The industries are wider than we can reach alone.

The part nobody tells you about AI agents

An agent is only as good as the knowledge behind it. That is not a disclaimer; it is the whole business.

Most companies' knowledge does not live in documents. It lives in the head of whoever has been there longest, in three years of chat history, and in a folder nobody has opened since 2019. Point an agent at that as-is and you get something fluent and confidently wrong — the failure mode everyone in this field recognises and nobody advertises.

Getting from there to an agent that a business can put in front of its customers takes someone who sits down with them and works through it: what the agent must know, what it must never say, which questions it should ask before it answers, and where a human takes over. That work is specific, it is unglamorous, and it cannot be done remotely at scale by a platform vendor who does not know the industry.

Which is exactly the opening

There are more industries that need this than we will ever reach, and each one has its own vocabulary, its own compliance line, and its own idea of what a good first conversation sounds like. We are not going to learn all of them. We would rather build the layer underneath and stay neutral across verticals.

So if you have spent years in an industry — if you know how deals actually close in it and you have the relationships to prove it — there is a business here that we are not in a position to do and you are.

What the work is

This is what an implementation engagement typically covers. It is worth reading closely, because this is where the margin is:

• Knowledge base construction. Auditing what the customer has, deciding what belongs in the agent and what doesn't, structuring it so retrieval actually works. The single highest-leverage step, and the one customers are least equipped to do themselves.
• Playbook design. Turning how the business handles an enquiry into something the agent runs consistently — qualification questions, the boundaries, the hand-off points.
• Integration. Channels, CRM, calendars, internal systems over MCP and APIs.
• Tuning and go-live. Testing against real enquiries, finding where it goes wrong, fixing it before customers see it.
• Training and handover. Getting the customer's own team able to run and extend it.

You deliver that work in your industry, under your own commercial terms.

Where the line sits

We keep the technical foundation; you keep the business fit. We are deliberately neutral across industries — we are not going to show up in your vertical competing for the same engagements.

Whatever we agree, it will be clear to the customer who owns the outcome — three-way finger-pointing is the way this model fails, and we would rather not build it that way.

Your own brand, if you want it

The platform is not open source, but every capability is reachable over the API — agents, knowledge bases, conversations, follow-ups, usage. Partners who want to run this as their own product can: your interface, your name, your customer relationship, with us in the background.

See White-label and API for what that actually involves technically.

Being straight about the stage

This is early. There is no tier table on this page, no certification programme, no partner portal — because inventing all of that before talking to the first few partners would just mean inventing the wrong thing. Commercial terms are deliberately open: industries differ enough in delivery effort that publishing one set of numbers now would be a guess.

This is not a sign-up form for resellers. It is a conversation. We would rather work with a small number of people who genuinely know their field than collect a directory of logos.