S12 E19: The Real Estate Draw Nightmare: Automating Underwriting and Risk Mitigation Across $3T in Real Estate Finance Data with Thomas Schlegel, Principal Software Engineer at Built

Episode Summary

Executive Summary: Thomas Schlegel, VP of Engineering and Head of Innovation at Built Technologies, explains how Built digitizes the messy, compliance-heavy money flow behind real estate and construction finance. He traces his path from services engineering to platform building, skunkworks innovation, AI-driven automation, and the tradeoffs of speed, scale, and humility in product development.

Main Topics: Built’s core business and the construction draw process (Priority: 5/5): Thomas explains Built as an end-to-end real estate and construction finance platform, with a focus on digitizing draw requests, compliance, lien releases, and reimbursements that move money through construction projects. Career path into Built and entrepreneurial transition (Priority: 4/5): He recounts meeting Built’s leadership as a customer, later joining the company, and quickly shifting from writing code to solving broader product and organizational problems. Home builder finance as the first major product challenge (Priority: 5/5): The team tackled a complex, niche lending product for large production homebuilders, involving borrowing bases, credit math, and many cascading calculations across thousands of homes. Building platform engineering and the reference architecture (Priority: 5/5): After proving the first product, Thomas founded platform engineering to create reusable architecture, processes, and guardrails so multiple teams could build faster across business units. Skunk Works innovation teams and talent model (Priority: 4/5): He describes his small innovation teams, rotational program, and the traits he values most: customer empathy, business acumen, curiosity, comfort with ambiguity, and speed. Scale, technical debt, and lessons from mistakes (Priority: 4/5): Thomas discusses balancing zero-to-one experimentation with scalability, giving examples of early database mistakes and a Flink-based architecture choice that created operational pain and debt. AI-enabled future of construction finance (Priority: 5/5): He predicts agents will increasingly handle back-office work, making the user experience more convenient while preserving precision and compliance for critical financial workflows.

Key Arguments: Built’s problem is not just software; it is turning a slow, archaic, document-heavy financial workflow into a searchable, automated system that keeps construction moving. The construction draw process is fundamentally about proving work was done and money should flow, which makes compliance and documentation as important as transaction processing. Homebuilder finance is uniquely complex because each home has its own economics, and large builders rely on borrowing bases that require continuous recalculation across many variables. A platform layer was necessary before Built could expand across multiple products and customer segments; product-by-product scaling alone would not have been enough. Innovation should prioritize learning over polish: most ideas will be discarded, and scale should only be optimized after validation. Effective skunkworks teams require humility and low pride of authorship, because the goal is to discover and hand off what works, not defend every prototype. AI will not eliminate the need for accuracy in construction finance, but it can radically improve convenience by letting agents do most of the repetitive work. Leadership at this stage is about inspiring people, solving hard problems, and building strong relationships, not just managing schedules and processes.

Data Points: Initial company size: about 30 employees - Thomas describes Built during the home builder finance project as a much smaller company than today. Revenue scale at the time: a few million dollars in revenues - He contrasts Built’s early stage with its later growth. Home builder finance project timeline: 14 months planned, 6 months delivered - The first major product project was originally scoped to take 14 months but was completed in about 6 months. Team pod size: about 10 people - Thomas describes the office setup during the 2019 era of the home builder finance work. Homebuilder scale: 500 homes at a time - He explains how production homebuilders operate assembly-line style across subdivisions. Borrowing base use case: 10,000 homes a year - Large builders use a large credit facility so they do not need to reborrow from the bank repeatedly. Idea validation window: 30 days - Thomas says skunkworks ideas often get about a month to reach production or be rejected. Rotation cadence: every 6 months - High-performing employees rotate into the Bets team on a six-month sabbatical-like program. Innovation discard rate: about 80% - He says roughly 8 out of 10 ideas are put away or abandoned. Database mistake timing: week one - Thomas recounts accidentally deleting a customer database during his first week. AI productivity example: 5 things in the airport, 2 on his phone - He uses this to illustrate how mobile AI tooling already enables work anywhere.

Pivotal Quotes: "You need to be aware of the scale of the type of thing that you're going to deal with, but you're certainly not optimizing for scale." — Thomas Schlegel: He explains how innovation work should prioritize learning and validation over premature optimization. "It's a very large problem hidden in plain sight." — Thomas Schlegel: He describes the construction draw and payment workflow as an enormous but overlooked industry pain point. "If something works, even if I love it, I've got to give it away because my job is on to the next one." — Thomas Schlegel: He defines the skunkworks mindset and the need to hand successful work off to other teams.

Implications: The episode shows that legacy, compliance-heavy industries are ripe for AI-assisted automation, but only if teams combine domain empathy, platform thinking, and disciplined experimentation. Built’s future likely centers on agentic workflows that reduce manual effort without sacrificing precision.