Every SaaS business will become a harness around a model, whether or not they’ve realized it yet. A harness includes all the infra, interfaces, context, and state that surround a stateless LLM. A “software factory” is a harness whose parts are smaller harnesses—one that writes specs, code, reviews—plus something on top deciding what runs when.
Following this trajectory, companies move from traditional SaaS to engineers pairing with agents, then to background agents in the cloud, and finally to proactive agents where harnesses orchestrate individuals. The company becomes a harness, with core tasks done by agents and the “product” entirely model output. The org chart inverts as humans become part of the harness, supplying taste and judgment.
It’s tempting to think AI-crafted products will be low quality, but the mitigation is having the harness pick where human inputs matter most. For example, an agent fanning out questions to reps and synthesizing a demo for a product lead, or a feature suggestion turned into an architectural decision for an engineering taste-holder. A good harness maximizes customer value while spending human attention only where needed.
Skepticism is fair—trusting agents with the outer loop is difficult today. But it’s not worth betting models won’t eventually do this, especially as planning-and-review work gets broken into tasks with verifiable rewards. See The Transposed Organization for more on “taste-holders.”
Source: Hacker News · Summarized by HeadlinesBriefing