I’m validating a product idea around AI agents and SaaS APIs. The thesis: As more users work through AI agents, SaaS products may need to make their APIs “agent-ready,” not just developer-ready. Today, most SaaS companies already expose REST APIs. But if a customer wants Claude, Cursor, or a custom agent to safely take actions inside that SaaS product, a normal API is not enough. You need: - agent-friendly tools - clean schemas - auth - scoped tokens - customer/user permissions...