Addy Osmani says the new software factory is really loops, harnesses and review gates
addyosmani · x · 2026-07-21
Software factories are loops at scale
Addy Osmani argues that the new agent era makes the old “software factory” dream worth revisiting, but with a sharper distinction between two modes:
- Light factory: humans stay in the loop, trading some speed and focus for judgment and fewer breakages.
- Dark factory: agents scope, build, and ship with little or no human reading, which risks people losing understanding of the codebase.
He breaks the system into three layers:
- Loop: one agent repeatedly gathers context, acts, verifies, and repeats until done.
- Harness: the sandbox, tools, memory, and completion gates around that loop.
- Factory: many harnessed loops running in parallel, fed by a queue and drained through review into production.
The core thesis is that the real job is no longer just writing code, but designing the checks, review gates, and autonomy boundaries that make agent-built software understandable and safe.
More from coding & agent
- Dev builds interactive 3D product experience with GPT-6 Astra + Hyper3D Rodin — nikola_mr64990 · 2026-09-11
- Codex tip: use Sol with Astra and Luna sub-agents to save usage — pvncher · 2026-09-11
- agents-best-practices: a provider-neutral Agent Skill for designing and auditing agentic harnesses — tom_doerr · 2026-09-11
- Cognition's SWE-2 uses a KKT duality argument in RL to shift the effort Pareto curve — YouJiacheng · 2026-09-11
- First-ever Three.js Conference lands in Paris, with a panel on AI-shortened design workflows — OdinLovis · 2026-09-11
- Data engineering, not agent frameworks, is the real bottleneck for enterprise AI agents — dhruv2038 · 2026-09-11