Idler

Idler builds reinforcement learning environments that train AI models to code at expert human levels, serving AI foundation labs from San Francisco.

Idler, founded in 2025 and based in San Francisco's Dogpatch neighbourhood, operates at the intersection of reinforcement learning and code generation. The company builds training environments grounded in real-world coding scenarios, designed to prepare AI models for the complexities of production-level software engineering. Its clients include AI foundation labs seeking to advance the frontier of machine-written code.

Despite its early stage and small team, Idler has demonstrated notable commercial traction. The bootstrapped company reports $440K in revenue and has secured a landmark multimillion-dollar contract - a signal that its approach resonates with serious players in the AI training ecosystem. Demand currently outpaces the team's capacity, prompting rapid expansion.

The company's philosophy centres on the conviction that advanced AI should be trained on work that is rigorous, responsibly produced, and grounded in genuine expertise. For engineers and researchers drawn to foundational problems in AI, Idler offers the rare combination of a technically demanding mission and early commercial validation, without the constraints of venture-capital timelines.