Blog
Perspectives on AI safety, cryptographic verification, and building trust infrastructure for autonomous agents.
Why Audit Logs Aren't Proof
Audit logs tell you what happened. Ed25519 signatures prove who authorized it — and make it impossible to deny. Here's why the distinction matters for AI agents, compliance teams, and incident response.
Securing CI/CD Pipelines with Human Approval Gates
How to add cryptographic human approval gates to GitHub, Bitbucket, and GitLab pipelines — so production deployments require real, verifiable authorization.
From Slack Approvals to Signed Proof: Why We Built SignedApproval
The origin story behind SignedApproval — why Slack-based approvals and email links aren't good enough for high-stakes AI actions, and what we built instead.
Integrating SignedApproval with LangChain, CrewAI, and OpenAI Agents
A practical guide to adding cryptographic human-in-the-loop approval to LangChain, CrewAI, and OpenAI agent workflows — with real code examples.
Ed25519 Signatures Explained: How SignedApproval Creates Unforgeable Proof
A deep technical dive into the Ed25519 signing flow — canonical payloads, asymmetric cryptography, and why anyone can verify a SignedApproval credential offline.
Why AI Agents Need Cryptographic Human Approval
As AI agents gain autonomy over code execution, financial transactions, and infrastructure, proving a human actually approved a high-stakes action becomes critical. Here's why current solutions fall short.