Two-plus years designing automations and AI agents in GHL, n8n, Make, and Zapier — connecting CRMs, payment tools, and inboxes so leads get answered, data stays accurate, and nothing depends on someone remembering to do it manually.
Nothing goes to a lead without one quick human check first — and if they don't book right away, they don't disappear, they go on the follow-up list.
Every project below shipped to production — not a demo. Each one replaced a manual step with something that runs itself and tells someone when it can't.
Handles lead capture and instant follow-ups through email and SMS.
Handles live customer conversations for an online shop, pulling business context and replying through Gemini in real time, day or night.
Engages leads in natural language, checks a live calendar, qualifies the prospect, and books the meeting — no human touches the conversation.
Connects proprietary business documents to a vector database so the AI answers from real company data instead of guessing — built to not hallucinate.
Worked around Xero's API limits by auto-generating an Account Transactions report and attaching it to the right Asana task the moment work finished — no manual exports.
Enriches inbound leads via Apollo, routes by priority, alerts the team in Slack, logs the top ones to Sheets, and drafts a personalized first reply with AI.
Platform depth on the left of each bar means daily, production-level fluency — not "used it once."
Two views from the Facebook Page AI chat agent project.
Real results from production automations solving real problems.
"The Facebook Page AI chat agent handles the repetitive back-and-forth so our team can focus on customers who need real attention. Response times went from hours to minutes, and our team actually has breathing room now. It's exactly what we needed to scale without the chaos."
This sequence exists because skipping a step is exactly how automations break quietly in production.
Not the process on paper — the one your team actually follows, exceptions included.
Wire the CRM, payment processor, and comms tools together with the logic the map calls for.
Missing fields, empty payloads, duplicate triggers — found before a client's lead is, not after.
Written clearly enough that a teammate can maintain it without a follow-up call to me.
Live automations get watched, not abandoned — breaks get caught and fixed fast.
Building automation tools used for testing and day-to-day engineering workflows.
Designing and shipping AI automation across Zapier, Make, n8n, and GoHighLevel for marketing, ops, and support teams.
Grab a 30-minute slot that works for you. If nothing fits, email still works too.
That's usually the first thing worth automating. Tell me what it looks like today.