Selected work

Building a company's technical backbone from scratch

Built and now run the technical backbone of an AI voice-receptionist company. Grew into the fractional CTO role, named publicly. 18 months and counting.

150,000+call minutes shipped
app.primesync.ai
PrimeSync · Client Portal
1,284
Calls this month
12,940
Minutes used
$48.2k
Est. revenue
312
Appointments set
Call outcomes
Appointment set
Quote requested
Message left
Transferred
General inquiry
Live status
Agent online · handling calls
Recent calls
Inbound · 2m 14sAppointment set9:42 AM
Inbound · 0m 51sQuote requested9:30 AM
Inbound · 3m 06sTransferred9:12 AM

Representative interface. Built in-house on the client's own accounts; shown with sample data.

The situation

PrimeSync set out to build an AI voice receptionist for auto repair shops, with TorQi as the managed service that delivers it. When I joined they had a clear product vision and no technical foundation to deliver it on. This was early in the voice-AI wave, so there was no playbook to copy. Someone had to decide how the whole thing would get built and run, from the agent itself to everything around it.

What I figured out

I worked out the product end to end. Starting from their vision, I chose the stack and the architecture that could actually ship it, at a time when very few companies were building voice AI this way. Two calls shaped everything after. First, because the auto-repair use case was so well understood, the agent setup could be productized instead of hand-built every time, without losing the ability to customize. Second, they would own their own dashboards rather than lean on white-label tools, because real visibility for clients and the internal team was going to be the difference between a demo and a business.

What I built

The technical backbone of the company, and the systems the team and clients run on every day. An Agent Factory that turns a prospect intake form into a customized demo agent fast, so a powerful agent gets in front of a prospect early. A client portal where each shop sees call counts, minutes, estimated revenue, call outcomes, full logs, and per-call transcripts and recordings. An internal admin dashboard, built in-house, that gives the team visibility and tooling across every account.

The result

About 20 shops live, running roughly 45 agents. The platform handles around 13,000 call minutes a month, and I have shipped over 150,000 call minutes to date. I grew from voice solutions manager into the company's fractional CTO, named publicly as CTO on PrimeSync's site, and this is my most consistent engagement over 18 months.

Stack

Supabase (Postgres, edge functions, auth, storage), Next.js, n8n, Python, Retell (voice), Twilio (telephony). Every dashboard built in-house, no white-label.