Front-Desk AI Copilot for a 26-Location Salon and Spa Group
Let's Connect
Overview
What we built
A 26-location salon and spa group in the US Midwest had front-desk coordinators juggling rebooking, retail recommendations and membership queries at every checkout, and new hires took weeks to get up to speed. We built an AI copilot embedded in the booking and point-of-sale screens that suggests the rebooking prompt and retail pairing for coordinators to approve.
In plain terms: one coordinator at checkout was expected to remember a guest's visit history, know the right rebooking prompt, recall which retail pairing suited that guest, and answer membership and policy questions correctly, all in the time it takes to close out a ticket. New hires took weeks to learn service menus and membership rules well enough to do this reliably, rebooking prompts got skipped whenever a rush period hit, and answers to the same policy question varied from one of the 26 locations to the next.
We developed an AI copilot embedded directly in the group's booking and point-of-sale screens. It surfaces each guest's visit history, suggests a tailored rebooking prompt and retail pairing at checkout, and answers membership and policy questions from a retrieval index of the group's own documents, with a coordinator reviewing every suggestion before acting. Rebooking at checkout rose from 31% to 44% of eligible visits within 4 months, and new coordinator ramp-up time was cut from 6 weeks to 3 across the 26 locations.
The Problem
Overloaded front-desk teams
Front-desk coordinators carried a lot in their heads at every single checkout: a guest's visit history, the right rebooking prompt for that guest, a retail pairing worth suggesting, and the group's membership and policy rules, all recalled correctly under the time pressure of a queue at the desk. Nothing in the booking or point-of-sale screens helped a coordinator remember any of it.
New hires felt this hardest. New hires took weeks to learn service menus and membership rules well enough to work a checkout confidently, which meant every location was regularly running with someone still learning the job. During rush periods, rebooking prompts were the first thing skipped, since a coordinator under pressure defaults to finishing the transaction rather than remembering to ask.
Consistency was the group-wide cost. Answers to policy questions varied from one location to the next, so a guest's experience depended on which of the 26 locations they visited and which coordinator happened to be on the desk that day, rather than on a single, reliable version of the group's own rules.
Too much to remember at checkout
Coordinators were expected to recall visit history, rebooking prompts, retail pairings and policy rules for every guest, all inside the time pressure of a checkout queue.
Slow new-hire ramp-up
New hires took weeks to learn service menus and membership rules well enough to run a checkout confidently on their own.
Rebooking skipped under pressure
Rebooking prompts were the first thing dropped during rush periods, as coordinators focused on finishing the transaction in front of them.
Inconsistent policy answers
Answers to policy questions varied from one of the 26 locations to the next, so a guest's experience depended on which desk and coordinator they reached.
What it was costing them
Every skipped rebooking prompt was a visit that did not get rescheduled, and every inconsistent policy answer chipped at a guest's confidence in the group as a whole. With new hires taking weeks to reach full competence at each of the 26 locations, the group was constantly running checkouts below the standard its own trained coordinators could deliver.
The Solution
Copilot inside every checkout
We developed an AI copilot embedded directly in the group's booking and point-of-sale screens, so it appears exactly where a coordinator is already working rather than as a separate tool to check. The copilot surfaces each guest's visit history the moment their record opens, giving a coordinator instant context without having to recall or look it up.
At checkout, the copilot suggests a tailored rebooking prompt and a retail pairing, and answers membership and policy questions by drawing on a retrieval index of the group's own documents, so the answer is the same regardless of which of the 26 locations asks it. Coordinators review every suggestion before acting, keeping the final call with the person facing the guest.
We extended the copilot beyond the desk itself. It drafts follow-up messages for approval, so a rebooking nudge or a thank-you can go out after the visit without a coordinator writing it from scratch. New hires lean on the same suggestions experienced coordinators use, which shortened how long it took a new hire to work confidently on their own.
Key decisions
Embed inside existing screens
The copilot lives directly in the booking and point-of-sale screens coordinators already use, rather than a separate application competing for their attention at checkout.
Surface visit history automatically
Each guest's visit history appears the moment their record opens, so a coordinator has context without needing to recall or search for it.
Suggest, never send automatically
The copilot suggests rebooking prompts, retail pairings and follow-up messages, but coordinators review every suggestion before it reaches a guest.
Ground policy answers in the group's own documents
Membership and policy questions are answered from a retrieval index of the group's own documents, so the answer stays consistent across all 26 locations.
Support new hires with the same suggestions
New hires see the same rebooking and policy suggestions experienced coordinators use, shortening how long ramp-up takes before they work independently.
Measurable Impact
What changed after launch
Rebooking behaviour changed quickly. Rebooking at checkout rose from 31% to 44% of eligible visits within 4 months, since the prompt now appeared automatically instead of depending on a coordinator remembering to ask, even during a rush period.
Ramp-up and consistency both improved. New coordinator ramp-up time was cut from 6 weeks to 3 across the 26 locations, membership and policy queries resolved 38% faster than the pre-launch baseline, and retail attach rate rose 17% versus control locations over a 12-week comparison, evidence that the same suggestions helping guests also helped the business.
Rebooking rate
31% of eligible visits rebooked at checkout
44% within 4 months of launch
Coordinator ramp-up
6 weeks to reach full competence
3 weeks across the 26 locations
Policy query speed
Pre-launch baseline resolution time
38% faster resolving membership and policy queries
Retail attach rate
Attach rate at control locations
Up 17% over a 12-week comparison
Headline results
Rebooking at checkout rose from 31% to 44% of eligible visits within 4 months
New coordinator ramp-up time cut from 6 weeks to 3 across the 26 locations
Membership and policy queries resolved 38% faster than the pre-launch baseline
Retail attach rate up 17% versus control locations over a 12-week comparison
Tech & Tools Used
What powered the build
Every tool below earned its place in this engagement. Here is the part each one played.
Next.js
Renders the copilot's suggestion panels inside the booking and point-of-sale screens, so rebooking prompts and retail pairings appear directly where coordinators are already working.
Node.js
Runs the backend service that pulls each guest's visit history and assembles the context the copilot uses to generate its checkout suggestions.
OpenAI API
Generates the tailored rebooking prompts, retail pairing suggestions and follow-up message drafts, working from the guest context the backend assembles for each checkout.
LangChain
Orchestrates the retrieval and generation steps, wiring the membership and policy document index to the generation calls that answer a coordinator's question.
pgvector
Indexes the group's membership and policy documents so relevant passages can be retrieved for any policy question a coordinator or the copilot needs answered.
PostgreSQL
Stores guest visit history, coordinator approvals and the suggestion history the copilot draws on across all 26 locations.
Redis
Caches frequently accessed guest records and in-progress checkout sessions, keeping suggestions quick to appear during busy rush periods.
Twilio SMS
Sends the follow-up messages the copilot drafts once a coordinator approves them, delivering rebooking nudges straight to the guest's phone.
Docker
Packages the copilot's services so the same environment runs consistently across development and the point-of-sale rollout to all 26 locations.
AWS ECS
Runs the copilot's backend services in production, scaling to handle checkout traffic across the group's 26 locations.
Ready to Build your Beauty, Spa & Wellness Business with AI Copilots
Ask Byte
Ask Byte
Typically replies instantly
just Now
Hi! I'm OrganByte's assistant. How can I help you today?
AI-generated content may be incorrect
