24 September 2026

For the Builders, Episode 1: How a History Major Built His Own AI Development Associate

Our new series starts with John McGurk of Flint Development, who went from never writing a prompt to running his own AI-powered deal platform on Giraffe in about two months.

▶ Watch the full episode

We're starting For the Builders, a new series about clients who are doing remarkable work, both on Giraffe and outside it. Our first guest is John McGurk, partner at Flint Development in Kansas City. In about two months he went from never writing a prompt to running his own AI-powered deal platform.

A non-linear path into development

John didn't take the usual route into real estate. He studied history and political science at the University of Wisconsin, then went to law school in Kansas City. He found land use law through the ABA's Urban Lawyer journal and went on to practise at Polsinelli, on and off, for 12 years. Along the way he worked on the Obama '08 campaign and spent four years as chief of staff to the Mayor of Kansas City.

He then moved from representing developers to being one. At Milhaus he built the Kansas City office into the company's largest pipeline and ended up running its entire western US operation. When the market cooled, he decided to back himself and joined Flint Development to launch Flint Residential, which he calls "a startup within a startup" with just two people chasing deals.

Lessons from City Hall: talk to the neighbours first

His time in the mayor's office taught him one rule above all: get out early with the neighbourhoods.

"What I like to do is take colored block massings and say, this is what we're thinking. I've spent virtually nothing on this."

He watched too many clients arrive at community meetings with

00,000 of near-final drawings, only to find they'd misread what the neighbourhood wanted. His approach is porch meetings, early conversations and real listening. In one activist neighbourhood, that approach turned likely opponents into people who wrote letters endorsing the project.

The Uber ride that started it all

Going out on his own meant losing Milhaus's in-house design, capital markets, construction and property management teams. John had to do all of it himself. Then, on the way to a charity event, a friend who works as a political pollster told him he had "a bunch of robots that do work for me."

That conversation started John on AI agents (he chose Hermes) and then on vibe coding in Replit. Before long he had built Dev Associate, an app that takes on the busy work a junior development associate would normally do.

Before building much, he got out pen and paper and mapped his whole workflow: sourcing, legal, massing, cost estimation, proformas, pitch decks and OMs. Then he asked three questions. Where could AI help? What could it take off his plate entirely? What should stay human?

What Dev Associate does

  • Knowledge base: bulk-upload broker PDFs and it automatically sorts them into OPEX, rent, vacancy and market data
  • Rent comps: a 3am nightly scrape across 250+ Kansas City multifamily properties, building a rent history he owns
  • Sales comps and cost estimation: historical and current bids, adjusted for construction cost inflation
  • Deal sheets: structured deal data that feeds straight into OMs and IC memos

The rent comp tool took around 8–10 hours to build, in five-minute bursts at his son's baseball games. John says it "saves me days and days and days down the road."

Giraffe as the platform his AI lives on

John first came to Giraffe for quick massing. When we told him we wanted Giraffe to be "the platform your AI lives on," he wasn't sure what we meant. It clicked when he discovered iframes.

He's a visual thinker, so he moved his tools onto the map:

  • A development tracker that pulls KCMO planning cases and links to Compass KC records
  • Market Mode: a census-tract heat map that scores demand, supply gap, feasibility, commute, schools, safety, amenities and employment anchors, then combines them into a composite score. It matched his instincts about Kansas City's growth corridors
  • Deal Mode: click a parcel and Claude runs an analysis in about 30 seconds. John can override its assumptions on zoning, lot size or ownership when he knows better than the data
  • Two-way sync: rent comps he picks on the Giraffe map are written back to his Dev Associate database

"Very rarely am I actually massing in it now."

Why build instead of buy?

The savings matter to him, but the bigger reason is owning the data. It lives on his own server, is backed up daily, and he can use it however he likes. He also knows it inside out. When an investor asks why a tract is red, he can walk them through the 50 data points behind it.

What's next

  • Connecting his rebuilt, input-driven proforma to Giraffe massing and his cost and rent data, so that updating one updates the other
  • Generating OMs automatically from the same data
  • Bringing on a computer science intern to turn the "history major vibe code" into a solid tool for the whole company, across both the residential and industrial sides

John's advice to other developers

"Just take the sledgehammer and put a hole through the wall and say, 'All right, I started. Let's see what we can do.'"

He's clear this isn't secret sauce. He sees it as existential: developers who don't adopt these tools will be "using a rotary phone when people have iPhones."

▶ Watch the full conversation with John McGurk