LatticeFlow AI is a Zurich-based AI governance platform, built on the argument that governance is still stuck in slide decks, checklists and PDFs. Atlas is a public registry of AI frameworks mapped to ready-to-run evaluations, covering more than 40 frameworks, 400 controls and 100 risks across OWASP, NIST AI RMF and ISO 42001. Evaluate runs use-case-specific tests with custom metrics defined in a readable, repeatable YAML spec, with more than 100 ready-to-run evaluations and traceability. Govern turns metrics into risk decisions using a deterministic risk methodology that accounts for impact, domain and deployment context.
For agentic systems it runs automated red-teaming and system-level security checks aligned with OWASP and MITRE ATLAS, and it addresses the EU AI Act through its COMPL-AI framework. The deterministic scoring is a deliberate design choice, since risk decisions that an auditor can reproduce are more defensible than one model’s opinion of another. The site cites Gartner recognition for AI governance platforms in 2026 and lists customers including Unique AI, Axpo and PastaHR; those are vendor-presented.
Pricing is not published and a demo is the route in, although a Get Started option and public documentation suggest some hands-on access. Test it against one real system: the useful question is whether the evaluations it runs match the risks your regulators and customers actually ask about.
