Developers working with large language models (LLMs) are experiencing an overwhelming number of automated pull requests (PRs) and issues, according to Jeremy Theocharis, co-founder and CTO at UMH. Speaking at the Local-First Conference in Berlin this week, Theocharis highlighted that even creators of LLM-based tools face this challenge, underscoring the scale of automated contributions generated by LLMs themselves.
During a session featuring Armin Ronacher, founder of Earendil and creator of Pi.dev—an open-source coding agent platform—Ronacher revealed that his team automatically closes nearly all PRs and issues submitted by LLMs. Despite this, he encouraged continued submissions, emphasizing that human input remains essential and that genuine contributions will stand out amid the automated noise, as reported by theocharis.dev.
This phenomenon reflects a broader tension in the AI development community, where the productivity gains from LLMs are tempered by the need to manage the volume of machine-generated code changes. Ronacher, known for his work on Flask and Sentry, represents a cohort of experienced engineers grappling with balancing automation and human oversight. The situation illustrates challenges in maintaining open-source projects amid rapid AI-driven activity.
The Local-First Conference, held in Berlin in July 2026, brought these issues to light, with key figures like Ronacher and Theocharis discussing the implications of LLM-generated contributions. Ronacher’s approach of auto-closing most PRs while valuing human input offers a practical model for managing AI-driven development workflows.