Quoting Jon Udell
Jon Udell challenges the framing of 'human in the loop' as a concession to machine authority, arguing instead that AI agents should be recruited into existing human workflows rather than replacing human judgment. His critique centers on agentic software development that remains transparent and reviewable, rejecting black-box systems that obscure decision-making. This reframes a core tension in AI adoption: whether automation subordinates human oversight or augments it within human-controlled processes.
Modelwire context
ExplainerThe sharper point here is linguistic: Udell is arguing that the phrase 'human in the loop' already concedes too much, implying the machine is running the loop and humans are guests in it. Flipping that framing, so humans own the workflow and agents are recruited into it, has real consequences for how developers design review interfaces and audit trails.
Modelwire has no prior coverage to anchor this to directly, so it sits somewhat on its own. It belongs to a broader ongoing conversation about agentic AI design philosophy, one that has been surfacing repeatedly in practitioner writing over the past year as coding agents and autonomous task runners have moved from demos to daily use. Willison has been a consistent voice in that space, and his choice to amplify Udell suggests this framing is gaining traction among developers who are actually building with these tools rather than just writing about them.
Watch whether Willison or Udell follows up with concrete tooling or workflow examples that operationalize the 'recruit into human process' framing. If that happens within the next few months, it signals the idea is hardening into practice rather than staying at the level of critique.
This analysis is generated by Modelwire’s editorial layer from our archive and the summary above. It is not a substitute for the original reporting. How we write it.
MentionsJon Udell · Simon Willison
Modelwire Editorial
This synthesis and analysis was prepared by the Modelwire editorial team. We use advanced language models to read, ground, and connect the day’s most significant AI developments, providing original strategic context that helps practitioners and leaders stay ahead of the frontier.
Modelwire summarizes, we don’t republish. The full content lives on simonwillison.net. If you’re a publisher and want a different summarization policy for your work, see our takedown page.