Enterprise AI and software factories
Diffusion
Diffusion builds software factories for large enterprises: systems that turn business intent into software and operational workflows. Its Factory Building service works with the people who own a workflow to develop patterns, evaluations and delivery paths, then trains operators to extend the system themselves. The aim is to address the work surrounding code generation—deciding what to build, proving it works, securing approval and running it. Diffusion Agent delegates implementation to frontier coding systems from Anthropic, OpenAI and Google; Diffusion Cloud provides a hosted runtime for agentic applications.
Justin McCarthy founded Diffusion after co-founding StrongDM and serving as its CTO, where he spent a decade building enterprise cybersecurity infrastructure. That background informs Diffusion’s emphasis on security and operational control alongside agent automation. Its factory approach includes critic and validator applications that judge outputs, pipelines that return to a known-good state, and digital twins that let teams rehearse failures unsafe to trigger in live operations.
Diffusion also researches how AI systems represent the knowledge needed to build software. Its September 2026 Specification Oracles research, by Atticus Cull and Justin McCarthy, compared models consulting external notes with models storing knowledge in trained adapter weights. In controlled, generated worlds, the adapters benefited more from structured relationships, while external notes required substantially less storage per world. The work examines a foundational factory problem: how a system can retain and answer questions about what software is supposed to do.
0 talks
Newest firstLoading more talks and speakers…
1 speaker at AIE
Affiliations reflect their AIE event listings and recorded appearances, not necessarily current employment.