2 ready-to-build workflows

AI agent workflows for Beauty, salons & spas

Booking, waitlist and no-show reminders, package renewal and win-back, review solicitation and service Q&A — on the channels clients already use.

Each recipe below is expressed only in VegaDūta's real workflow building blocks — Trigger, Agent, Knowledge, Condition, Approval, Output, Tool (MCP), Voice, Device, Loop, Code and Parallel — so it maps 1:1 to something you can assemble in the Workflow designer. Consequential actions always pass a human Approval step.

1. Booking with waitlist, no-show reminders and service Q&A

Clients book by DM and call while the front desk is mid-blow-dry; the waitlist lives on a sticky note, cancellations aren't backfilled, no-shows leave a stylist's chair empty, and half the messages are just 'how long does balayage take?' or 'do you do gel removal?'.

Trigger: WhatsApp / Call (booking request or service question)

How the workflow runs

  1. Trigger

    Client messages. A WhatsApp or Call trigger captures a booking request or a service question.

  2. Agent

    Confirm service, stylist & timing — or answer the question. An Agent node identifies the client, the service, a preferred stylist and time and confirms the details; for a service question (duration, what's included, aftercare, prep) it answers grounded in the salon's menu and policy — never inventing a price or a service you don't offer.

  3. Condition

    Full? go to the waitlist. A Condition node checks live availability: if the stylist/slot is full it adds the client to the waitlist and auto-offers the spot when a cancellation frees it; otherwise it proceeds to book.

  4. Tool (MCP)

    Write to the booking / POS system. A Tool (MCP) node records the appointment (or waitlist position) in your salon booking / POS system — no first-party connector, MCP calls yours.

  5. Output

    Confirm + no-show reminder. An Output node sends the confirmation, then a day-before reminder on WhatsApp/SMS with one-tap confirm or cancel — a cancel instantly releases the chair to the next waitlisted client.

Channels & connectors

  • WhatsApp
  • Voice / Call
  • SMS
  • Knowledge base
  • Tool (MCP → booking/POS)

Outcome

Appointments book and backfill themselves, the waitlist works without the front desk, service questions answer themselves from the real menu, and reminders turn silent no-shows into freed-up chairs.

Why it helps

Freed cancellations and timely reminders are what keep chairs full — the manual waitlist skips exactly the backfills a busy front desk never gets to — while grounded service answers stop the endless 'how long / how much' back-and-forth from interrupting the floor.

Build spec

1 agent — 1 booking + Q&A agent; the capacity/waitlist branch, the booking write and the reminders are workflow nodes.

System prompt (paste-ready)

You are a booking and front-desk assistant for {{salon}}. Identify the client, the service they want, any preferred stylist and time, and confirm the details before booking. Only offer services, stylists, durations and prices that the booking system and salon menu return as real — never invent a slot, a stylist's availability, a price, or a service you don't offer. If the slot is full, offer the waitlist and explain they'll be auto-offered the next cancellation. For service questions (how long, what's included, prep, aftercare), answer only from the salon menu/policy in Knowledge and cite it; if it isn't there, say so and offer to check with the team. Do not give dermatological, medical or allergy advice — for skin/allergy concerns, advise a patch test or a professional.

MCP connectors

  • Salon booking / POS system — via a Tool (MCP) node (no first-party connector)
  • WhatsApp + SMS — via the messaging channels

Built-in tools

  • http_request (check live stylist/slot availability + write the booking)
  • knowledge_search (service menu, durations, prices, policies)

Guardrails

  • Only books slots the booking system confirms as available, and only quotes services/prices from the salon menu — no invented slots, stylists, prices or services
  • A cancel or no-show instantly releases the slot to the next waitlisted client; the booking write is idempotent so a repeated confirm never double-books a chair
  • No dermatological, medical or allergy advice — skin/allergy concerns are referred to a patch test or a professional

Cost strategy

Booking and menu Q&A is structured, grounded work, so keep it on the economy/standard tier — reserve any premium tier for genuinely open-ended reasoning this flow doesn't need. Cap tool-calls per conversation so a chatty booking exchange stays cost-predictable.

Output & delivery

The agent confirms the request or answers the service question from the menu → the Condition node books directly or adds to the waitlist by live availability → a Tool node records the appointment/waitlist position in the booking/POS system via MCP → an Output node sends the confirmation and a day-before reminder with one-tap confirm/cancel on WhatsApp/SMS.

