← Trust Directory

Cognition

cognition.ai · 1 assessment

Cognition - organisation
Do not approve on current evidence
2 blocking issues to resolve before signing
Assessed Oct 21, 2026 · public evidence coverage 50% · evidence confidence medium · methodology 0.7.1

Certifications

What the vendor claims, and how far each claim has been independently corroborated. A claim is only ever as strong as the rung it reaches.

SOC 2 Type 2
Vendor claimed only

Listed on the Trust Center with report and bridge letter available under NDA; referenced in the Security Exhibit and DPA (the DPA calls it a 'certification'). Audit period, auditor and in-scope services not public.

  • ✓ Vendor claimed
  • ✓ Evidence cited
  • — Registry corroborated
  • — Scope verified
  • — Current
ISO/IEC 27001:2022
Vendor claimed only

Listed on the Trust Center and referenced in the Security Exhibit. Certificate, certification body and scope statement not public.

  • ✓ Vendor claimed
  • ✓ Evidence cited
  • — Registry corroborated
  • — Scope verified
  • — Current
FedRAMP High (In-Process)
Vendor claimed only

Stated as 'FedRAMP Class D (High) In-Process' and listed on the FedRAMP Marketplace for 'Cognition's entire platform' (July 2026). In-Process is not an Authorized status.

  • ✓ Vendor claimed
  • ✓ Evidence cited
  • — Registry corroborated
  • — Scope verified
  • — Current
ISO/IEC 42001
Not claimed

Not claimed in any public source scanned. This is the AI-management-system certification — in its absence, the vendor’s AI governance rests on its general security and privacy certifications.

  • — Vendor claimed
  • — Evidence cited
  • — Registry corroborated
  • — Scope verified
  • — Current

Supply chain

Third parties this vendor’s AI depends on, as disclosed in its own public material. Responsibility transfers; accountability doesn’t.

Models: SWE-2 (Cognition first-party)
AI providers: OpenAI, Anthropic, Microsoft/Azure (model processing), Google/Vertex, Databricks, xAI, Snowflake
Infrastructure: Microsoft Azure, AWS, Google Cloud
Subprocessors: Auth0, Hex, Sentry, Datadog, Zendesk, Pylon, Decagon, Agumbe (Cognition India)

How to read this

This assessment is automated and point-in-time, built only from evidence the vendor publishes publicly plus checks against official certification registries. It is not an audit, not a certification, and not an endorsement. A low score means public evidence was thin or uncorroborated - which is a finding about disclosure, not proof of a weak control environment. Where too little was found to characterise a product at all, no score or grade is published rather than a low one: absence of evidence is not a number, and a letter grade would read as a verdict on the vendor when it would only be a verdict on what we could collect. This record reflects the evidence available on Oct 21, 2026.