Blog
Highlights from the Currai blog: the posts worth reading first.
A complete, code-first guide to turning OpenAI Realtime voice conversations and text-to-speech generations into readable Currai User Stories and correctly nested traces.
A step-by-step guide to importing completed Retell AI calls into Currai with signed call-ended webhooks—without adding capture code to the live voice agent.
A step-by-step guide to importing completed Vapi calls into Currai with an authenticated webhook—without adding capture code or sharing a Vapi private API key.
Browse implementation notes, observability guides, product decisions, and workflow ideas by topic.
A risk-weighted evaluation framework for deciding which agent actions can run automatically, which need review, and which should remain blocked.
Read moreContext trimming, tool edits, retry policies, and model-specific tuning can degrade agents without breaking tests. Use paired trajectory evals to catch silent harness regressions.
Read moreComputer-use agents turn model decisions into clicks, typing, uploads, and transactions. Evaluate permissions, sensitive data, reversibility, recovery, and honest handoff before deployment.
Read moreCoding agents can retrieve benchmark answers, exploit weak graders, and optimize for artifacts instead of the task. Here is how to build evals that still measure real capability.
Read moreAgent failures cross model, retrieval, tool, orchestration, and runtime boundaries. Use first-divergence analysis to assign the failure to the layer that can actually fix it.
Read more