Built by the people who built the evals.
Lattice/AI is founder-led. Started by an ex-ML infrastructure and policy practitioner who previously built internal evaluation suites for LLM labs and enterprise teams. We start small and serious — and grow the team alongside our clients, not ahead of them.
We turn down work we can’t put our name on. That’s the filter.
AI should not be deployed on trust alone. We exist to make trustworthy evaluation practical — turning uncertainty into clear frameworks, evidence-based findings, and actionable runbooks. Whoever ships the AI, we’re the ones who sign that it’s ready.
Founder-led. By design.
Lattice/AI is small by discipline, not by accident. Every artifact on this site was written by the person who did the work — the same person you brief is the same person who signs the report.
The core practice covers:
- LLM and agent evaluation methodology
- Adversarial red-teaming and prompt-injection research
- Retrieval and RAG grounding evaluation
- Governance mapping (EU AI Act, NIST AI RMF, ISO 42001)
- Engineering integration — shadow gates, regression suites, CI evals
When the work calls for it, specialists join — regulated-industry counsel, vertical domain experts, security researchers — named in the brief and named on the report. Out when the engagement closes.
Tell us what’s under contract.
Three sentences is enough. The thing you’re trying to ship. The deadline. The thing that scares you about it.