---
metadata:
    description: Notes on running LLM agents in production, and what building Millstream has taught so far.
    title: Blog
---

# Blog

Notes on running LLM agents in production, and what building Millstream has taught so far.

---

## Latest

[Self-improving software starts where the data has labels](/blog/self-improving-pipelines). Agents write the code, and
people still carry production signal back to them. What the backward pass looks like in a running pipeline engine, and
what it proves before a change ships. 5 November 2026.

---

## Also worth reading

[Why Millstream](/why). Why the work teams build around LLMs deserves a runtime of its own.

[Agents in production, a checklist](/docs/production-checklist). What to check before an agent step goes live.

---

## Next

[How pipelines learn](/platform/self-improvement) · [See the platform](/platform/) · [Roadmap](/roadmap/)
