Refolk
FrameworkSales and go-to-market

The Account AI-Review Gate Read: Council, Security-Led, Ad Hoc, or Open

Grade any target account's AI approval gate from six public signals and leave with a gate type, named seats, a review-time band, and a pre-wire move before first touch.

23 min readLast reviewed November 22, 2026Read as Markdown

Key takeaways

  • Headline proxy figures overstate AI gates: in 2024 more than 31% of the S&P 500 disclosed some AI oversight, but only 11% disclosed explicit board or committee oversight.
  • A verified enterprise-scope ISO/IEC 42001 certificate is a strong gate signal because only 271 existed worldwide at the end of 2025, against 125,622 ISO 27001 certificates.
  • In Refolk's index there are 703 current US Chief AI Officers but only 73 current AI-governance title holders, so a CAIO title alone does not prove an AI council exists.
  • Supplier terms that require approval before a vendor uses AI tools imply an intake gate; terms that require disclosure only upon request do not.
  • Vendor-reported AI reviews run 8 to 12 weeks for high or regulated tiers but 5 to 10 business days for vendors holding SOC 2 Type II, ISO 27001 and documented AI governance, so your own evidence pack sets much of the cycle.
  • Score six dimensions 0 to 2 on quotable evidence only, assign the gate type with a first-match rule, and have a second rep grade blind before the verdict counts.

This is the outside-in read I run before spending first touches on a target account for an AI product. It tells you which AI approval gate the account runs, who staffs it, and whether to pre-wire it, plan around it, or move the account to next quarter. It is written for founders selling their own product, account executives, SDR leads, and partnerships teams. You get six scored evidence dimensions, a first-match rule that turns the scores into one of four gate types, and a seat list, review-time band, and pre-wire action for each type. It is tight enough that two reps grading the same account reach the same verdict.

What is an AI-review gate, and what are the four types?

An AI-review gate is the approval step between "the champion wants this" and "we can sign", where someone other than the buyer checks how a vendor uses AI. From public evidence, an account's gate falls into one of four types: Council, Security-led, Ad hoc, or Open.

The usual advice is to ask your champion who reviews AI purchases. That works on call two, but by then you have already chosen who gets the first touch and how to forecast the deal. The stall this read prevents is a council or security veto nobody saw until after the champion said yes. Grading the gate before contact turns that surprise into an account attribute you can plan around.

Gate typeWhat the public evidence showsWho staffs itDefault action
CouncilNamed management body or enterprise-scope certificateGovernance lead, report signer, security reviewerPre-wire: governance seat in first wave
Security-ledApproval verb in supplier terms, mandatory exposure, or board oversight onlyThird-party risk, infosec, privacyPlan around: champion requests vendor pack in meeting one
Ad hocA policy page or a posting, no standing bodyPrivacy, legal, or IT, whoever holds itPlan around: arm the champion, map three reviewers
OpenNothing on any dimensionChampion and standard procurementProceed, re-grade next quarter

"Deprioritize" is not a fifth gate type. It is an action that can land on a Council or Security-led account when your own evidence pack is not ready and the review band runs past the weeks left in the quarter. The verdict section gives the exact rule.

How much public evidence of AI oversight actually exists?

Less than the headline numbers suggest. ISS-Corporate found more than 31% of the S&P 500 disclosed some board oversight of AI in 2024, but only 11% gave specific disclosure of full board or committee oversight. That roughly 3x gap is mostly director expertise being counted as oversight.

ISS counts three things as "oversight": a board committee with oversight responsibility, a director knowledgeable about the technology, or an AI ethics panel. Only the first tells you anything about who might review a vendor. Read every headline figure with that in mind.

CohortYearMetricValue
S&P 5002022-23Any AI oversight15%+
S&P 5002024Any AI oversight31%+
S&P 5002024Explicit board or committee oversight11%
S&P 1002025Board-level oversight54%
S&P 1002025Oversight plus formal AI policy28%

Sources are ISS-Corporate (rows 1 to 3), the Glass Lewis S&P 100 review (row 4), and a secondary summary (row 5). Definitions differ between sources, so treat the rows as direction, not as a time series.

S&P 100 AI oversight disclosure, 2025 proxies

  1. S&P 100 companies
    100%

    All proxy statements reviewed

  2. Board-level AI oversight disclosed
    54%

    Full board or a committee

  3. Oversight plus a formal AI policy
    28%

    Secondary summary of the same cohort

Even among the largest US companies, only 28% disclosed both board oversight of AI and a formal AI policy.

