ClientPilot AIDemo build · fictional business, simulated dataA form fill becomes a booked appointment. Nobody touched it.
AI intake that answers from the business’s own documents, follows up by SMS, and books the calendar — with a visible audit trail for every step.
Live demos are password-gated — request access via WhatsApp, takes a minute.
Reply cites the clinic’s own policy PDF10 automated steps, timestampedThe Google-Calendar event it createdThe challenge
The office manager’s day is a queue: read the inquiry, look up the policy, write the reply, chase the booking. Multiply by fifty a month and intake becomes a part-time job nobody was hired for.
Automation without evidence is a hard sell — the owner has to be able to see what the machine said and where the answer came from before they trust it with clients.
The solution — three decisions
Citations as a first-class UI element
Every AI reply carries a small source tag naming the document and page; clicking opens the exact passage. Grounding is visible, not claimed.
The automation thread
Webhook fired → reply drafted → SMS sent → booking created — rendered as a stitched vertical trail with timestamps. It is the “nobody touched this” proof, replayable on camera.
Escalation as a feature
Clinical and claims questions are deliberately held for staff with a visible reason — the assistant knowing what not to answer is part of the pitch.
How it works
Flow diagram: the inputs listed first feed into the ClientPilot engine engine in the middle, which produces the outputs listed last.
In
ClientPilot engine
Out
The demo implements this shape end to end with a simulated service layer — the “extend for production” section below lists what swaps in for live deployment.
Product tour
What the demo shows
- The same inquiry visible on both surfaces — client phone and staff CRM
- An SMS thread rendered natively, typing indicators and all
- Inbox rows whose reference IDs descend with recency — the details technical buyers check
- A knowledge base with indexing states powering every answer
Under the hood
- Inquiry templates with per-template message variants keep 50 mock inquiries free of duplicate rows
- Escalation logic models a real routing decision: clinical/claims topics never get auto-replies
- A single automation-event model renders both the per-inquiry thread and the 30-day activity feed
- Funnel math is computed from the same store the tables read — every KPI reconciles
Built as a demonstration — on purpose.
Demo Physio Clinic is fictional and labeled as a demo on every screen. Nothing here is presented as client work: no client names, no outcome metrics, no testimonials. The proof is the running product — open the live demo above (password on request) and check every claim.
What I’d extend for production
- Real RAG over uploaded documents with chunk-level citation offsets
- Twilio SMS + Google Calendar integrations behind the same event model
- Feedback loop: staff edits to drafts fine-tune reply style