Dragan J.

AI Voice Agent Architect | Retell, Twilio, n8n, Supabase | 250K+ Min

I build voice AI systems that answer real phone calls every day. Across the systems I've shipped there are about 250,000 minutes of live call time and roughly 75 agents running for 30+ businesses, and for the last 18 months I've been the fractional CTO of a voice AI company, where I built the entire technical side myself. WHAT I BUILD INBOUND RECEPTIONISTS. Agents that answer the phone 24/7, handle questions the way a front desk would, qualify callers, book appointments against the real calendar, log everything to the CRM, and pass the call to a human when it needs one. One of mine has been running at a multi-sport facility for over a year and still books appointments every week without needing a rebuild. OUTBOUND CAMPAIGNS. Lead reactivation and qualification at volume. The largest was 30,000+ calls for a Canadian auto dealer, booking appointments at 10% and bringing the cost per booked appointment from $15 down to about $4.50. FULL PLATFORMS. This is what I did as a CTO. When a company runs voice agents for many clients at once, the agents themselves are the easy part. What I built is everything around them: a portal where the team builds up and configures a new client's agent from an intake form, a dashboard where each client sees their own calls, transcripts and bookings, an admin side that tracks costs and monitors every agent in production, and billing and onboarding plumbing that takes a new client from signed to live in days. If you're an agency trying to get to that point, this is work I know well. HOW I WORK The first call is me giving you an honest read on whether voice AI makes sense for your case and what it would actually cost to run, because a fair share of the people who reach out don't need it yet, and I'd rather tell you that up front. When we do build, everything runs in your own accounts and gets documented, so you own the system outright and aren't dependent on me after handover. Tools I work with: Retell, Vapi, LiveKit, Pipecat, Twilio, Supabase, n8n, GoHighLevel, Stripe, CRM integrations (HubSpot, Salesforce and others), and custom dashboards and backends when off-the-shelf doesn't fit. On the model side, OpenAI and Anthropic. If your stack is different, that's usually fine. The architecture matters more than the vendor. A GOOD FIT - You have real call volume with revenue attached to answering the phone. - You built a version yourself that kind of works and want someone senior to take it over. - You're an agency that needs white-label builds your clients fully own. If that sounds like your situation, message me with what you're trying to do and I'll tell you how I'd approach it.