Obsessing over linear model benchmarks is becoming obsolete, akin to comparing dial-up speeds. The real value and locus of competition is moving to the "agentic layer." Future performance will be measured by the ability to orchestrate tools, memory, and sub-agents to create complex outcomes, not just generate high-quality token responses.
Related Insights
When everyone can generate content with AI, the basic version becomes table stakes. The new competitive edge comes from creating advanced agent workflows, such as a "critic agent" that constantly evaluates and improves output against specific quality metrics.
Simply offering the latest model is no longer a competitive advantage. True value is created in the system built around the model—the system prompts, tools, and overall scaffolding. This 'harness' is what optimizes a model's performance for specific tasks and delivers a superior user experience.
The LLM itself only creates the opportunity for agentic behavior. The actual business value is unlocked when an agent is given runtime access to high-value data and tools, allowing it to perform actions and complete tasks. Without this runtime context, agents are merely sophisticated Q&A bots querying old data.
Just as standardized tests fail to capture a student's full potential, AI benchmarks often don't reflect real-world performance. The true value comes from the 'last mile' ingenuity of productization and workflow integration, not just raw model scores, which can be misleading.
While language models are becoming incrementally better at conversation, the next significant leap in AI is defined by multimodal understanding and the ability to perform tasks, such as navigating websites. This shift from conversational prowess to agentic action marks the new frontier for a true "step change" in AI capabilities.
An AI coding agent's performance is driven more by its "harness"—the system for prompting, tool access, and context management—than the underlying foundation model. This orchestration layer is where products create their unique value and where the most critical engineering work lies.
Traditional AI benchmarks are seen as increasingly incremental and less interesting. The new frontier for evaluating a model's true capability lies in applied, complex tasks that mimic real-world interaction, such as building in Minecraft (MC Bench) or managing a simulated business (VendingBench), which are more revealing of raw intelligence.
The most significant gains from AI will not come from automating existing human tasks. Instead, value is unlocked by allowing AI agents to develop entirely new, non-human processes to achieve goals. This requires a shift from process mapping to goal-oriented process invention.
Elias Torres argues that the current AI paradigm, which focuses on tools that assist humans (e.g., summarizers, drafters), is fundamentally limited. He believes true value is unlocked when you can instruct an AI to perform a task *infinitely* on its own, without requiring a human to type into a chat box for every action.
OpenAI's new GDP-val benchmark evaluates models on complex, real-world knowledge work tasks, not abstract IQ tests. This pivot signifies that the true measure of AI progress is now its ability to perform economically valuable human jobs, making performance metrics directly comparable to professional output.
