Tribo · Case Study · 2025–2026

AI Automation Suite for an
Aesthetic Medicine Clinic

End-to-end intelligent lead capture, advisor derivation, and CRM orchestration — deployed at zero infrastructure overhead on a global edge network.

Deployed · Cloudflare Pages · Cali, Colombia
9AI Bots Live
73Services Scoped
70%CRM Integration
24/7Lead Capture

Executive Summary

Transforming a Manual Intake Process
into an Autonomous Revenue Engine

An aesthetic medicine clinic operating across multiple specialized service lines faced a critical operational bottleneck: patient lead intake was entirely manual, dependent on staff availability, and unable to scale with growing Meta Ads traffic. Each ad campaign directed prospective patients to generic communication channels that could not qualify leads, collect clinical pre-screening information, or route inquiries to the appropriate specialist advisor.

Tribo was engaged to architect and deliver a comprehensive automation suite that would transform this fragmented process into a unified, intelligent system operating around the clock without human intervention.

Scope of Engagement

Full-Stack Automation Delivery

AI-powered chatbot suite (Fibot) · Custom CRM system · Corporate website · Meta Ads integration architecture · Lizto CRM data pipeline · Centralized analytics dashboard

Client Profile

Aesthetic Medicine Clinic · Cali, Colombia · 9+ specialized service lines in facial harmonization, aesthetic dermatology, and non-surgical procedures · Active Meta Ads campaigns per service vertical

Business Challenge

The Problem We Were Hired to Solve

Fragmented Lead Capture at Scale

The clinic ran independent Meta Ads campaigns for each of its 9+ service verticals — alectomy, lip profiling, rhinomodeling, botox, bichectomy, under-eye correction, nasolabial fold correction, hydroxyfill, and Jalupro biostimulation. Each campaign generated prospective patient inquiries that arrived through generic channels with no automated pre-screening, clinical history collection, or intelligent routing mechanism.

Response Latency

Manual response dependency meant leads went unattended during non-office hours, resulting in prospect attrition and lost conversion opportunities from Meta campaign spend.

📋

No Clinical Pre-Screening

Advisors received unqualified inquiries with no prior clinical history, no reference photography, and no service context — making each consultation operationally expensive and time-consuming.

🔗

Disconnected Data Ecosystem

Patient data resided in siloed systems — Lizto CRM, WhatsApp threads, and paper-based records — with no unified view of the patient journey from ad impression to booked appointment.


Solution Architecture

The Ecosystem We Built

A multi-layer automation architecture connecting Meta Ads traffic directly to an AI-driven lead qualification engine, with bidirectional CRM synchronization and advisor derivation via WhatsApp.

Full System Architecture — Tribo × Aesthetic Medicine Clinic

META ADS 9 Active Campaigns One per service vertical Traffic FIBOT SUITE AI-Powered Lead Capture 9 Deployed Bots: → Alectomía → Perfilamiento Labial → Rinomodelación → Botox Full Face → Bichectomía → Corrección de Ojeras → Corrección Surcos → Hydroxyfill → Jalupro 5.5 Powered by Groq LLM llama-3.3-70b-versatile ⚡ Cloudflare Pages CDN Lead Data CUSTOM CRM 1,888 Patient Records Search · Filter · Export WhatsApp Per Patient Lizto CSV Compatible HTML · Standalone · No Backend Data Sync MAKE AUTOMATION Email Parsing Module HTTP Connector Workflow Orchestration Engine LIZTO CRM Client Appointment Manager CSV Export Integration Email Notification Interception WHATSAPP ADVISOR Instant Lead Derivation +57 350 248 0590 Specialist per service line CORPORATE WEBSITE Service Catalog Video Banner · WhatsApp CTA Embedded Fibot Entry Points Cloudflare Pages · Global CDN ANALYTICS DASHBOARD Meta Ads Performance Lizto Data · Bot Activity In Development Active Integration Data Pipeline (70% complete) Planned / In Development Production · Deployed TRIBO — FULL SYSTEM ARCHITECTURE · AI AUTOMATION SUITE Aesthetic Medicine Clinic · Cali, Colombia · 2025–2026

Fibot Conversational Suite

State-Machine Bot Architecture

Each Fibot instance operates as a deterministic finite-state automaton for core flow steps, while delegating open-ended natural language queries to a large language model inference engine. This hybrid architecture eliminates hallucination risk in critical patient-facing flow steps while preserving conversational fluidity for FAQ handling.

Single-File HTML Groq LLM (llama-3.3-70b) Cloudflare Pages Zero Backend Required iOS Safari Compatible Photo Upload

FAQ Interception Pattern

A dedicated FAQ dictionary intercepts patient questions at any flow step. After resolving the query via Groq LLM, the bot executes a repeatCurrentStep() call, returning the user to their exact position in the flow — ensuring FAQ handling never breaks the lead capture pipeline.

