A rep at a home-services shop, a law-firm intake desk, or a dealership BDC lives on the phone. Their CRM sits in one tab. Their dialer in another. ChatGPT in a third. After every call, they retype what just happened. Half the time, they forget.
We rebuilt the stack so the call is a first-class object. Pipelines shape around calls. Contacts know which calls they came from. The CRM doesn't summarize the call after the fact — the call writes the CRM as it happens.
And we drew a line through it: humans close the sale, AI agents do everything else. Customers want to talk with real people. So real people own the conversation. AI sits next to them with live help, and AI agents handle the busy work that eats half a rep's day — before, during, and after.
Underneath, every action — placing a call, drafting a follow-up, moving a stage — is an MCP tool. Reps run the product through the UI. Power users run it through Claude Code. Our own AI features are just MCP clients. This is what AI-Agent-Native means: built for agents to do real work, not bolt-on demos.