2. Package renewal, win-back and post-visit review requests

Prepaid packages and memberships lapse with no nudge, clients who haven't been back in months are never re-invited, and the five-star review that a delighted client would happily leave never gets asked for — so reputation and repeat revenue both leak quietly.

Trigger: Webhook (renewal/lapse run + visit-completed event)

How the workflow runs

  1. Trigger

    Renewal run + completed visits. A scheduled Webhook trigger each day evaluates packages/memberships nearing their end and clients who've lapsed; a second Webhook fires when a visit is completed, to ask for a review. The booking/CRM has no first-party connector, so these arrive by Webhook.

  2. Loop

    Per client. A Loop node iterates the list so each client gets the right message — a package/membership renewal reminder, a win-back note tailored to how long they've been away, or a post-visit review request.

  3. Agent

    Personalise the message. An Agent node writes a renewal reminder, a warm win-back referencing the service they usually book, or a friendly review request referencing today's visit and stylist — never inventing an offer or a service history.

  4. Approval

    Manager approves discounts. An Approval node holds any win-back discount, package upgrade or fee waiver until a manager signs off — a pricing concession is a financial decision that stays with a person.

  5. Output

    Reach out — review link or pay link. An Output node sends the message on WhatsApp: a review link for the post-visit request, or a Razorpay UPI link so an approved renewal/package can be paid in one tap.

Channels & connectors

  • Webhook
  • Loop
  • WhatsApp
  • Approval / HITL
  • Razorpay UPI

Outcome

Lapsing packages get a timely, personal nudge, dormant clients get a warm re-invite, and happy clients actually get asked for the review they'd have left anyway — with every discount held for manager approval.

Why it helps

Consistent, personal follow-up and a well-timed review ask are the mechanism that plugs the leak — the manual process skips exactly the renewals, win-backs and review requests a busy salon never gets around to, while pricing concessions stay human-approved and payment stays on a Razorpay UPI link.

Build spec

1 agent — 1 outreach agent per client; the per-client Loop, the discount Approval and the pay-link/review-link send are workflow nodes.

System prompt (paste-ready)

You are a retention and reputation assistant for {{salon}}. Depending on the record type, write ONE short, personal message in the client's language: a package/membership renewal reminder, a win-back note for a lapsed client (referencing the service they usually book and how long it's been), or a post-visit review request (referencing today's service and stylist). Use only the client's actual history from the record — never invent a visit, a service, or a loyalty balance. Only mention a discount, package upgrade or waiver AFTER it has been approved at the approval step; never promise or invent pricing on your own. For a review request, invite honest feedback and route anything negative to the manager rather than pushing for a public rating. Honour opt-outs and don't over-message someone who hasn't replied.

MCP connectors

  • Booking / CRM / membership system — via a Tool (MCP) node (no first-party connector)
  • Razorpay UPI — pay-link for approved renewals/packages
  • Completed-visit + renewal-due events — inbound via Webhook

Built-in tools

  • knowledge_search (current packages, prices, membership records)

Guardrails

  • No discount, package upgrade or fee waiver goes out until a manager signs off at the Approval node — the agent never invents an offer or a price
  • A review request invites honest feedback and routes negatives to the manager privately — it never incentivises or gates a public rating; frequency caps and opt-outs are honoured per client
  • Payment is collected only through the Razorpay UPI link on an approved renewal/package — the agent never handles card or account details itself

Cost strategy

This is per-client outreach that runs at list scale, so the drafting agent belongs on the economy tier — that one choice dominates total cost since it runs once per client. Batch the Loop and cap tool-calls per record so a large renewal run stays linear and predictable.

Output & delivery

A Loop runs the agent per due/lapsed client and per completed visit → the agent drafts the matching message from the client's real history → any discount is held for the manager Approval node → an Output node sends it on WhatsApp with a review link (post-visit) or a Razorpay UPI link (approved renewal/package).

Related industries

Build one of these in minutes

The sandbox gives you a live agent workspace — no account, no card. Or head back to the full catalogue and compare patterns across every industry.

Try this in the sandbox — no signupBrowse all industries
See how VegaDūta compares to n8n, Dify and BotpressWhy VegaDūta's architectureEstimate your WhatsApp API costs