📊 Full opportunity report: The Model Is Only 10%: The Real Lesson of the New SDLC on ThorstenMeyerAI.com — validation score, market gap, and execution plan.
TL;DR
A recent Google whitepaper argues that in AI-assisted development, the AI model itself is only 10% of the system’s behavior. The focus should be on the harness and context engineering, which constitute 90%, impacting reliability and costs significantly.
A new Google whitepaper, “The New SDLC With Vibe Coding,” emphasizes that the AI model accounts for only about 10% of system behavior in AI-assisted development. Instead, the harness and context engineering are where most of the control and value lie, fundamentally shifting how organizations should approach AI integration.
The whitepaper, authored by Addy Osmani, Shubham Saboo, and Sokratis Kartakis, reports that 85% of professional developers now use AI coding agents regularly, with 51% using them daily. It states that roughly 41% of all new code is generated by AI. The key insight is that the AI model itself is only 10% of what determines system behavior; the remaining 90% depends on the harness — the prompts, tools, rules, and observability layers around the model.
This perspective challenges the common focus on upgrading models, suggesting instead that organizations invest more in configuration, tooling, and context management. Evidence from experiments with coding agents shows that changing the harness can significantly improve performance, even with the same underlying model.
The model is only 10%
A Google whitepaper argues software’s biggest shift is from writing code to expressing intent. Its sharpest claim: the model you obsess over is the smallest part of the system — the scaffolding around it does the real work.
The clearest map yet of how serious AI development works — and mostly tool-agnostic. But it’s a Google funnel: the concepts are neutral, the on-ramps point to Gemini, Jules & the ADK. If the harness is 90% and it’s yours, your moat and your costs both live there — so own your scaffolding, route across models, and remember: AI amplifies whatever engineering culture it lands in.
Implications for AI Development Strategies
This shift means that organizations should prioritize building robust harnesses and context management rather than solely focusing on acquiring the latest AI models. It impacts cost management, reliability, and security, as the majority of failures and inefficiencies stem from configuration errors. The insight encourages a reallocation of resources towards system design and tooling, which can lead to cost savings and better control.
AI development tooling kit
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Background on AI-Assisted Development and Evolving Best Practices
As of early 2026, AI coding agents have become mainstream, with a significant portion of new code generated by AI. Previously, emphasis was placed on adopting more advanced models. However, the whitepaper underscores a growing recognition that model improvements alone are insufficient. The real challenge lies in how the models are integrated and controlled. Past approaches often overlooked the importance of harness design, context provisioning, and verification, which now emerge as critical components of effective AI development.
“The model is only 10% of what determines behavior; the rest is in the harness.”
— Addy Osmani
The AI Prompt Playbook: Master AI Prompt Engineering with 140 Ready-to-Use Templates for ChatGPT, Claude, Gemini & Copilot
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Unresolved Questions About Implementation and Cost
While the whitepaper presents compelling evidence that harness and context dominate system behavior, it does not specify precise best practices for building and maintaining these components. The optimal strategies for cost-effective harness design and context management remain under development, and organizations may face challenges in scaling and standardizing these practices across teams.
AI Engineering: Building Applications with Foundation Models
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Next Steps for Organizations Adopting the New SDLC
Organizations should evaluate their current AI development workflows, emphasizing harness and context engineering. Investment in training, tooling, and process refinement is likely to yield better reliability and cost savings. Industry leaders may start publishing best practices and standards for harness design. Further research will clarify how to balance upfront engineering costs with long-term efficiency gains.
AI model harness configuration software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
Why is the model only 10% of system behavior?
The whitepaper explains that most of the system’s behavior depends on how the AI is integrated, configured, and controlled through prompts, tools, and rules, which constitute 90% of the effective system.
How does this shift affect AI development costs?
Focusing on harness and context management may involve higher upfront engineering costs but can lead to lower ongoing operational costs and better reliability, as configuration errors are a major source of failures.
What does this mean for AI model upgrades?
The whitepaper suggests that upgrading models alone is insufficient; organizations should invest more in system design and tooling, which has a larger impact on performance and cost.
Is this approach applicable to all AI systems?
While the principles are broadly applicable, the specific strategies for harness and context engineering may vary depending on the system’s complexity and use case.
What are the main challenges in implementing this shift?
Challenges include developing scalable best practices for harness design, managing dynamic contexts, and balancing upfront engineering efforts with long-term benefits.
Source: ThorstenMeyerAI.com
