Time to be B2A

One of the best insights I’ve had recently, after talking with friends and listening to AI podcasts, is extremely simple, but I still think many companies are missing it: Your company may not need to offer an AI agent to your customers. Your customer might be someone’s AI agent.

OpenClaw was a great proof of concept for this idea. People want their own agents, customized to their needs, connected to services, MCPs, and APIs that give them enough autonomy to actually get things done. In that world, the user chooses the model they want to use, how much they want to spend, and the best way to interact with it. The agent knows them. It understands the context of their work. It can combine tools, services, and workflows across different products. And because of that, it can often serve them better than an AI experience locked inside a single centralized product.

This changes how we should think about product strategy. Your product is no longer designed only for human users. It also needs to be easy for AI agents to understand, access, and use. That means extremely clear documentation, great connectors, accessible MCPs or APIs, and strong SEO may become a real competitive advantage.

A simple example: try building an app with Claude. There is a good chance it will suggest tools like Supabase and Vercel. That is not random. Those products are easy to explain, easy to integrate, and easy for AI models to reason about.

So maybe the better question is not: “How do we add AI to our product?” Maybe the better question is: “How do we make our product ideal for AI agents to consume?”