A few more facts change how you read a filing:

  • Committee versus full board. Of the S&P 100 companies disclosing board oversight, a secondary summary puts 63% with a specific committee, most often audit or technology, and 37% with the full board.
  • Committees are growing. EY's 2025 proxy review found the number of S&P 500 companies naming a committee with AI oversight more than tripled, with audit committees the primary choice.
  • Expertise is not a gate. Nearly half of Fortune 100 proxies cited AI in director qualifications in 2025, almost double the 26% in 2024. That is the easiest thing to misread as oversight.
  • Sector matters. In the 2022-23 cycle, IT companies disclosed AI oversight at 38% and Health Care at 18%. A later sector breakdown is not established publicly.

Certification is rarer still. ISO/IEC 42001 is the international standard for an AI management system, meaning the policies and controls an organisation uses to govern its own AI. The ISO Survey counted 271 ISO/IEC 42001 certificates worldwide at the end of 2025, against 125,622 for ISO 27001.

1 in 464
ISO/IEC 42001 certificates per ISO 27001 certificate worldwide, end of 2025
271 against 125,622, which is why a verified enterprise-scope 42001 certificate is a strong marker of a formal AI gate.

The practical consequence is that most accounts, including large ones, score low on filings, and private companies file no proxy at all. A blank filing is weak evidence that there is no gate. That is why the read uses six dimensions rather than leaning on the proxy.

How do you score the six evidence dimensions?

Score each dimension 0, 1, or 2, and only on evidence you can quote with its source and section. The total runs from 0 to 12 and tells you how much evidence you hold. The gate type comes from which dimensions scored, using the first-match rule in the next section.

Dimension012
D1 Board or committee oversight in filingsNone, or director expertise onlyFull board holds AI oversightNamed committee whose charter or description names AI
D2 AI policy or responsible-AI reportNonePrinciples page, no named owner or bodyReport names a management body or signing executive
D3 ISO/IEC 42001 certificationNone, or "aligned with"Product scope, or accreditor unverifiedEnterprise scope, certifier accredited for 42001
D4 Governance seatsNoneOpen posting onlyCurrent holder of a governance title
D5 Sector and jurisdiction exposureNoneRegulated sector, but your use is out of scope or not yet applicableYour use falls in a mandatory regime now in effect
D6 AI terms in supplier pagesNoneDisclosure duty ("inform" or "upon request")Approval required before vendor AI use

The dimensions sit at different depths of the organisation. Evidence at an outer layer does not prove the inner layers exist.

Where each dimension sits in the gate

  1. External mandate
    Regulation that forces a review regardless of disclosure (D5)
  2. Board
    Committee or full-board AI oversight in the proxy (D1)
  3. Management body
    Responsible-AI report naming a council, and 42001 scope (D2, D3)
  4. Named seats
    Current governance title holders and open postings (D4)
  5. Intake desk
    Supplier terms and the security questionnaire a vendor actually meets (D6)
A board committee that watches AI risk is not proof of a management body or an intake desk beneath it.

D1: Board or committee oversight in filings

What it proves: the board has assigned someone to watch AI risk. Read the risk-oversight section of the latest DEF 14A, the annual proxy statement US public companies file, along with the committee charters. Delegated-committee language looks like Citi's Technology Committee, which "reviews trends... including... Artificial Intelligence and Machine Learning."

When it lies: the only AI mention is in director bios, or a board committee exists with no management body beneath it.

D2: Published AI policy or responsible-AI report

What it proves: someone in management owns AI conduct in public. Record who signs it and whether it names a council, committee, or office.

When it lies: it is a principles page with no owner. That scores 1, not 2.

D3: ISO/IEC 42001 certification with verified scope

What it proves: when the certificate is enterprise-scoped and the certifier is accredited, a formal AI management system exists and has been audited. Verify in two passes:

  1. Confirm the certification body is accredited for 42001 itself, on the accreditation body's own register (ANAB, UKAS, RvA, JAS-ANZ, SAC or IAS) or on IAF CertSearch. Accreditation for ISO 27001 does not extend to 42001.
  2. Read the certificate's scope statement and class it as enterprise or product.

When it lies: the certificate covers a product line. AWS's initial November 2024 certificate named Bedrock, Q Business, Textract and Transcribe. That shows those services are governed, not that procurement runs an AI gate. "Aligned with 42001" is a self-declaration and scores 0. How often certificates cover only a product line is not established publicly.

