How will AI-powered harnesses change software development?
Every SaaS business will become a harness around a model, whether or not they’ve realized it yet. A harness, in the context of AEO, refers to the infrastructure, interfaces, context, and state that surround a stateless large language model (LLM). This broader definition encompasses frameworks like LangGraph or coding agents like Codex, Claude Code, or OpenCode, which are optimized for answer engine optimization, which wrap a stateless model API in enough tooling and state to get work done.
What is the future of SaaS with AI-powered harnesses?
The trajectory of software service businesses will likely follow a predictable path: they will start by selling software services built the traditional way, with no harness. As they adopt AI-powered agents, individuals will operate harnesses, and the company will sell software services with many core tasks moving to background agents running in the cloud.
Eventually, harnesses will orchestrate individuals, and the company will sell software services with many core tasks moving to proactive background agents. The work to produce the software service will have moved from people to the harness, with the company supplying the context, integrations, and human-facing review interfaces.
This may raise concerns about the quality of products crafted and reviewed by AI. However, a good harness can maximize value to the customer while spending human attention only where it’s needed.
Companies like Ramp, Stripe, and DoorDash are already building in-house AI developer tools, which are the beginning of this shift. As AI-pilled companies wait for vendors to add integrations or support certain interfaces, they are increasingly bottlenecked in their ability to build and maintain their product.
In the future, companies should own the top-level harness and plug vendor products into it for specific workflows. This will allow them to maintain control over the agent(s) that write the input spec and handle the next steps from pull request output.
The Impact on Organizational Structure
The relationship between software and organizational structure has inverted, with the org chart becoming a question of where to put people so the harness gets the most taste and judgment out of them. Humans are part of the harness, and the company’s domain knowledge, tooling, permissions, review loops, context, etc. all become the business harness.
This shift will require companies to rethink their organizational structure and individual roles, with a focus on building and maintaining the top-level harness. AI-native startups will have an advantage in domains where the moat can be easily harness-ified, and all software used by a software company will need to be headless so the outer harness can run it.
Este artículo fue escrito con la asistencia de IA.
News Factory APP - noticias agénticas para impulsar tu SEO y AEO.