See sample plans

A studio concept · design prototype · sample data throughout

The front desk,
without the 9 a.m. queue.

MediFlow AI is a concept appointment assistant for small clinics — it answers scheduling questions, proposes open slots, and hands anything clinical to a human. This page is a design study: every conversation shown is written sample copy, not real patient interaction.

  • No medical advice, ever
  • Human handoff by default
  • Sample data only on this page

Design principles

Trust is the feature. Everything else is scheduling.

Three rules shaped every screen of this concept.

Stay in the lane

The assistant books, reschedules, and answers logistics. The moment a question touches symptoms, medication, or diagnosis, it says so plainly and routes to a human. No hedging, no almost-answers.

Design constraint, not a capability claim — this concept never went near patient data.

Calm by default

Health software gets louder when stakes are high; this gets quieter. Soft sage, generous line-height, one idea per screen. Urgency is conveyed with words and clear next steps — not red banners.

Type: Fraunces for warmth, Figtree for the work. Both chosen for readability at small sizes on clinic tablets.

The front desk stays in charge

Every automation has an override. Staff can pause the assistant mid-conversation, edit proposed slots, and review a plain-language log of what it said and why. It drafts; the clinic decides.

The concept's admin screens are built around review queues, not dashboards-of-dashboards.

How the concept works

Four steps, one human checkpoint

Connect the calendar

The concept reads the clinic's existing schedule — providers, rooms, buffers — and never invents availability it can't see.

Answer the routine

Hours, parking, "do I need fasting?", reschedules — the repetitive 80% gets a fast, sourced answer drawn from the clinic's own handbook.

Propose, confirm, log

Bookings are proposed with real slots, confirmed by the patient, and written to a plain-language log staff can audit line by line.

Hand off, always

Anything clinical, distressed, or ambiguous moves to a human with full context attached — the patient never repeats themselves.

What the concept does

In scope

  • Appointment booking, rescheduling, and cancellations against real slots
  • Pre-visit instructions from the clinic's own handbook (fasting, arrival time, forms)
  • Waiting-list notes: "a spot opened Thursday" — with human confirmation
  • Post-visit logistics: pharmacy hours, lab locations, paperwork status
  • Quiet escalation: symptom questions become care-team tasks, not answers

What it deliberately won't

Out of scope, forever

  • Assess, triage, or interpret any symptom
  • Recommend medication, dosages, or supplements
  • Make promises about diagnosis timelines or outcomes
  • Store conversations beyond the clinic's own retention policy
  • Replace a nurse line — it routes to one

Sample plans

Pricing, imagined honestly

A concept's pricing page should say what a real one often hides: who the product is actually for.

Single practice

$149/month concept price

One provider team, one calendar, up to ~400 assisted conversations a month.

  • Scheduling + logistics answers
  • Staff review queue & transcript logs
  • Handbook import (documents, not scraping)
Fit check questionnaire
Where the design spends most care

Group clinic

$399/month concept price

Up to 12 providers with shared rooms, multi-line routing, and per-provider tone settings.

  • Everything in Single practice
  • Waiting-list workflows with human confirm
  • Front-desk takeover mid-conversation
  • Quarterly transcript audits with your team
See a longer sample log

Care network

Custom— priced per site, not per seat

Multi-location groups needing local phone numbers, translated flows, and regional handbooks.

  • Per-site calendars & policies
  • Language packs designed with clinic staff
  • On-site onboarding weeks, not webinars
Design-partner interest form

Honest answers

Questions a real clinic would ask

Is this an actual product clinics can use?

No. MediFlow AI is a studio concept build — an end-to-end design study of what a trustworthy scheduling assistant could look like. Nothing here processes real patient data or connects to real calendars.

Where did the demo conversation come from?

We wrote it. The clinic, doctor, and slots are invented names, and the conversation is designed to show the most important moment: the assistant declining to answer a clinical question and routing to the care team instead.

Why no testimonials or outcome numbers?

A concept has no users to quote and no outcomes to count. Publishing invented "90% fewer no-shows" figures would be exactly the kind of overclaim this product's whole design exists to avoid.

What would a real build need to clear before launch?

At minimum: a HIPAA-compliant architecture with a business associate agreement, security review, clinician-led escalation policies, accessibility audit, and pilots where the front desk can shut it off in one click. The concept is designed so those constraints are visible in the UI, not bolted on later.

Which parts of this page are real work?

All of it, as craft: the type system, the chat layout, the responsive behaviour, and the copy decisions are the deliverable. It's built as a single static page — semantic HTML, hand-authored CSS, and a few lines of enhancement JavaScript.

Software for waiting rooms
should feel like one.

Quiet, clear, and always one step from a human. That's the thesis of this concept.

Revisit the sample conversation