D4: Governance hiring and named roles

What it proves: there are people whose job is the gate. Current holders count for more than open postings. One posting analysis found around 71 new US AI governance postings a week, 56% of them from companies with 10,000+ employees.

When it lies: a single req at a mid-size firm is a plan, not a standing body. Growth estimates also conflict: IAPP, citing Stanford HAI, reports 17% year-over-year growth, while another source cites 150%. Do not use hiring momentum as a score input.

D5: Regulated-sector and jurisdiction exposure

What it proves: an outside rule may force a gate regardless of what the account publishes. Check HQ and EU operations, the sector (banking over $30 billion in assets, insurance, health, HR tech), and whether your product falls in an EU Annex III high-risk use or a consequential-decision use.

When it lies: the dates or scope in the rep's head are stale. These are the dates as published. Re-check each one against the regulation text or a current legal summary before you score it or cite it to a buyer.

RegimeDate as publishedWhat it means for the gate
EU AI Act, Article 50 transparencyMost obligations apply 2 Aug 2026Transparency duties, not a high-risk gate
EU Digital Omnibus, Regulation (EU) 2026/1744In force 27 Jul 2026; Annex III 2 Dec 2027; Annex I 2 Aug 2028High-risk duties postponed
Colorado revised AI lawEffective 1 Jan 2027Consequential-decision uses
California ADMT regulationsEnforcement from 1 Jan 2027Automated decision-making uses
US banks, SR 26-2 / OCC Bulletin 2026-13Issued 17 Apr 2026, replacing SR 11-7Generative and agentic AI left out of formal scope

Two details catch people. Generative AI already on the EU market before 2 Aug 2026 has until 2 Dec 2026 for machine-readable marking. One source claims Article 50 applied from 2025; that is an outlier, so do not use it.

D6: AI-vendor terms in procurement or supplier pages

What it proves: the account has written an intake rule for vendors that use AI. Record the verb, because the verb is the score:

  • Approval, scores 2. OpenAI's supplier code says suppliers "shall disclose and obtain approval for AI tools, model providers, or subprocessors."
  • Affirmative disclosure, scores 1. UPS's AI Addendum says the vendor "shall inform UPS in writing about the usage of any AI Technology."
  • Disclosure on request, scores 1. OpenText's supplier code requires disclosure "upon request", the weakest form of intake.

When it lies: disclosure language gets read as an approval gate. None of these documents, in the excerpts I have seen, names the function that owns AI intake. D6 tells you a gate exists, not who staffs it.

Who staffs the gate, and how do you build the seat list?

Start the seat list with current holders of AI-governance titles, add the executive who signs the responsible-AI report, and add a Chief AI Officer only when a governance seat reports to them. These seats are rare, so finding even one named person changes the first-touch plan.

MarketAI-governance titles"Head of AI Governance" in sample of 25"Responsible AI Lead" in sample of 25
US7322
UK4345

In Refolk's index of professional profiles, there are 73 current holders of AI-governance titles in the US and 43 in the UK, a UK-to-US ratio of 0.59. Those counts use exact-title match. That matters, because the title is unstable: "AI Governance Lead" competes with AI Compliance Officer, Responsible AI Lead, Head of AI Ethics, and AI Risk Manager. Search all five.

9.6 to 1
US Chief AI Officers per current AI-governance title holder, in Refolk's index
703 current US Chief AI Officer title holders against 73 AI-governance title holders, so the CAIO title is far more common than the seat that actually runs a gate.

Where the seats sit differs by market. In the UK sample of 25 from Refolk's index, the top employers include NatWest Group (two profiles), Legal & General, AXA UK and Howden, and 4 of the top 10 employers are banks or insurers. The US sample mixes public sector and consultancies, including JPMorganChase, the Defense Intelligence Agency and Maryland DoIT. The UK pattern is consistent with regulator pressure on financial firms, but that is my inference and it is unverified.

Use these seat lists by gate type:

  • Council: the governance lead under any of the five titles; the signer of the responsible-AI report; the security or third-party risk reviewer who runs the questionnaire; the CAIO only if a governance seat reports to them.
  • Security-led: the third-party or vendor risk manager; the information security reviewer; privacy. For a generative AI product at a US bank over $30 billion in assets, SR 26-2 leaves generative and agentic AI out of formal model-risk scope. One practitioner summary says those systems fall back on enterprise risk, information security, privacy and vendor controls. Plan on vendor risk and security as reviewers, not model validation, but confirm locally.
  • Ad hoc: IAPP data splits AI governance ownership across privacy (22%), legal (22%) and IT (17%). Name one person in each until the champion tells you who holds it.
  • Open: the champion and standard procurement.

