Intelligo Group

The removal test returns a different business rather than a diminished product. Clarity's models collect and read across millions of structured and unstructured records, apply natural language processing to media and filings, resolve entities and identify patterns, and the company describes itself as the first platform to train artificial intelligence to collect and analyse data with the mindset of a human due diligence researcher, refined through tens of thousands of feedback points.

Strip that and what remains is manual analyst research, which is the incumbent model the company was built to displace and the reason it can price well below the traditional market. Human validation sits on the output rather than underneath it. Same shape as Kobalt Labs, where a reasoning engine reads counterparty material and analysts adjudicate.

Human review is structural rather than aspirational and the company sells it as part of the product. Reports are described as automated and analyst enhanced, with machine output validated by human analysts before it reaches the client, and the positioning throughout blends artificial intelligence with human expert analysis rather than presenting the model as sufficient.

For a product whose output is an adverse finding about a named person, an identifiable analyst standing behind each report is the right control and more than most of this index publishes. What is not described is the boundary inside that process: no statement of what an analyst reviews versus samples, whether low risk reports pass unreviewed, or how a disagreement between model and analyst is resolved.

Two real mechanisms and one absent number. The company states that its technology has been trained through tens of thousands of feedback points specifically to reduce false positives, and human analyst validation sits on every report, so a client receives filtered rather than raw output. What is not published is any measure of either error direction, and the asymmetry recorded across this index applies with unusual force here.

Due diligence exists to catch the thing nobody else found, so its defining failure is the false negative, the fraud or the undisclosed history the report missed, and the vendor markets improvement on false positives while publishing no detection rate, no coverage measure and no benchmark against known cases.

References are named with roles rather than logos, which is the useful kind: the head of operational due diligence at Hamilton Lane, a major private markets firm, appears on the company's own site alongside compliance counsel at Techstars and a named principal at Stonehaven. A distribution partnership with Dasseti embeds the checks inside an operational due diligence platform investors already use.

Roughly 44 million dollars raised across four rounds, and the company was acquired by Carrick Capital Partners in September 2025, which is a diligence event of its own since a private equity buyer examined the business before purchasing it. What is missing is measurement: no client count, no report volume, no coverage figure and no accuracy or detection outcome anywhere.

No data boundary statement was located. The published account of model improvement points inward rather than across customers, with algorithms described as improving through continuous feedback from the company's own research and product teams rather than from client activity, which is a better default than unbounded cross client learning and stops short of foreclosing it.

What is not addressed is whether a report commissioned by one institution, or the subject profile built to produce it, informs anything served to another, which matters because two competing funds diligencing the same manager is an ordinary occurrence in this market.

No privacy policy detail, retention schedule, subprocessor list or deletion commitment was located, and the subject matter makes that absence unusually consequential. The platform assembles a comprehensive profile of a named individual from public records, legal filings, regulatory data, news media and social media, and continuous monitoring means the profile is maintained indefinitely rather than produced once.

One stated source deserves specific attention: checks run against data leak records, which means material obtained through breaches of other organisations is used as research input, and nothing published addresses the provenance, lawfulness or accuracy standards applied to it. The person profiled has no relationship with Intelligo, is not told a profile exists, and cannot see what it contains.

No attestation, certification, trust centre or dedicated security page was located, and no service organisation control report or international information security standard certificate is announced or offered on request. That is a conspicuous gap for a platform holding assembled dossiers on executives, fund managers and portfolio company principals, since a breach here would expose not the clients' data but detailed risk profiles of third parties who never consented to their creation. Institutional allocators run vendor security reviews as a matter of course and this is the first document they would request.

Anti bribery and anti money laundering are named as use cases the platform supports and no supervisor, statute or admission process appears anywhere. One regime is directly and obviously engaged and never mentioned: background reports used for employment decisions in the United States fall under the federal consumer reporting statute, which imposes accuracy obligations, adverse action notice duties and a dispute right for the subject, and the company markets candidate screening alongside its investment diligence products. Whether it operates as a consumer reporting agency for that line, and what that means for subjects, is unaddressed in located material.

The output is an adverse finding about a named individual and nothing published addresses how evenly the system produces one. Every technique involved carries a known and uneven error profile. Name matching and entity resolution degrade with transliteration, patronymics and non Western naming conventions, which is the finding already recorded for Quantexa and Bretton AI.

Adverse media coverage is heavily English language weighted, so a subject from a thinly covered media market generates few hits while one from a saturated market generates many, and neither reflects conduct. Social media analysis penalises both large digital footprints and absent ones. Legal record availability varies by jurisdiction to the point of incomparability.

The consequence for a wrongly flagged subject is losing an investment, a board seat or a job, decided by a report they never see. No demographic or jurisdictional error analysis, no fairness testing and no correction path was located.

Nothing was located on either half. No guarantee, indemnity or falsifiable accuracy commitment binds the vendor to a report a client will act on, and no route of any kind exists for the person the report is about. The subject is not notified that a check was run, cannot obtain the report, cannot see which sources produced an adverse finding and has no described mechanism to correct an error, and continuous monitoring means the profile persists and updates without their knowledge indefinitely.

The gap is sharpest where the same platform is used for employment screening, because the federal consumer reporting regime grants exactly those rights to a candidate and nothing published describes how they are honoured.

Sources are described by category and never by name. Reports are stated to draw on publicly available information plus premium sources, spanning global corporate registries, legal and asset records, regulatory data, news media, social platforms and data leak records, and not one provider, aggregator or licensed database behind those categories is identified.

That matters more than usual because the reliability of an adverse finding depends entirely on which underlying source produced it, and a client cannot weigh a hit without knowing where it came from. No model provider, hosting arrangement or subprocessor is named either.

One integration is named and it is well chosen: a partnership with Dasseti places the checks inside an operational due diligence platform that institutional allocators already run, so a client can commission a background check without leaving the workflow where manager diligence happens. Beyond that the platform is a standalone software as a service destination.

No integration to a customer relationship system, compliance case management platform, investor portal or data warehouse is named, and no developer documentation or application programming interface reference was located, so a firm cannot determine how reports and monitoring alerts reach the systems its analysts actually work in.

No hosting provider, region selection, residency commitment or private deployment option was located. The company originates in Israel with a United States commercial presence and serves institutional investors internationally, which places subject data, much of it about European and other non United States individuals, across several transfer regimes at once.

For a platform whose entire payload is profiles of named people, where those profiles rest and under whose law is a question a client's own privacy review would raise, and nothing published answers it.

No rate card or pricing basis is published and the site states only that pricing options vary by use case and need. One relative claim exists and it is old, reported at launch stage as pricing at roughly half the average market rate for a comparable check, which tells a buyer the positioning but not the number and dates from a period when the product and company were much smaller. For a product sold per report or per monitored subject, the unit and the rate are exactly what a prospective buyer needs and neither appears.

The financial buyer set is wide and consistently named: banks, investment funds, insurance companies, institutional investors allocating to managers, and private equity firms evaluating portfolio company management. Use cases span pre investment diligence, merger and acquisition diligence, ongoing monitoring, asset tracing, anti bribery and anti money laundering work and executive checks, so the platform is bought by compliance, operational due diligence and investment teams inside the same institution.

The dilution is that the same platform also serves employment screening for job candidates and general corporate vetting, which is a different market with a different regulatory frame, and the company does not maintain them as visibly separate propositions.