Most automation breaks within weeks of delivery - a chatbot that hallucinates, an n8n flow that dies on the first malformed input, a system nobody can maintain. The problem usually isn't the AI. It's that nobody understood the process before automating it.
I spent over a decade running sales and operations - managing pipelines, closing deals, building supplier and client networks - before becoming an AI Automation Specialist. That background is what separates my systems from a set of connected API calls: I map what the process actually needs to do first, then build around that.
WHAT I BUILD
🤖 AI Voice Agents & Sales Automation
Inbound and outbound voice agents that qualify leads, score them, and update your CRM automatically, using Happ.tools, n8n, and LLM-based scoring logic.
📩 Support & Email Automation
Multi-step systems that classify, route, respond to, and escalate incoming tickets and emails against SLA deadlines, with logic for anything that shouldn't be handled by AI alone.
📊 CRM & Sales Pipeline Intelligence
Workflows that track deal health, flag at-risk pipeline automatically, and replace manual reporting with automated forecasts.
🔗 Backend & Payment Automation
Python services that monitor inboxes, parse non-standard files, and sync records into your CRM - fully autonomous, with duplicate-prevention and retry logic.
RECENT RESULTS
- B2B Email Support System: daily handling time cut from 4 hours to 20 minutes (85% reduction), with priority-based routing and instant escalation for high-risk messages
- Support Ticket Router: 50-80 tickets/day classified and routed automatically with a 3-level escalation chain, triage time cut by 85%
- AI Voice Sales Qualifier: lead response time cut from 20 minutes to under 2 minutes (98% reduction), 100% of inbound leads auto-qualified and scored into CRM
- Voice AI for Retail: 70% of inbound calls resolved with zero human involvement, 24/7 coverage
- Sales Pipeline Intelligence: manual CRM review eliminated, deals scored weekly, at-risk deals flagged automatically
- Payment Reconciliation System: processing time cut from 60 minutes to under 3 seconds, zero duplicate-processing errors since deployment
BUILT TO NOT BREAK
Every system I build is designed to run without me watching it: structured output validation so the AI can't return malformed data, duplicate-prevention on every trigger, and a full audit trail of every action taken.
TECH STACK
Automation: n8n (self-hosted), Make, Python
AI & Voice: Claude, GPT, Whisper, Happ.tools
CRM & Marketing: HubSpot, Zoho CRM, Pipedrive, KeyCRM, Klaviyo
Integrations: Google Workspace, Telegram Bot API, Gmail API, Slack, Supabase
HOW I WORK
I scope your actual process first, not a generic template. Then I build with proper error handling and escalation logic, document everything, and hand over a system your team can run without depending on me.
If a process in your business is eating hours every week, message me with what it is and the tools you're already using. I'll tell you how I'd automate it.
