Build Daily
Tinley Park · October 13, 2026Why "best model" is the wrong question now
Cognition President Russell Kaplan argues that as AI models reach saturation on common software tasks, the question of which model is 'best' is no longer the right one; speed and cost now outweigh raw capability for developers building agents. The clip is from Max Agency, a podcast hosted by LangChain CEO Harrison Chase that explores how real-world AI agent systems are designed, deployed, and debugged.
The outline
Every point the video makes, in the order it makes them.
- 01
Model saturation
Once every available model can handle a given software task well, raw capability ceases to be the differentiating factor.
- 02
Shift in developer priorities
With capability no longer the bottleneck, developers increasingly optimise for speed and cost rather than chasing the highest-scoring model.
- 03
Why 'best model' is the wrong question
Russell Kaplan reframes the common 'which model is best?' question as outdated, arguing the real decision criteria have flipped.
- 04
Max Agency podcast context
The clip comes from Max Agency, hosted by Harrison Chase (CEO of LangChain), which interviews builders designing and deploying real AI agent systems.
- 05
Agent architecture and tooling
The broader podcast covers architecture decisions, evaluation methods, tooling, and failure modes encountered when building useful agents in production.
- 06
Coding agents in the wild
The discussion is situated in the context of AI coding agents, where task saturation makes operational efficiency the primary concern.
When every model clears the bar on common engineering tasks, the optimization target shifts from capability to price per call. For a solo builder shipping agents, that means the architecture question stops being which model is smartest and becomes what is the cheapest path that still works. The moat was never the model selection; it was the system wrapping it.