Built for operators who need to know what's actually changing.

Bitter Frontier reads projects across the coding-agent stack together: what general agents just made cheaper, what became the new bottleneck, and where operators should spend scarce human attention.

Two named ideas guide that read. Bitter Lesson Maxing means building where improving general agents compound your advantage, not where their next release erases it. Amdahl Maxing means treating human attention as the scarce serial resource and spending it where judgment changes the outcome.

This is not a newsletter summarizing blog posts. Claims trace back to primary sources: commits, releases, versioned documentation, security advisories, or described reproducible observations. A signal should change what you build, delegate, configure, test, measure, keep human, or stop relying on. The evidence chain is public and auditable.

Software helps gather and check the source record. Michael Ruescher sets the editorial doctrine, verifies the reading, and decides what deserves publication. Automation expands coverage; editorial judgment stays accountable. The research trail preserves what was read, what was found, and what the review caught.

Bitter Frontier is published by Bitter, which builds a harness-agnostic operating environment for coding agents. The publication has its own method: claims need sources, and signals need operator consequences.

Founder and editor Michael Ruescher previously built a long-horizon agent at Blueshift. That work led to the layer decision behind Bitter: the company is betting that provider harnesses should remain replaceable while the operating environment and accumulated company state stay durable. Frontier is the upstream intelligence function that follows from that decision. Read the pinned founder's letter.