Portfolio

Three builds, three bottlenecks.

Every one of these started the same way: a person doing something by hand, over and over, until it capped what the business could take on. None of them started as a product we already had.

These are three of our builds, not all of them. The screens below are intentionally simplified — built to show the shape of the fix, not the full workings underneath.

Inbound operations

AI voice agent

Real-time voice AI that runs the call from hello to hang-up.

The bottleneck

Every call had to be taken by a person, verified by a person, paid to a person and typed up by a person. After five, calls just stacked up until morning.

What we built

A voice agent that answers in real time and holds an actual conversation. It verifies the caller, handles the request, takes payment, and writes the full call — transcript, outcome, money — back into the CRM the second it ends. Whoever's running the floor watches it live in an ops console; the system reports its own health so nobody's guessing whether it's up.

What it changed

The phone stops being a staffing problem. Same answer at 2 p.m. and 2 a.m., and no call ends without a record.

Real-time voice Caller verification Payments CRM writeback Live ops console Diagnostics
Expand
The live ops console — calls in flight, outcomes, system health.
Expand
Quote generated from the pin, before anyone drives out.
Concrete contractor

Geolocation estimator

Instant remote quotes, priced by where the job actually is.

The bottleneck

Every quote meant a truck, a driver and most of a day. Anything past an hour out wasn't worth bidding, so it didn't get bid.

What we built

An estimating tool that prices a job from its location — distance, drive time, access and site conditions run against the contractor's own rate sheet — and returns a real quote on the spot, remotely.

What it changed

Service range stopped being limited by windshield time. Work that used to be "too far to bid" is now just another quote sent before lunch.

Geolocation pricing Rate-sheet logic Remote quoting Customer-facing
Office furniture reseller

Liquidation lead agent

Sorts the junk, prices the job, matches the stock.

The bottleneck

Liquidation leads come in fast and most go nowhere. Reading them, pricing install and removal, then checking what's on the floor ate the whole selling day.

What we built

An AI agent that qualifies every incoming liquidation lead, quotes install and removal against real labor and mileage, and matches buyers to the refurbished stock actually in the warehouse.

What it changed

Sales opens a ranked shortlist with a price already on it instead of an inbox.

Lead qualification Install & removal quoting Inventory matching
Expand
Qualified leads, priced and matched to stock on hand.

How a build goes

01

First look

Free

Tell us what's eating your week, then we sit down — your shop if we can drive it — with a proof of concept already running. Built from what you gave us and digging of our own. The shape of the fix, not the full logic yet, and we'll say so.

02

Scope and a number

One week

Written down: what it does, what it won't do, what it costs. The rules the proof of concept skipped get pinned down here. You approve it before we write code.

03

Build, in the open

2–6 weeks

A deliverable every week — something running, not a status update. If something's off, we hear it in week two, not week six.

04

Hand off, stay reachable

Ongoing

The accounts are in your name and the license is yours for good. We're on the phone when something needs changing.