Agent infrastructure: memory, context, governance & security

Visa used Mythos to hunt for bugs in its own payment network, then open-sourced the harness that made it possible. Visa uses Anthropic’s Mythos to find deep exploit chains and open‑sources the agentic harness, giving engineers a concrete pattern for governed security hunts and audit-ready agent workflows (Principles 07, 14).

MinIO pitches persistent memory for agents with work to finish. MinIO launches AIStor Memory to give agents persistent, enterprise-controlled memory that unifies storage, workspaces, and secrets—making stateful agent behavior, retrievability, and governance practical for production (Principles 06, 10).

Fenergo bets on governed AI to fix compliance at scale. Fenergo launches Fen‑AI, a governed agentic orchestration platform that automates KYC/CLM while preserving auditability and regulatory controls—an example of Principle 09 applied to regulated workflows (Principles 09, 15).

Modus Raises $10M Seed Led by Insight Partners as Enterprise AI Creates Demand for a New Context Layer. Modus builds a Context Warehouse to help enterprise agents prioritize which information matters, illustrating the essential context layer outcome engineers must design to avoid brittleness and scaling failures (Principles 06, 11).

OpenAI releases early open-source Codex Security CLI for scanning repos and CI/CD checks. OpenAI publishes an open-source CLI to scan repos, verify fixes, and add CI checks—giving teams a practical tool to embed model-driven code audits and automated security gates into delivery pipelines (Principles 14, 15).