The inconsistency and 'laziness' of base LLMs is a major hurdle. The best application-layer companies differentiate themselves not by just wrapping a model, but by building a complex harness that ensures the right amount of intelligence is reliably applied to a specific user task, creating a defensible product.

While AI models get the headlines, they are becoming commodities. The true competitive advantage lies in building a custom "harness"—the surrounding application, data integrations, and specialized tools that direct the model's power to solve a specific user problem effectively.

The competitive advantage for vertical AI isn't just data, but creating increasingly difficult, proprietary evaluation benchmarks. By creating and continuously improving performance against a moving target for specific tasks, vertical AI companies build a durable product advantage that general models cannot easily replicate.

Startups like Cognition Labs find their edge not by competing on pre-training large models, but by mastering post-training. They build specialized reinforcement learning environments that teach models specific, real-world workflows (e.g., using Datadog for debugging), creating a defensible niche that larger players overlook.

Early-stage AI startups should resist spending heavily on fine-tuning foundational models. With base models improving so rapidly, the defensible value lies in building the application layer, workflow integrations, and enterprise-grade software that makes the AI useful, allowing the startup to ride the wave of general model improvement.

Large AI labs focus on solving problems in the most generalizable way, which can be an 'intellectually lazy' approach for specific enterprise needs. Startups can win by building the necessary last-mile components—harnesses, orchestration, and software—that labs are not structured or incentivized to create.

Relying solely on expensive frontier models is unsustainable. Vertical AI companies must build a portfolio of smaller, specialized models that match frontier performance on specific tasks but cost 100x less, effectively allocating intelligence where it's needed most.

While the "bitter lesson" suggests powerful general models will dominate, vertical AI solutions can thrive by deeply integrating with a company's specific data, workflows, and project context. The model can't know this proprietary information; value is created by the application that bridges this gap.

As base model capabilities converge, the key differentiator is shifting to the "agent harness"—the infrastructure, tools, and skills built around the model. For vertical AI, this is where domain expertise is injected, creating specialized agents with custom tools that outperform generalist models.