The work of the development associate is being rebuilt in software
A partner at a residential developer in Kansas City spends his evenings on Replit. He has built himself a development associate.
Every night at 3am it updates and stores rent comps. It scores every census tract in the metro on demand, supply gap, schools, safety, commute and amenities, then renders the result as a heatmap. Click any parcel on the map and his AI writes the pursue-or-pass case.
The whole apparatus runs inside Giraffe.
His name is John McGurk. He's a partner at Flint Residential, the residential arm of Flint Development.
His background is unusual: history, then law, then politics, then development. That breadth shows up in how he sequences projects.
His AI moment arrived in an Uber. A pollster friend mentioned he had a fleet of robots doing work for him, and told McGurk that if he didn't get on the train he'd be left behind.
A week later he was building.
Rent comp ingestion. Sales comp ingestion. Site lookup. Census scoring. He tried building massing too — in his words, an "utter failure."
He evaluated TestFit. He kept building. Then he found Giraffe.
What changed wasn't that he found another tool. It was that he found somewhere for the tools he'd already built to work together.
Today, market mode ranks 248 Kansas City census tracts.
Deal mode lets him click a parcel and get spatial intelligence, rent comps, massing, and an AI-written pursue-or-pass case in one place. The proforma is next.
He is not an engineer.
This is a partner at a mid-market developer, working evenings, rebuilding the work of the development associate in software.
We sat down with John to talk about what he's building, why he thinks developers are underestimating AI, and what happens when one person suddenly gets leverage that used to require an analyst, a GIS consultant, and weeks of work.
Watch below.
Postscript A
John calls something out which is worth noting.
Before spending real money, he goes to neighbourhood associations with cheap colour-block massings and asks what they think. As a lawyer he watched the opposite play out: developers arriving with
His version flips it.
Massing first. Conversation second. Capital third.
We made people that would have been our enemies into our friends. We actually got letters of endorsement from them.
This is also why Giraffe is intentionally simple.
The job of early-stage software is not to impress people with complexity. It is to make engagement cheap, iteration fast, and conversations happen before positions harden.
Postscript B
John also called Giraffe's chat AI "the best I've used."
That one belongs to our customer success team.
They've spent an enormous amount of time training it on real workflows, real edge cases, and real user problems so people can actually get unstuck and build things.