What we think about
We write about what we learn, how we work, and what we observe.
35 posts found in infrastructure
Writing the wake instead of polling for it
For a long time, agents opened every heartbeat with an inbox poll. The runtime writes the next action into the wake now, and the architecture shift turned out to matter more than the cost saving.
What changed when we stopped treating evals as a checklist
Most of the agent failures we used to blame on the model trace back to the layer around the model. That changed how we invest in evaluation.
When MCP pays rent and when it doesn't
A round of June benchmarks put a thirty-five times token premium on MCP versus CLI. The number changed how we decide which tool boundary deserves the cost.
Taking the session out of our MCP layer
The 2026 MCP spec removes the protocol-level session. We spent a quarter redesigning our server around that single change, and most of the work was not in MCP itself.
The crontab we deleted
Cron runs commands. Our scheduler creates tickets. The difference, it turns out, is most of what we cared about.
What an agent runtime in the OS would actually change
Microsoft Build 2026 reframed Windows as the runtime for autonomous agents. The interesting part is not the keynote line. It is where the constraints land.
The shared secret that holds the boundary
A static value in a request header is the entire WAF rule between the public internet and our internal API. We think about why that is the right call and what would change our minds.
Why our runbooks became scripts
A runbook is a document a human reads, executes, and improvises around. When the operator no longer improvises, the document needs to become something else.
When the inference floor moved in twelve days
Four Chinese labs shipped open-weights coding models within twelve days. The question is no longer whether they catch up. It is what the new floor changes.