Why Cognition bought Poke: AI personality is becoming a competitive advantage

Cognition just acquired Poke, and the reason tells you everything about where AI product competition is heading. According to the announcement, they want Poke's conversational style and interaction patterns inside Devin, their AI coding agent. This isn't about better models or faster inference. It's about personality.

The underlying bet is simple but significant. As foundation models commoditize and everyone gets access to similar capabilities through APIs, the way an AI assistant talks to you becomes the moat. Poke built a distinct interaction model that users clearly responded to, and Cognition decided that was worth acquiring rather than building from scratch.

This mirrors what we've seen in consumer AI products. Character.AI built an entire business on personality. Inflection's Pi differentiated purely on tone. Even OpenAI's ChatGPT voice mode updates focus heavily on conversational flow and emotional resonance, not just accuracy. The pattern is consistent. Technical capability gets you to the table, but interaction design keeps users coming back.

For developer tools like Devin, personality might seem like a nice-to-have. But if you're spending hours pair programming with an AI agent, the difference between one that feels like a helpful colleague and one that feels like a stack trace parser matters more than you'd think. Cognition clearly believes that advantage is real enough to acquire a company for it.

The acquisition also reveals how quickly AI product strategy is maturing. A year ago, everyone was focused on evals and benchmarks. Now serious companies are buying interaction design and conversational architecture. That's a market signal that the purely technical phase of AI competition is giving way to product and experience differentiation.

What this means for you: if you're building with AI or choosing tools, start paying attention to how they communicate, not just what they can do. The agents you'll actually use long-term will be the ones that match your working style. When evaluating a new AI tool, try this prompt: "Explain your interaction model. How do you decide when to ask clarifying questions versus making assumptions? How do you handle uncertainty?" The quality of that answer will tell you how much thought went into the experience beyond the model.