The booking & retention engine,
explained end to end.

01System overview

A lead becomes a booked, reminded, retained customer, mostly on autopilot. Emily (the AI) handles the conversation over both SMS and voice; one n8n engine does the real booking in MindBody; and a lifecycle layer keeps customers showing up and coming back.

The big picture

A new lead is synced into GoHighLevel and tagged. Emily, the AI, greets them over text or on a call, answers with real open times, and captures their intent. One n8n engine then books the session in live MindBody and writes back the exact confirmation the customer is sent, the same engine for SMS and voice, so a booking made on a call is visible to a text seconds later. From there the lifecycle runs itself: confirmation and photo reminders, no-show recovery, reviews, referral and membership nurture, and reactivation of cold leads. Every step is written back to the CRM.

The moving parts

Five layers do the work; n8n sits in the middle — it listens to GoHighLevel, runs the booking engine against MindBody, and orchestrates Retell and the reminders.

Lead intake

Outcome Every new lead lands in the CRM as a tagged contact within minutes, so the rest of the system can pick it up. No lead sits in MindBody or an ad platform unseen.

n8n · Beem - Mindbody Leads to GHL — actual build

→ feeds Emily, the calls, and the nurtureLead in · booked session

Trigger

Runs on a schedule, pulling new MindBody (and funnel) leads into GoHighLevel and tagging them gf-new-lead.

Module flow

1. Trigger—Picks up new leads to import.
2. Fetch leads—Reads the new MindBody / funnel leads. MindBody
3. Map + upsert contact—Creates or updates the contact in GoHighLevel. GHL
4. Tag gf-new-lead—Tags the contact so the lead engine takes over.

The call firer

Outcome Any workflow that needs to call a customer hits one webhook, and Emily dials with the right opening line for the situation. No more per-workflow auth bugs.

n8n · Beem - Retell Call Firer — actual build

← called by callback / no-show / reactivationLead in · booked session

Trigger

POST to webhook/beem-fire-call from any workflow that needs to place a call.

Module flow

1. Webhook—Accepts both a simple {phone, reason} body and the full call body.
2. Fire Retell call—Places the call via Retell with Bearer auth and a reason-aware opener. Retell call_reason

The MindBody booking engine

Outcome Whatever Emily promises, book Friday 2pm, move it, cancel, "what's my appointment?", happens for real in MindBody, and the customer gets a truthful confirmation. One engine, both channels, always in sync.

n8n · Beem - MindBody Booking (ghl-book) — actual build

← used by Emily (SMS) AND the voice agentLead in · booked session

Trigger

POST to webhook/beem-mindbody-book from Emily (the GHL SMS Conversation-AI) and the Retell voice agent's function tools.

Module flow

1. Webhook—Receives the action: book / check / reschedule / cancel.
2. Resolve MindBody client—Finds the customer in MindBody by pinned id, then phone / email, to avoid duplicate profiles. pinned id
3. MindBody action—Books, checks, moves, or cancels the session against live MindBody. MindBody timezone
4. Write reply + tag + pin—Writes the exact SMS reply (booking_reply), tags booked, and pins the MindBody client id. booked
5. Respond—Returns the result so the channel can speak it.

After every call

Outcome Every call leaves a visible record on the contact (summary + transcript), a no-answer is logged as "called, no answer," and the customer is never stranded behind a stale AI flag.

n8n · Beem - Retell Post-Call (notes + tags) — actual build

→ writes the record · cleans the sessionLead in · booked session

Trigger

Retell calls this webhook when a call is analyzed or ended (all three agents point here).

Module flow

1. Webhook—Receives call_analyzed (connected) or call_ended (no answer).
2. Note + tag + hygiene—Finds the contact, writes summary + transcript, tags the outcome, removes ai_active. visible artifact

Reminders & no-show recovery

Outcome Customers get timely, friendly reminders with real photos of where they are going, which cuts no-shows. If they still miss it, recovery kicks in.

n8n · Beem - Reminder Dispatcher — actual build

→ confirmation · 24h · 1h · recoveryLifecycle · reminders & retention

Trigger

Runs on schedule, dispatching the right reminder for each upcoming appointment.