Building this list by hand means running five title variants against each account, then checking whether each match is current and at the right company. Refolk takes the seat as a plain-English request and returns current holders across LinkedIn and the open web, so the variant problem is handled in the query rather than in your spreadsheet.

How do you turn scores into a gate type, a review band, and an action?

Apply the first-match rule: work down the table, and the first row whose condition is true sets the gate type. The gate type then sets the review band and the pre-wire action. Adjust the band for your own attestations, because much of the delay sits with the vendor, not the gate.

OrderIf this is trueGate typeReview band to forecast
1D2 = 2, or D3 = 2, or D4 = 2 with D2 = 1Council8 to 12 weeks
2D6 = 2, or D5 = 2, or D4 = 2, or D1 is 1 or 2Security-led3 to 6 weeks; 8 to 12 if D5 = 2
3Any other non-zero scoreAd hoc4 to 8 weeks
4Every dimension at 0OpenNo AI-specific leg established

The bands are my mapping of vendor-reported durations onto gate types. Here are the source figures:

ConditionDurationSource type
Standard risk tier3 to 6 weeksVendor blog, methodology unpublished
High or regulated tier8 to 12 weeksVendor blog, methodology unpublished
Vendor with no attestations4 to 8 weeksVendor blog, methodology unpublished
SOC 2 Type II + ISO 27001 + documented AI governance5 to 10 business daysVendor blog, methodology unpublished

The AI-specific block is described as typically 30 to 60 items inside a broader enterprise security review. CSA's AI-CAIQ, a standardised AI questionnaire, runs to 320 questions, according to a vendor blog. How many extra reviewers an AI review adds is not established publicly.

Attestation modifier. If you hold SOC 2 Type II, ISO 27001 and documented AI governance, shorten Security-led and Ad hoc bands toward 5 to 10 business days. I would not shorten a Council band below 3 to 6 weeks until your own closed deals show it.

Pre-wire actions by gate type:

  • Council: pre-wire. Put the governance seat in the first wave alongside the champion. Send the AI governance pack before anyone asks: model providers, subprocessors, attestations, and pre-filled answers to the AI questionnaire block.
  • Security-led: plan around. Have the champion request your vendor pack in meeting one rather than waiting for the questionnaire to arrive.
  • Ad hoc: plan around. Give the champion the pack plus a one-page note written for privacy, legal and IT, since any of them may end up holding the review.
  • Open: proceed. Run the normal motion and re-grade next quarter.

Gate formality against your pack readiness

Pack ready (attestations plus AI governance docs)Pack not ready
Ready, informal gate
Move fast; ship the pack with the first meeting
Ready, formal gate
Pre-wire the governance and security seats now
Not ready, informal gate
Plan around; arm the champion while you build the pack
Not ready, formal gate
Deprioritize this quarter unless the band fits
Informal gate (Open, Ad hoc)Formal gate (Security-led, Council)
A formal gate is only a reason to wait when your own evidence pack is not ready for it.

What is the step-by-step procedure for grading an account?

Run the eight steps below in order, about 60 minutes of research without a certificate to verify and about 95 with one, then score and calibrate. Each step ends with a dimension scored on the gate card.

Grade one account's AI-review gate

  1. Screen regulatory and sector exposure
    Check HQ, EU operations, sector, and whether your product falls in an Annex III or consequential-decision use, re-checking the regime dates. About 10 minutes; done when D5 is scored.
  2. Read the supplier terms for an AI intake verb
    Find an AI addendum or supplier code and quote the verb: approval, inform in writing, or upon request. About 10 minutes; done when D6 is scored.
  3. Read the proxy and committee charters
    Classify the AI disclosure in the latest DEF 14A as committee, full board, expertise-only, or none. 15 to 20 minutes; done when D1 has a quoted phrase and section.
  4. Check for an AI policy or responsible-AI report
    Note who signs it and whether it names a management body. About 10 minutes; done when D2 is scored.
  5. Verify any ISO/IEC 42001 certificate and its scope
    Confirm the certifier's 42001 accreditation on the accreditor's register, then read the scope statement. About 30 minutes; done when scope is classed enterprise or product.
  6. Scan for governance seats and postings
    Search current holders and open postings across all five title variants, plus any CAIO with a governance seat reporting in. About 15 minutes; done when D4 is scored and seats are named.
  7. Score, assign the gate, and set the action
    Apply the first-match rule, then fill the seat list, review band with attestation modifier, and pre-wire action. Done when the gate card is complete.
  8. Calibrate with a blind second grade
    A second rep grades without seeing the first card, and both reconcile dimension by dimension on the quotes. Done when there is one agreed verdict.

