Due diligence, answered.
The questions platform teams ask before an integration call.
What exactly are we buying?
A white-label agent engine integrated into your practice-management backend. Every practice on your platform can then take bookings through their patients' AI assistants (MCP + A2A), a phone line and WhatsApp — one engine, one integration, every channel, under your brand and on your invoice.
What are MCP and A2A?
The open protocols AI assistants use to talk to services: MCP powers ChatGPT/Claude-style connectors, A2A is the agent-to-agent protocol. A patient asks their assistant to book; the assistant converses with the engine; the patient confirms an SMS code; the appointment is written into your system with a full transcript. No app, no login — the patient adds the connector once and keeps it.
Do you sell to practices directly?
No — never. We work exclusively through platform partners. Your practices stay your customers: your brand, your invoice, your relationship. That's a structural commitment, not a courtesy.
Our software has no public API — is this still possible?
Yes. Our first production integration ran against a closed system with no official interface at all — reverse-engineered, with genuine read and write calendar access. With your cooperation the path is much shorter: a calendar layer, an internal API or a documented data model gets us to a plan measured in weeks.
How is patient identity verified before anything is booked?
Nothing is booked or disclosed until the patient confirms a one-time SMS code. On top of that, each practice controls which practitioners and appointment types are agent-bookable, and can run approve-first mode where a human confirms every agent booking before it lands.
Who is the data controller? Where does data go?
The practice stays the controller, you're the processor, we come in as a sub-processor — so you need a DPA with us and a sub-processor notice to your customers, not a new legal framework. Data is encrypted in transit and at rest with configurable retention. Some sub-processors (for example the AI models) sit outside the EU and operate under EU Standard Contractual Clauses — we say that plainly and walk your DPO through the full picture. DPA draft, sub-processor list and deletion concept come from us.
Are calls recorded?
No — by design. Conversations are transcribed and classified, never stored as audio recordings. Every calendar action lands in a write-ledger the practice can inspect.
Who handles support?
The proven split: your team takes first line (setup, customer questions), we take everything technical behind it — monitoring, model behaviour, integration maintenance. Your customers keep one point of contact: you.
Do you require exclusivity?
No exclusivity without consideration. Terms — model, support split, pilot scope, success criteria — go in writing before anything is built.
What does the integration cost, and how long does it take?
The integration is built and maintained at our cost; typical build time is 2–3 weeks with your cooperation. Commercials for the module itself are rev-share or per-practice wholesale — worked out on the call, because a 40-practice vendor needs different terms from a 4,000-practice one.
How do we know the engine actually works?
Call it. The reference deployment at a Hamburg physio practice has answered 224 calls at 100% with 0.8s pickup, writing into a closed calendar system — and on the integration call we hand you the live agent connector so you can book from your own ChatGPT or Claude while we watch.
Put your platform on the agent web.
One integration call: your backend, the channels, the commercials — and a live booking you can watch.