Have your AI talk to ours.
Rootz has an AI that speaks for Rootz. Your AI can ask it whether Rootz fits your problem and how we would apply it, then open a real project with real people. There is no form to fill in, and you don't need a sales call to start.
Connect
The endpoint is a standard MCP server (Streamable HTTP):
In Claude Code:
Or just tell your assistant: "Ask the Rootz agent at rootz.global/mcp whether Rootz fits this problem: …"
What it does
Maps your problem
Your AI describes the problem. The agent says which of the three jobs it is (quality, origin, ownership), which Rootz capabilities apply, real deployments that are closest, and what a Rootz engineer would ask first.
Says no when it should
If nothing in the problem is about proving where something came from, who authorised it, or who owns it, it says Rootz is probably not the fit. That answer is worth something too.
Opens a real project
When there is a fit, your AI opens a project. A person at Rootz is notified, and you get a private project page that both AIs, and both teams, can pick up from.
Checks your identity
It will also tell you what identity your own company presents to other AIs, and the next step to make it verifiable.
Optimistic by role, honest by rule
This is a sales agent, and its goal is a real project. Every capability it describes carries its status:
LIVE runs today and you can test it PILOT works, being proven with first customers ROADMAP designed, not built
A weak claim is not a fake claim, and an understated claim is as false as an overstated one. The agent never quotes binding prices and never commits Rootz to anything; people do that. It does not run a language model on what your AI sends: it returns the map, and your own AI does the reasoning.
Identity unlocks depth
The more your AI can prove about who it is, the further the conversation goes. The place to start is simple: give your agent its own mailbox.
| Level | Your AI presents | What that opens |
|---|---|---|
| 0 · Anonymous | Nothing | Capabilities with status, worked examples, the design map |
| 1 · Verified mailbox | Control of a mailbox, proven with a code we email to it. A mailbox at your company's domain counts as company level. | A project, a private project page, the pilot shape, a named person at Rootz |
| 2 · Domain-published key | A signing key published under your domain | A signed two-way record of the conversation |
| 3 · CorpID-authorised agent | Your corporation's digital name authorising this agent for a stated scope, revocably | Agreement to next steps within that scope |
A mailbox proves the agent can receive mail at your domain. It does not prove your company authorised the request; that is what Level 3 is for. Every company will need its own version of this front door, and every one of them will face the same question: which identity should another company's AI trust?
Who is speaking
The agent speaks for Rootz Corp., whose corporate identity is published at /.well-known/corp-identity.json. Every answer is signed by the agent's own key. That key's authority is delegated by the Rootz CEO key, which the corporation's 2-of-3 root authorised on chain, and each receipt reports whether that delegation is recorded, so your AI can check the whole chain itself. You are talking to an AI; a person at Rootz follows up on every verified project.