lists advanced user permissions as a licence inclusion, and the MCP server is described as read-only, but neither is presented as a segregation control and no document describes how any boundary is enforced.

No located public material addresses ethical walls, matter-level segregation or tenant separation. The security page publishes role-based access controls, SSO and full audit logging, and the privacy policy states employees are prohibited from viewing customer data except where necessary to diagnose a problem. Both describe access control in general terms rather than segregation between matters or between customers, and neither describes how any boundary is enforced at query time. Checked 1 September 2026.

The privacy policy states Ontra may access, preserve and disclose any information it stores to external parties where it believes in good faith that doing so is required or appropriate, giving law enforcement and national security requests and legal process such as a court order or subpoena as examples. The standard is Ontra’s own good-faith judgement and extends to what it considers appropriate, not only what is legally required. No commitment to notify the customer, and no carve-out for notice where lawfully permitted, was located on 1 September 2026. The policy does commit to reasonable efforts to notify before a change of control, which is a different event.

The privacy policy lists government agencies, regulatory bodies and law enforcement agencies among the parties to whom personal information may be disclosed, on a standard that extends past legal compulsion to what is authorised or permitted by law. No commitment to notify the customer, and no carve-out for notice where lawfully permitted, was located on 1 September 2026, and no transparency report is published. The policy does commit to notice in a different situation, where assets or businesses are transferred to a third party.

The product does not retrieve primary law, so there is no legal corpus to source. Ontra’s AI operates on the customer’s own fund documents, side letters, credit agreements and playbooks, plus the Ontra Market Playbook used as a negotiation baseline. No public material identifies any primary law source, licence basis or update cadence, checked on 1 September 2026. Recorded as not addressed because the question does not arise for this product class rather than because the vendor declined to answer it.

The product does not retrieve primary law, so there is no legal corpus to source. PERSUIT’s corpus is its own transaction dataset of law firm proposals and pricing, which the vendor describes and dates precisely as covering more than $22 billion in proposals from over 4,800 firms, sourced from platform transactions as of February 2026. That is a commercial dataset rather than primary legal authority, and no public material identifies any statutory or case law source, licence basis or update cadence for legal content. Recorded as not addressed because the question does not arise for this product class.

No citator, and none would apply. Ontra’s outputs are summaries, clause comparisons and obligation data drawn from the customer’s own agreements rather than from case law or legislation, so there is no authority whose subsequent history could be checked. Nothing on the AI capabilities, product or security pages addresses currency of underlying legal authority. Checked 1 September 2026.

No citator, and none would apply. Persi’s outputs are fee benchmarks, firm shortlists, proposal comparisons and spend dashboards derived from commercial transaction data, not citations to legal authority whose subsequent history could be checked. Nothing on the Persi, platform or data intelligence pages addresses legal authority at all. Checked 1 September 2026.

No located public material addresses what the product does when it cannot ground an answer. The security page describes subject-matter experts who review outputs and flag potential hallucinations, but that is human detection after generation rather than a documented abstention path in the product, and no confidence signal or no-answer behaviour is described. The AI capabilities page, the product pages and the security page were checked on 1 September 2026. Notable because AI Search is marketed as answering questions across a firm’s full agreement set.

No located public material addresses what Persi does when it cannot ground a recommendation. The Persi page illustrates six workflows and every one returns a confident answer with figures attached, including benchmark ranges and comparable matter counts; none shows or describes an abstention path, a confidence signal, or behaviour where the dataset holds too few comparable matters to support a benchmark. Checked across the Persi, platform, data intelligence and security pages on 1 September 2026.

No court order, opinion or disciplinary record naming this product has been located. The AI Hallucination Cases database maintained by Damien Charlotin was searched on 1 September 2026 on the product name and on the former company name InCloudCounsel, alongside general sanctions coverage, and nothing naming the product was found. This is a statement about the public record rather than a finding about the product. Ontra’s outputs are contract and fund-document work product rather than citations to legal authority.

No court order, opinion or disciplinary record naming this product has been located. The AI Hallucination Cases database maintained by Damien Charlotin was searched on 1 September 2026 on the product name alongside general sanctions coverage, and nothing naming the product was found. This is a statement about the public record rather than a finding about the product. Persi produces sourcing and pricing recommendations rather than citations to legal authority, so the failure mode this signal tracks is not one the product exhibits.

No located public material engages with bar or ethics guidance. ABA Formal Opinion 512 is not mentioned and no state bar or overseas regulator opinion is named across the home page, AI capabilities page, security page, pricing page or Terms of Service, checked 1 September 2026. Terms clause 12.2 states Ontra is not a law firm and forms no attorney-client relationship, which is a positioning statement rather than engagement with the guidance its buyers and its own legal network members are bound by.

No located public material engages with bar or ethics guidance. ABA Formal Opinion 512 is not named, and no state bar, Law Society or overseas regulator opinion appears across the home page, Persi page, platform pages, security page, integrations page, proof page or privacy policy, checked 1 September 2026. Notable because the platform is sold to law firm partners as well as in-house teams, and because outside counsel guideline compliance is a workflow the product itself enforces.

Public materials claim cost control and efficiency — outsourcing routine negotiations to control costs, decreasing entity management costs, and freeing teams for higher-value work — without addressing how AI-assisted work should be recorded or disclosed on a bill. The published pricing structure is the cost of the technology and service to the buyer, which under the standing rule is not what this signal tracks. No audit record of AI-assisted work and no fee or disclosure guidance was located on 1 September 2026.

Savings claims are the most prominent content on the site — 26% average savings per matter, 30% with competitive sourcing, and a three-year savings target reached in nine months at SAP — and no public material addresses how AI-assisted work should be recorded or disclosed on a bill. The product does produce a durable record of agreed fees against actuals and flags invoices that breach an AFA cap, but that record covers outside counsel billing generally rather than AI-assisted work specifically, which is the distinction this signal turns on. No guidance is published on fee or disclosure treatment where AI compresses the work.

A subprocessor list is published on Ontra’s trust centre at trust.ontra.ai, which also invites subscription to email notifications when subprocessors change. The portal is publicly linked from the security page and no agreement or sales conversation was required to reach it. Its body did not render to automated retrieval on 1 September 2026, so the contents of the list were not read and it is not assumed that model providers appear on it; no model provider is named anywhere on the readable site. Recorded on the existence of the published list and its change-notification mechanism.

The material exists and is explicitly gated. The security page enumerates what the SafeBase Trust Center holds — subprocessor list, penetration test report, network diagram, data flow diagrams, SOC reports, ISO certificate and Statement of Applicability — and states that most documents are available under NDA as part of a standard evaluation, with access granted rather than immediate. No subprocessor list is published openly and no model provider is named anywhere on the site, so a firm could not evidence whose model sees client content without first entering an NDA. Checked 1 September 2026.

No located public material addresses court disclosure or AI-use certification. Nothing describes an exportable record of which model produced which passage or who verified it, checked across the AI capabilities, product, pricing and security pages on 1 September 2026. Insight publishes side-letter compliance exports and Ontra cites an SEC production at Linden Capital Partners, but both are regulatory and investor reporting artifacts rather than a filing disclosure record. The product serves fund legal and compliance teams rather than litigators, so the question rarely arises in practice.

No located public material addresses court disclosure or AI-use certification. The platform produces a full audit trail of firm selection decisions, proposals and pricing changes, which the security page describes as covering every firm selection, every decision and every change, but nothing frames that as a disclosure record and nothing identifies which model produced which recommendation or who verified it. Checked 1 September 2026. The product governs procurement of legal services rather than producing court filings, so the question rarely arises for its buyer.