A note on order. The procedure as first assembled put the supplier-terms check sixth. Practitioner guides on security review argue for security evidence first, with the champion requesting the vendor pack rather than waiting for the questionnaire. I moved supplier terms to step 2 for that reason: an approval verb tells you, before you open a proxy, that a vendor pack has to ship with the first meeting.

AI-review gate card
Account:
Graded by:                Second grader (blind):
Date graded:

D1 Board or committee oversight (0/1/2):
   Quote + proxy section:
D2 AI policy or responsible-AI report (0/1/2):
   Quote + signer + named body:
D3 ISO/IEC 42001 (0/1/2):
   Accreditor register checked:      Scope statement:
D4 Governance seats (0/1/2):
   Named holders:                    Open reqs:
D5 Exposure (0/1/2):
   Regime:                           Date re-checked on:
D6 Supplier AI terms (0/1/2):
   Intake verb (approval / inform / upon request) + source:

Total (0-12):
First-match row fired: 1 / 2 / 3 / 4
Gate type: Council / Security-led / Ad hoc / Open
Seat list:
Review band forecast:            Attestation modifier applied: yes / no
Action: Pre-wire / Plan around / Proceed / Deprioritize this quarter
Re-grade trigger:

One card per account in the CRM account record. Every non-zero score needs a quote and a source line.

How does the gate read go wrong?

Most errors come from reading a signal at the wrong layer: board for management, product for enterprise, disclosure for approval. Each failure mode below says what happens and the check that catches it.

  1. Proxy "oversight" is really expertise only. ISS counts AI-skilled directors as oversight, so headline figures include accounts with no assigned body. Check: require a charter or committee-description sentence that names AI. Director bios alone score zero.
  2. A committee exists, but only at board level. An audit committee overseeing AI risk is not evidence of a management-level intake council. Check: look for a management body named in the responsible-AI report. Without one, the account cannot reach Council through D1.
  3. The certificate is scoped to a product. AWS's first certificate named four services, not procurement. Check: read the scope statement. "Aligned with 42001" is a self-declaration and scores zero.
  4. A false negative from the register. IAF ceased operations on 1 January 2026, CertSearch still runs, and one directory reports it may not show ISO/IEC 42001 in accredited scopes. A missing entry proves nothing. Check: go to the accreditor directly, such as ANAB or UKAS.
  5. The CAIO title is noise. The 703 US Chief AI Officer profiles in Refolk's index include small firms and mis-titled CEO and CTO profiles. Check: confirm company size and whether a governance seat reports to the CAIO before adding them to the seat list.
  6. The posting is aspirational. A single "AI Governance Manager" req at a 500-person firm is not a standing body. Check: look for current holders plus a policy page. How long a posting takes to become a disclosed committee is not established publicly, so do not forecast one.
  7. Stale regulatory read. Reps who cite SR 11-7, or who carry the original August 2026 EU high-risk dates, will mis-forecast regulated accounts. Check: re-check every D5 date against a current source and record the re-check date on the card.
  8. Supplier code mistaken for intake. OpenText requires disclosure "upon request" while OpenAI requires "approval". Check: only approval language scores 2 and implies a gate.
A board that watches AI risk is not a desk that approves AI vendors.

There is also a structural limit. Every review duration here is vendor-reported with unpublished methodology, and no public source documents this procedure end to end. The scoring rule is a standard for consistency, not a validated predictor. Treat the bands as priors until your own data replaces them.

How do you keep the read calibrated and current?

Calibrate by blind double-grading a sample of accounts each quarter, and keep the read current by re-grading on specific triggers rather than on a calendar alone. Over time, replace the vendor-reported bands with review durations measured on your own deals.

Calibration. When two reps disagree on the gate type, the cause is almost always one dimension. Reconcile on the quotes. If a quote does not settle it, score the lower value and log the open dimension as the first question for the champion. That keeps the verdict conservative and turns the disagreement into discovery.

