Get ready: our MCP server is coming. It should be ready next week, and I’ll be sharing a lot more about what it can actually do. But int his long post, I want to share some thoughts on MCP and where everything is getting. We’ve been getting more and more requests for MCP support, and it’s kind of crazy to think about how quickly this shift has happened. A year ago, the idea that an AI agent could just go into your software and do the work for you, without you ever logging into the UI still felt pretty futuristic. Now, I think it’s one of the biggest changes coming to SaaS. Software companies are already seeing it affect things like per-seat pricing. If an agent is doing the work, does every human still need a seat? When people first started asking me for a LeadCapture MCP server, my reaction was basically: Why? We’re already building AI directly into LeadCapture. Our Copilot can create things, make changes, analyze funnels, etc. Sometimes it screws up, but its' getting better. But then it clicked. Forcing everyone to use our AI inside our UI is actually counter to where things are going. I think we’re going to end up with two groups of people: People who want to log into software, click around, and have a UI. And people who want to tell Claude, Codex, or whatever agent they use what they want done, and have the agent go do it. And maybe even people in between, like me.. .who like options. We need to support both. So we started looking into what it would take to expose LeadCapture through MCP. And there was a pretty big silver lining from all the work we’ve already done on Copilot: We were probably 90% of the way there without realizing it. The hard part wasn’t really “building an MCP server.” Forms and funnels look simple from the outside, but users make an unbelievable number of tiny customizations: visibility rules, integrations, validation, tracking, routing, styling, field behavior, and hundreds of other little things we’ve added over the years. Then there's security, uptime, etc.