Module flow

1. Trigger—Fires on cadence to find due reminders.
2. Build + send reminder—Sends the confirmation, the 24h storefront-photo MMS, and the 1h parking-photo MMS. MMS photo

No-show & session tracking

Outcome MindBody is the source of truth for who showed up. Completed visits feed the review and referral flow; no-shows trigger an automatic rebooking SMS and a recovery call. Last-visit and session-count fields stay current.

n8n · Beem - Attendance Poller — actual build

→ session-complete · no-show recoveryLifecycle · reminders & retention

Trigger

n8n schedule, every 30 minutes (and on demand).

Module flow

1. Cron—Runs every 30 minutes.
2. Scan MindBody attendance—Reads ended appointments across all rooms. MindBody
3. Tag + bump fields—Tags session-complete or no-show, bumps last-visit / session-count, writes a CRM note.

Slot prefetch

Outcome Emily quotes real, current openings the moment a customer asks, with no "let me check on that" dead air.

n8n · Beem - Availability Prefetch (cron) — actual build

→ keeps Emily's answers instantLifecycle · reminders & retention

Trigger

n8n cron, refreshing open slots for active (ai_active) contacts.

Module flow

1. Cron—Refreshes on a short cadence.
2. Read MindBody slots—Pulls the real open slots for active customers. MindBody
3. Write available_slots—Stores them on the contact so Emily answers instantly.

Scheduled callbacks

Outcome A requested callback happens at the time the customer actually asked for, in their words, with a final check so it is skipped if they have since cancelled or opted out.

n8n · Beem - Callback Scheduler — actual build

→ waits, then fires via the call firerLifecycle · reminders & retention

Trigger

POST to webhook/beem-callback-schedule when a contact is tagged callback_requested.

Module flow

1. Webhook—Fires when a callback is requested.
2. Parse callback time—Reads the customer's own words for the time, Phoenix-anchored. America/Phoenix
3. Wait until the time—Pauses precisely until that exact moment.
4. Recheck + fire—Skips if they cancelled / opted out, else fires the call. GR-001

Error alerting

Outcome Any broken run pings a monitored inbox immediately, so issues get caught and fixed fast instead of silently dropping customers.

n8n · Beem_error_alert — actual build

fires on any workflow errorOps & alerts

Trigger

Set as the error workflow on the Beem workflows; fires on any failure.

Module flow

1. Error trigger—Catches a failure in any linked workflow.
2. Email the alert—Sends the error to a monitored inbox. jjcavada1@gmail.com

GoHighLevel · the CRM layer

GoHighLevel holds the conversation, the contacts, the tags, and the reminders. It runs “Emily,” the SMS bot, and drives the lifecycle workflows. The booking system of record is MindBody, the n8n engine books there; GoHighLevel is where the lead lives and where every step is written back.

The tag lifecycle that drives everything

A contact moves through these tags; each transition fires an automation.

Key custom fields

GHL-side workflows (WF-1 to WF-10)

Lead ignition, the nurture sequence, Emily takeover, the MindBody booking handler, reminders, post-visit review, referral + membership nurture, member lifecycle, and non-booked reactivation — all tag-driven, all calling the n8n engines.

Retell · the voice agent

Retell is Emily's voice. A GPT-4.1 outbound agent calls leads (and handles callbacks, no-show recovery, reactivation), an inbound agent answers the line, and a membership agent runs the warm upsell. Mid-call they reach into the same n8n MindBody engine the SMS bot uses, so a booking made on a call is visible to a text seconds later.

Functions — the tools the agent calls mid-call

Each is a live webhook into the n8n MindBody engine. The prompt names these as the only functions that exist, so the agent never invents a capability.

Dynamic variables — context passed in at call time

Injected so the agent knows the caller and why it's calling before it speaks:

Prompt structure

The prompt leads with a hard, authoritative override so the agent never hallucinates a booking. Its backbone:

Full workflow inventory

Beem n8n workflows with live status. firing = running now · standby = built, on demand · off = inactive / superseded. The nine firing workflows are detailed in full above.

Firing — live production firing

Standby & building blocks

What triggers what

The master map: every source event, the workflow it hits, and what happens next.

Key IDs & config

Identifiers the workflows rely on. Secrets (keys/tokens) are redacted to their shape — real values live in the credential vault.

Who to contact