FlatClaw, Private AI Platform
All use case spotlights
Voice AgentsCustomer EngagementBeauty & WellnessFranchise & Retail

Franchise beauty and wellness group

≈ $300M+ system-wide sales (salon brand) · 700+ locations · two franchise brands

Inbound booking voice agent

A private two-way voice agent answers the booking line, matches each caller to one to three professionals and texts them booking links. Text and web chat reuse the same matching service. The day-one tenancy handles twenty-plus simultaneous calls.

Organization
Franchise beauty and wellness group, hundreds of locations
Channel
Inbound booking calls on each location's existing number
Runs on
The customer's cloud tenancy, two GPUs
Scope
Voice agent first; text and web chat on the same brain
The situation

Where they started.

Consumers call a location to book, and the booking branch of the phone tree mostly ends in voicemail: locations are run by independent professionals, not a front desk. The group wanted every caller matched to the right professional and handed a way to book, without touching the leasing side of the same phone number and without owning anyone's calendar.

What FlatClaw does

What was built.

  • A private two-way voice agent attached to the booking branch of the existing phone platform, location by location; leasing, inboxes and voicemail behave exactly as before, and voicemail remains the fallback.
  • A real conversation: streaming speech recognition, an open-weight language model and natural speech synthesis running on two GPUs in the customer's cloud, with interruptions handled the way a person would.
  • A matching and routing service built as its own API: it holds each location's professional directory and services, recommends one to three professionals with reasons, and texts the caller booking links; text and web chat become new front doors onto the same service.
  • Every call lands as a transcript, a reasoning trace and a lead record with attribution, so the group can prove which leads became bookings.
Results

What changed.

  • The booking line is answered every time, at every hour, instead of landing in voicemail.
  • Callers leave with a text naming the right professionals and a link to book.
  • The day-one tenancy handles twenty-plus simultaneous calls; capacity grows by adding voice workers.
  • Lead attribution good enough to build a per-lead business model on.
Why private

Caller audio, phone numbers and conversation history are consumer data. They never leave the group's own cloud account, and the branded voice is an asset the group owns rather than rents by the minute.

Runs on

The customer's cloud tenancy

The stack

Components involved.

FlatClaw voice serviceStructured conversation graphMatching and routing servicePhone platform media streamsOpen speech and language models on dedicated GPUsReview console
Your workflow

Have one like it?

Every spotlight started as a conversation about a process nobody liked doing, under a data-locality constraint.