We built this for Seer Medical.
Not a slide deck. A working AI agent that answers study-status, device-wear, and pre-test questions — using mock Seer patient data. Result interpretation never happens here; anything diagnostic routes to your clinical team. Try chat or voice below.
Ask about your removal appointment, when your report goes to your doctor, or if you can shower with the device. Clinical questions hand off.
How we built this
Honest version. Here's what's behind this chat:
Wrote Seer-shaped knowledge
Five hero articles covering fitting day, wearing the device, report timing, billing, and at-home support. The agent quotes from these — never speculates on results.
Built workflows for two real flows
Natural-language workflows for study status (when's removal, when's the report) and device / pre-test support (with clinical callback path). Diagnostic questions never get answered — they route to the clinical team.
Connected mock Seer tools
Three mock tools — patient study lookup, upcoming clinic visits, and clinical callback request — that mirror real systems. In production these point at the Seer study record.
Went live with guardrails on
Published to the live environment with the no-weak-uncertainty guardrail steering every response. Tested across 10 scenarios — see the report.
Want to build the real one?
This took us a focused afternoon with one engineer. Production with your real study record, Medicare item lookups, and the referring-doctor integration takes a few weeks. We've done this for regulated healthcare subscribers across AU, NZ, US, and EU.
Discuss with our experts →
Demo built with