GoHighLevel · AI Automation · Integrations

I build the systems that make sure no lead ever leaks.

I'm Lexter Andrew "Druu" Rivera, an automation specialist from the Philippines. I design, build, and document CRM and AI systems end to end: GoHighLevel environments, WhatsApp automation on the official Meta Cloud API, AI voice agents, and the n8n integrations that hold it all together.

Antipolo, Rizal, Philippines · Remote · Overlaps AU and US hours · Open to full time roles

Druu Rivera, GoHighLevel and AI automation specialist
2+ yearshands-on GoHighLevel
15 workflowsin one production clinic system
113 leadsthrough an AI voice outbound dialer
Official Cloud APIlive Meta app + native WhatsApp Flows
Proof of work

Real systems I personally built

Not mockups and not coursework. These run in GoHighLevel, n8n, and the Meta WhatsApp Cloud API, and every one ships as a cloneable snapshot with SOPs so a team can run it without me.

Conversation AI · Live Paying Client

Messenger + Instagram AI Receptionist

Problem
Clinic inquiries arrive at night on Messenger and Instagram; by the time the front desk replies, the patient has booked with whoever answered first. And Meta's 24-hour rule blocks reminder messages exactly when the clinic needs them.
What I built
A GHL Conversation AI receptionist, in production for a paying dental clinic: instant Taglish-friendly replies, qualification that writes tags into the CRM, real calendar booking, and human handoff at any point. The reminder chain re-opens Meta's 24-hour window with a one-tap OK reply loop, so confirmations run with zero SMS cost.

GHL Conversation AI · Messenger + Instagram DM · GHL calendars & workflows

Running live for a paying client right now; the OK reminder loop is proven in real production conversations, not a demo.

💬 Talk to the bot yourself

Live demo, not a video. Ask a price. Try to book. Try to break it. It runs on my demo clinic page, because I never send strangers into a real client's inbox.

Messenger AI receptionist conversation: instant Taglish reply, qualification, calendar booking, and the OK reminder loop that solves the Meta 24-hour window
Conversational CRM

Aesthetic Clinic WhatsApp System

Problem
Clinics were losing after-hours inquiries and letting warm leads go cold between WhatsApp chats and the front desk.
What I built
A 15-workflow GoHighLevel system: button-menu qualification by treatment type, automated FAQ paths, nurture and booking sequences, reminders, no-show recovery, and human handover at any point. Fully bilingual English and Arabic.

GoHighLevel · n8n (self-hosted) · WhatsApp bridge · EN/AR

Zero-leakage discipline throughout: every opportunity update re-finds its record first, so the pipeline never silently drops a lead, even after wait steps.

Clinic system workflow list in GoHighLevel
Outbound Sales Engine

Solar Lead Engine with AI Voice Dialer

Problem
Outbound lists sat untouched because manual dialing does not scale for a solo operation.
What I built
A six-workflow engine: lead import and tagging, pre-qualification, an AI voice outbound dialer with 113 leads enrolled, a call-outcome router that sends every disposition to the right next step, plus email outreach and reply handling.

GoHighLevel · Vapi · Twilio · call-outcome routing

The dialer is an AI voice agent making real outbound calls, with outcomes written straight back into the pipeline.

Solar outbound engine workflow list in GoHighLevel
Official WhatsApp Platform

Meta Cloud API + Native WhatsApp Flows

Problem
Real WhatsApp automation, native in-chat forms included, only works on the official Meta platform, which most builders never touch.
What I built
A live Meta business app on the official WhatsApp Cloud API, plus a native booking Flow: name, service, date, and time captured entirely inside the chat, no external links.

Meta WhatsApp Cloud API · WhatsApp Flows · Flow JSON · n8n webhooks

Honest status: the Flow is built and previewing; publishing sits behind Meta business verification, which is exactly the kind of platform constraint I flag to clients up front.

Native WhatsApp booking Flow previewing in WhatsApp Manager
AI With Guardrails

Claude Document Agent

Problem
AI that reads documents will eventually hallucinate, and in finance-adjacent work a confident wrong answer is worse than no answer.
What I built
A Python agent on the Claude API that extracts structured data from English and Arabic invoices, then validates every number deterministically: totals, tax math, line items, duplicates. Anything uncertain is flagged for a human.

Python · Claude API · structured JSON outputs · deterministic validation

My rule for every AI build: the AI drafts, the code verifies. AI never gets the final say unchecked.

Claude Document Agent pipeline: extract with Claude, validate deterministically in Python, route to verified or human review
Integration Layer

Self-Hosted n8n Integration Bridge

Problem
GoHighLevel cannot natively talk to every platform a business runs on, and rented connectors add monthly cost, rate limits, and someone else's roadmap.
What I built
A self-hosted n8n integration layer running two-way data flow GHL does not support natively. The workflow shown: a WhatsApp reply lands on a webhook, the contact is found in GHL via the REST API, upserted if new, and tagged so the right GHL workflow fires next, all automatic.

n8n on Linux/Docker · REST APIs · webhooks · PostgreSQL · Redis

I run the full stack myself: n8n, PostgreSQL, and Redis behind a Cloudflare Tunnel on my own server.

n8n integration workflow canvas
In the ecosystem

Serious about the platform

I build on GoHighLevel every day and I show up for the ecosystem around it. In August 2026 I attended HighLevel LIVE Manila to sharpen the exact systems I build, and to meet the people behind the platform.

Druu Rivera with the HighLevel founders at HighLevel LIVE Manila 2026

With the HighLevel founders at HighLevel LIVE Manila 2026.

How I work

The discipline behind the builds

Zero leakage

Find before every update

I diagnosed a silent GoHighLevel failure where opportunity context drops after wait steps by reading the actual run logs, then built a re-find pattern that eliminated the whole class of bugs. It is now a standard in every build.

Repeatable

Everything ships as a snapshot

Every system is built to clone into a new account in minutes, with merge-field templates so personalization is configuration, not rebuilding. That is how one build serves many clients.

Documented

SOPs a team can actually run

Each build hands over with written SOPs and walkthroughs so non-technical teammates can operate, update, and troubleshoot without me. If only I can run it, it is not finished.

Honest

Truth over overclaiming

I tell clients what a platform cannot do before they pay to find out, Meta approval timelines, API limits, template gates. The proof on this page is labeled exactly as it is, nothing inflated.

Contact

Let's talk about your systems

I'm open to full time remote roles and serious project work. The fastest way to evaluate me: a 20 minute call where I screen-share a live build and you ask anything.