Instead of interacting with a single LLM, users will increasingly call an API that represents a "system as a model." Behind the scenes, this triggers a complex orchestration of multiple specialized models, sub-agents, and tools to complete a task, while maintaining a simple user experience.

Performance gains increasingly come from the "harness"—the surrounding system of tools, data connections, and agentic workflows—not the underlying model. Stanford's "meta-harness" concept shows a 6x performance gap on the same model, suggesting the product layer is where real innovation and competitive advantage now lie.

An AI model's operating environment—its "harness"—is now the primary driver of capability. Benchmarks show the same model achieves vastly different results in different harnesses, proving that the runtime, tools, and state management are as critical as the model's internal weights for achieving results.

The success of tools like Anthropic's Claude Code demonstrates that well-designed harnesses are what transform a powerful AI model from a simple chatbot into a genuinely useful digital assistant. The scaffolding provides the necessary context and structure for the model to perform complex tasks effectively.

Google's strategy involves the core AI model progressively absorbing the surrounding tooling and infrastructure (the "scaffolding"). This creates a standardized, extensible "harness" that accelerates development and ensures a consistent, high-quality agentic experience across Google's vast and diverse product landscape, from Search to consumer apps.

Building on AI involves a "tick-tock" cycle. First, engineers create a complex "harness" of prompts and skills. Then, a new, more powerful base model is released that performs those skills natively, "eating the harness" and forcing engineers to simplify and build a new layer of more advanced heuristics.

An AI model alone is like a brain without a body. To become a useful agent, it needs a "harness" or "scaffolding" consisting of four key components: domain-specific knowledge, memory of past interactions, tools to take actions, and guardrails for safety.

Raw AI models are not useful on their own. A critical new software layer, dubbed a 'harness,' has emerged to make them effective. These harnesses (like OpenClaw or Codex) provide the structure for models to think in patterns and accomplish complex tasks, acting like an operating system for AI.

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).