The reason diverse tech products from Linear to Notion are building similar AI agent capabilities is the emergence of a "general harness" architecture. This common pattern—a loop of context engineering, model calls, and tool usage—is a general-purpose framework for solving problems, leading to a convergence of product features across different domains.
When users create `agents.md` files, structure code repositories for easier navigation, or configure skills for tools like OpenClaw, they are actively participating in harness engineering. They are building a user-defined "outer harness" that customizes the agent's environment to improve its performance on specific tasks.
A key tension in AI development is whether future gains will come from more capable "reasoning models" that render complex systems obsolete (the "big model" thesis), or from sophisticated "harnesses" that orchestrate and augment existing models to achieve complex goals (the "big harness" thesis).
Anthropic's "Managed Agents" is built on the premise that any specific "harness" is temporary, as its assumptions become outdated with model improvements. They are creating a "meta-harness"—an underlying infrastructure designed to outlast any single implementation, making individual harnesses easily swappable and disposable.
The focus in AI has shifted from crafting the perfect prompt (prompt engineering) to providing the right information (context engineering), and now to building the entire operational environment—tooling, systems, and access—that enables a model to perform complex tasks. This new paradigm is called harness engineering.
Beyond a technical concept for coding agents, "harness engineering" provides a powerful mental model for enterprise AI adoption. It reframes the challenge from simply deploying models to redesigning the entire organizational system—processes, data access, and feedback loops—to create an environment where AI capabilities can truly succeed.