R

Responsiv

The models do the interpretive work, and the platform around them would still stand without them. The vendor's material describes the models breaking hundreds of pages of regulatory text into an executable checklist of requirements, mapping those requirements against a firm's policies and controls to surface gaps, and answering regulatory questions with citations. Policy Scanner, released 17 June 2026, reads an uploaded policy and ranks the rules driving it, section by section.

Underneath sits a compliance system a firm could run without any of that: a register of applicable rules, a policy library, alert routing by business unit and jurisdiction, task assignment and an audit trail. Removing the models would leave that system standing.

Output traces to the rule it came from, and no accuracy figure is published. The Platform page says every answer cites the exact regulation, article and paragraph so a reader can verify it in seconds. The Policy Scanner release shows each suggested obligation citing a specific rule, such as 15 U.S.C. section 78m, with a short rationale and a relevance score marking it a direct policy driver or a control reinforcer. Section 3(b) of the Terms acknowledges that machine learning is probabilistic and that use of the service may produce incorrect output that does not accurately reflect real cases or facts.

No measurement is published: no test set, no accuracy figure, and no evaluation of how often a driving rule is missed, on a product sold on exam defensibility.

A person decides, and the decision point is built into the product. The Policy Scanner release says the tool does not replace judgment and that the user still decides what a mapping means. Suggestions arrive as a ranked list a reviewer accepts or dismisses one at a time, with the option to adjust the scope of the analysis with a prompt and regenerate. Section 3(b) of the Terms tells the customer to evaluate output as appropriate, including by human review by a knowledgeable attorney or expert.

Alerts and obligations are routed to named owners who confirm completion. No limit the system imposes on itself is published: nothing states when it declines to suggest, what a low relevance score means in practice, or what its output may not be used for.

Real users speak on the record, and neither the customer nor the result can be pinned down. The homepage carries four quotes with names and titles, three from legal or compliance leaders, including a General Counsel describing lower spending on outside counsel; none names an employer. The one case study, dated 3 June 2026, describes an independent investment bank of about 1,000 employees across the US, UK, Ireland and the EU that built a register of its obligations.

It quotes the bank's Director of Compliance, but the firm is not named and the only figures given are headcount and footprint. Nothing published states time saved, coverage found or any other measured outcome.

The homepage promises and the agreement point in different directions, and the agreement is what a customer signs. The homepage says data remains the customer's. The Terms of Service let Responsiv train on what a customer puts in, and section 3(a) lets it keep using Output within the service and in its business generally, in both cases unless a separate enterprise agreement says otherwise. The material a firm uploads, including its policies and procedures, sits outside the Privacy Policy, the only published data commitment. Nothing addresses segregation between customers, privilege or work product, or what any model provider may retain.

What the product is and who it is for are stated in full in the Terms. Section 2(d) calls it an informational system intended for attorneys, paralegals, legal researchers and others engaged in legal research. Responsiv is not a licensed attorney, the output is not legal advice, no attorney client relationship forms, and users are told not to disregard or delay professional advice on the strength of it. The review instruction in section 3(b) reaches supervision too.

Jurisdiction is the gap: the platform covers more than 50 jurisdictions, and nothing states where its analysis is reliable, where local qualification matters, or which regimes it does not cover.

A Responsible AI section sits on the homepage, with nothing behind it that could be audited. It makes three claims: that the models deliver precise, reviewable, audited information specific to the industry, that customer data is never used for training, and that sensitive data can be removed before analysis. No one inside Responsiv is named as accountable for model behavior, nothing describes what is tested before a release ships, and no evaluation results are published.

Nothing addresses whether the models perform unevenly across regulators, jurisdictions or document types, on a product sold for its breadth of coverage.

What is published covers the company's handling of personal data, not the firm's documents. The Privacy Policy, last updated 14 May 2024, describes appropriate technical and organizational security measures, says Responsiv may notify of a breach consistent with applicable law, and keeps personal data only as long as necessary. It then says expressly that it does not apply to Customer Data, which is what a compliance team actually uploads.

The Terms describe collecting performance data that may include Content to monitor and improve the service, and allow parts of it to be subcontracted. No retention period or deletion commitment for uploaded policies, no access control model, no subprocessor list and no incident notification timeline is published.

The published position puts the risk of a wrong answer on the customer. Section 6 of the Terms disclaims all warranties, including fitness, uninterrupted operation, any level of security and any business result, and caps direct damages at one hundred dollars in aggregate across the customer, its employer and its affiliates. Indirect and consequential damages are excluded entirely. Section 5 runs an indemnity from the customer to Responsiv, and none runs back.

No warranty attaches to the requirements, gap findings or citations the product generates, and no insurance position appears. The separate agreement the Terms refer to, which would carry an enterprise allocation, is not published.

No integration into the systems a compliance or legal team already runs is published. The platform is a place to upload policies, maintain the register and assign work, with dashboards and alerts set per team. Nothing names a connection to a document management system, a GRC or risk platform, a policy repository, email or a ticketing tool, and no API, developer documentation or integrations page exists. The case study describes policies scanned after upload, a manual route in. Searches for integration and API material found none.

Where the software runs and where a firm's policies are stored are not published. The site is specific about regulatory geography, covering all 50 US states, the Americas, Europe, Asia and the Middle East, but that is the coverage of the content rather than the location of the processing. No tenancy model is described, no hosting provider or region is named, and neither the Terms nor the Privacy Policy states a processing or storage location.

The vendor's own case study describes a customer operating across the US, UK, Ireland and the EU, and nothing published answers where such a customer's data would sit.

No independent security attestation is published. There is no security page and no trust center, and no SOC 2, ISO 27001, named auditor, penetration test or other certification is mentioned anywhere on the site. The only security language is the Privacy Policy's statement that appropriate technical and organizational measures are in place, and its acknowledgment that absolute security cannot be guaranteed. Searches for a trust center or SOC 2 report found none.

The models are presented as the vendor's own and never identified. No model, version or provider is named, nothing states whether inference runs on Responsiv's own infrastructure or a third party's, and no commitment to notify customers when the models change is published. The Terms reserve the right to subcontract portions of the service without naming anyone. Who processes a firm's policies cannot be learned from the published record.

No price, rate or unit of charge is published. There is no pricing page in the navigation or the footer, and every route through the site ends at a demo request. The Terms of Service are the only commercial material. They say Responsiv may charge fees at its discretion, create different levels of access and offer free trials, and they refer to a free tier a user may not open multiple accounts to exploit. No tier is named, described or priced, and fees are otherwise left to a separate agreement.

The buyer and the regulatory scope are set out in detail. Three industry pages address financial services, from banks and broker dealers to asset managers; health and property and casualty insurance; and consumer technology, with rules on minors' online safety, dark patterns and privacy among them. Subject areas are named down to AML, recordkeeping and market conduct, across more than 50 jurisdictions. The boundary is not stated: nothing says which regulators or regimes fall outside the coverage, how current the content is for a given source, or what size of firm the platform is built for.