Re-grade triggers:

  • A new proxy statement or committee charter is filed.
  • A responsible-AI report is published or re-signed.
  • A governance title holder joins, leaves, or a new req appears.
  • A D5 regime date passes or is amended.
  • The supplier code or AI addendum is revised and the intake verb changes.

Of these, the seat trigger is the hardest to watch by hand, because governance titles move between five variants and people change roles without notice. Re-running the same plain-English seat query in Refolk each quarter and diffing it against the seat list on the card is the cheapest way to catch it.

Replacing the bands. For every closed or stalled deal, record the gate type you assigned, the date the AI or security review started, and the date it cleared. After a handful of deals per gate type, your own medians are better than any figure in this guide.

Before the verdict counts

  • Every non-zero dimension on the gate card has a quoted phrase, a source, and a section.
  • D1 is scored from a charter or committee description, not from director bios.
  • Any 42001 certificate was checked against the accreditor's own register and its scope classed as enterprise or product.
  • D5 dates were re-checked against a current source and the re-check date is on the card.
  • D6 records the exact intake verb, and only approval language scored 2.
  • The seat list names people under all five governance title variants searched, and any CAIO listed has a governance seat reporting to them.
  • The review band reflects the attestation modifier for your current evidence pack.
  • A second rep graded the account blind and any unresolved dimension is logged as a champion question.

Questions practitioners ask

How do I find the AI governance committee at a target account?

Start with the proxy statement's risk-oversight section and the committee charters, then look for a responsible-AI report that names a management body. A board committee that oversees AI risk is not the same as a management council that reviews vendors, so you need the second document to confirm an intake body. For private companies with no proxy, rely on the responsible-AI page, current governance title holders, and the supplier terms.

Who approves AI vendors in an enterprise?

It depends on the gate type. Where a standing council exists, an AI governance or responsible-AI lead usually sits on it alongside a security reviewer. Where there is no council, AI questions are folded into the security or third-party risk review. Where ownership is ad hoc, IAPP data splits AI governance ownership across privacy (22%), legal (22%) and IT (17%), so map all three until you know which one holds it.

How long does an AI vendor risk review add to the sales cycle?

Vendor-reported figures put a standard-tier review at 3 to 6 weeks and a high or regulated tier at 8 to 12 weeks, falling to 5 to 10 business days for vendors holding SOC 2 Type II, ISO 27001 and documented AI governance. None of these sources publishes its methodology, so use them as starting bands and replace them with durations measured on your own closed deals.

Does a Chief AI Officer at the account mean there is an AI council?

No. Refolk's index holds 703 current US Chief AI Officer title holders against 73 current AI-governance title holders, about 9.6 to 1, and the CAIO pool includes small firms and mis-titled CEO or CTO profiles. Count a CAIO toward the gate only when the company is large enough to run one and a governance seat demonstrably reports to them.

Is an ISO 42001 certificate proof the buyer runs a formal AI gate?

Only when the certifier is accredited for ISO/IEC 42001 specifically and the certificate's scope covers the enterprise rather than a product line. Accreditation for ISO 27001 does not carry over, and "aligned with 42001" is a self-declaration. Because only 271 certificates existed worldwide at the end of 2025, a verified enterprise-scope certificate is a strong signal of a formal gate.

Why not just ask the champion who reviews AI purchases?

Ask anyway, but grade first. By the time you can ask, you have already chosen who gets the first touch and how to forecast the deal. The stall this read prevents is a council or security veto that surfaces after the champion says yes. A pre-contact grade lets you put the governance seat in the first wave and ship the vendor pack with meeting one.

Try it on the search you came here for

Stop building boolean strings. Just describe the person.

Type one sentence. I plan the search, read GitHub, public LinkedIn and Crunchbase records, and the open web as it is right now, and hand back a ranked list with the reason next to every name.

  1. 01Describe them

    One plain sentence. Role, city, stack, stage, whatever matters to you.

  2. 02I read the web live

    GitHub, public LinkedIn and Crunchbase records, the open web. Not a database that went stale last quarter.

  3. 03You read the shortlist

    Ranked, with the reasoning under every name. Open a profile, ask a follow-up, narrow it down.

  • No boolean, no filters, no seat to buy. One box.
  • Read at search time, so a profile updated yesterday counts today.
  • Every step visible as it runs, every name with its reason.

500 free credits on sign-up. No card, no demo call. See real searches.

Read next