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.
Related Insights
When evaluating AI startups, don't just consider the current product landscape. Instead, visualize the future state of giants like OpenAI as multi-trillion dollar companies. Their "sphere of influence" will be vast. The best opportunities are "second-order" companies operating in niches these giants are unlikely to touch.
Unlike cloud or mobile, which incumbents initially ignored, AI adoption is consensus. Startups can't rely on incumbents being slow. The new 'white space' for disruption exists in niche markets large companies still deem too small to enter.
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 notion of building a business as a 'thin wrapper' around a foundational model like GPT is flawed. Truly defensible AI products, like Cursor, build numerous specific, fine-tuned models to deeply understand a user's domain. This creates a data and performance moat that a generic model cannot easily replicate, much like Salesforce was more than just a 'thin wrapper' on a database.
Incumbent companies are slowed by the need to retrofit AI into existing processes and tribal knowledge. AI-native startups, however, can build their entire operational model around agent-based, prompt-driven workflows from day one, creating a structural advantage that is difficult for larger companies to copy.
AI labs like Anthropic find that mid-tier models can be trained with reinforcement learning to outperform their largest, most expensive models in just a few months, accelerating the pace of capability improvements.
Small firms can outmaneuver large corporations in the AI era by embracing rapid, low-cost experimentation. While enterprises spend millions on specialized PhDs for single use cases, agile companies constantly test new models, learn from failures, and deploy what works to dominate their market.
As reinforcement learning (RL) techniques mature, the core challenge shifts from the algorithm to the problem definition. The competitive moat for AI companies will be their ability to create high-fidelity environments and benchmarks that accurately represent complex, real-world tasks, effectively teaching the AI what matters.
Perplexity's CEO argues that building foundational models is not necessary for success. By focusing on the end-to-end consumer experience and leveraging increasingly commoditized models, startups can build a highly valuable business without needing billions in funding for model training.
Investing in startups directly adjacent to OpenAI is risky, as they will inevitably build those features. A smarter strategy is backing "second-order effect" companies applying AI to niche, unsexy industries that are outside the core focus of top AI researchers.