Fibot State Machine — Conversation Flow

START GREETING Bienvenid@ + Fibot intro ANTECEDENTS QUESTION Prior procedures? [Sí / No] No TIME REF How long ago? Regex detection PHOTO REQUEST Frente + Perfil (2 required) Reference image + upload zone 2 photos received NAME COLLECTION Validates non-question inputs WHATSAPP ADVISOR CARD Specialist derivation · Tap to contact FAQ Intercepts at any step Groq LLM resolves query repeat Current Step() LEAD CAPTURED

Technical Infrastructure

Zero-Infrastructure-Overhead Stack

The entire suite operates on a cost-efficient serverless architecture with no dedicated server provisioning, no monthly hosting fees, and global CDN distribution at enterprise-grade performance thresholds.

01
Presentation Layer
HTML5 · CSS3 · Vanilla JavaScript — Single-file architecture per bot, deployable as a flat ZIP to any static host
02
AI Inference Engine
Groq API · llama-3.3-70b-versatile — High-throughput language model inference for FAQ resolution and conversational fallback
03
Edge Deployment
Cloudflare Pages — Global CDN distribution, optimized operational margins, unlimited bandwidth, sub-100ms TTFB
04
Workflow Automation
Make (ex-Integromat) — Multi-step workflow orchestration for Lizto data interception, CRM synchronization, and notification routing
05
CRM Data Layer
Custom HTML CRM + Lizto — 1,888 patient records, CSV pipeline compatibility, WhatsApp direct-dial per record
06
Paid Media Integration
Meta Business Suite — One bot per campaign, direct ad-to-bot traffic routing with no intermediary landing page friction

CRM Integration Architecture

Lizto × Make Data Pipeline

Lizto operates as the clinic's appointment management platform with no public API or webhook infrastructure. Tribo engineered a two-path integration strategy that achieves automated data synchronization without native connector availability.

Integration status: 70% complete — email interception pipeline active, CSV export/import path validated, real-time webhook bridge in final configuration.

Lizto × Make Integration Flow — Dual-Path Architecture

PATH A — AUTOMATED EMAIL INTERCEPTION (PRIMARY) PATIENT BOOKS Appointment in Lizto app.lizto.co LIZTO SENDS Booking confirmation email notification MAKE Email Watch Module Parses + extracts fields PATH B — CSV EXPORT SYNC (SECONDARY) LIZTO EXPORT Manual CSV download Client data snapshot CRM IMPORT CSV mapping layer Field normalization CUSTOM CRM Unified Patient Database 1,888 records · HTML standalone ADVISOR ALERT WhatsApp notification DASHBOARD Unified analytics view ⚙ Integration Status: 70% Complete Path A (email interception via Make) · Path B (CSV import) both validated and operational. Real-time bridge configuration in final stage.

Additional Deliverables

Corporate Website & CRM System

Corporate Website

Clinic Digital Presence

Full-featured corporate website deployed to Cloudflare Pages with service catalog, video production banner, embedded WhatsApp CTA, appointment form, and direct Fibot integration points. Built to the clinic's cream/gold visual identity system.

Cloudflare PagesResponsiveVideo BannerWhatsApp CTA

Custom CRM System

Patient Record Management

Standalone HTML CRM system pre-loaded with 1,888 patient records. Features advanced search, service-line filtering, multi-column sorting, 50-record pagination, CSV export in Lizto-compatible format, and one-tap WhatsApp contact per patient.

1,888 RecordsLizto CompatibleNo BackendWhatsApp Per Row

Outcomes & Impact

Measurable Results Delivered

9
AI bots in production — one per active Meta campaign, fully deployed to Cloudflare global edge network
24/7
Uninterrupted lead capture — patient inquiries qualified and routed to advisors with zero human intervention during non-office hours
100%
Pre-screening automation — every lead arrives at the advisor with clinical history, reference photos, and service context already captured
70%
CRM integration complete — Lizto data pipeline operational via Make automation, eliminating manual data transfer overhead

Operational Efficiency

Advisor time previously spent on initial patient intake re-allocated entirely to high-value consultation activities. Bot handles 100% of pre-qualification at scale.

Infrastructure Overhead

Zero dedicated server costs. Zero devops maintenance overhead. Global CDN performance with optimized operational margins across the entire bot suite.

Scalability Architecture

Each additional service line requires only replication of the bot madre template — a 2-hour deployment cycle per new service vertical, infinitely scalable across the 73-service roadmap.


About the Firm

Built by Tribo

Tribo is a boutique automation and AI integration studio specializing in end-to-end digital infrastructure for service businesses in Latin America and beyond. We architect, build, and deploy conversational AI systems, CRM pipelines, and marketing automation ecosystems — always with production-grade quality and measurable business outcomes.

AI Chatbot Development

WhatsApp bots, web chatbots, lead capture automations with LLM integration

CRM & Automation

Custom CRM systems, Make workflows, data pipeline architecture

Meta Ads Integration

End-to-end paid media to bot to CRM pipelines