Travis Noll - Challenges in Valuation

We sat down with Travis, a tech entrepreneur who built his career at the intersection of Silicon Valley and local government, to explore what it really takes to modernize property appraisal workflows. From being pulled into city council meetings as a kid to building software used by assessors and clerks across North America, Travis shared a grounded look at how public sector innovation actually happens.

Travis' path into the assessment world started the way many civic tech journeys do, with permits and planning. It is the most visible part of local government, so it attracts early attention from founders. But a simple observation opened a much bigger opportunity. Staff were physically driving stacks of paperwork from one office to another, week after week, just to keep processes moving. That question, "Why is this still paper?" led Travis and his co founder to the property appraiser's office and into a niche that had massive operational complexity but far less technology investment than it deserved.

Like most startups, the first big bet did not stick. Travis described an early proof of concept focused on predicting and resolving appeals earlier, something that clearly addressed a real problem. The issue was business fit. Appeals were a sharp pain for only a couple of months each year, which meant customers could forget about the product for the other ten months. The lesson was decisive. If you want to build a durable relationship with a government customer, you have to solve a problem that exists year round. That thinking led them to deeds automation, a daily workflow that is high volume, time consuming, and foundational to everything else the office does.

A big part of the conversation focused on how they learned a highly specialized domain without pretending they knew it on day one. Travis was candid that there is still plenty he does not know, especially on valuation. Starting with deeds helped because it is complex, but it is not valuation complex. The learning loop was simple and relentless. Sit with customers, watch them do the work, ask why, and repeat. Over time, the team built real fluency in how different states handle tenancy, naming conventions, exemptions, and document nuances. They also confronted the reality that land records in the United States can be deeply antiquated, with legal descriptions that read like instructions from another century. The work is still fundamentally about accuracy, even when the source material is messy.

Travis then unpacked how AI fits into these workflows in a practical way. He described deeds processing as a chain: OCR to pull text, natural language processing to interpret it, rule based structuring to match local requirements, and validation steps like address checks. Before modern LLMs, the team used earlier transformer models for parts of this pipeline. Today, LLMs help most where text interpretation and flexible structuring are needed, but not everything is best solved by the same tool. The larger shift is not just better extraction. It is end to end automation that connects back into the system of record. When that integration is strong, a portion of documents can move from intake to completion without manual data entry, freeing staff to focus on review, quality control, and higher value work.

The conversation also addressed a reality assessors know well: exemptions and special programs are where complexity multiplies. Travis described how policy changes often land on local offices that are already operating within software that was never designed for the latest rules. The result is workarounds, manual manipulation, and "patch the holes" processes that are hard to maintain and hard to explain to the public. He also pointed out the education gap. Residents often do not understand tax years, eligibility timing, or why a form in 2025 impacts 2026. That confusion creates inbound volume, frustration, and risk.

While Travis' company does not focus on outbound public education, he shared how they are tackling the inbound side with a product designed to centralize questions and draft responses using AI. The key point is that these responses are not only generated from a generic knowledge base. They can also pull from local data and internal notes, so staff can explain what happened in a specific case, such as a reassessment triggered by a permitted renovation. The workflow is designed to keep humans in control, with AI drafting and staff reviewing before sending.

That idea connects to one of the most important themes of the episode: trust and validation. Both Travis and Will emphasized that AI works best when you treat it like a capable assistant that still needs supervision. Travis described a crawl, walk, run adoption model. Start with full review, then gradually automate the simplest cases as confidence grows. He also highlighted a unique public sector challenge. People are often less tolerant of mistakes made by an automated system than mistakes made by a person, even if the overall error rate is lower. In government, being able to explain what happened matters, and "the AI did it" is not a satisfying answer to most constituents.

When asked for advice to newer teams building in this space, Travis kept it simple. Sit with customers as much as you can. There is no substitute for being shoulder to shoulder, learning the language, and watching the real workflow. He noted that remote work made this harder during Covid, and reconnecting with in person implementation has been a major priority.

Looking forward, Travis described a roadmap driven by customer pull. Deeds was the entry point, but the broader mission is workflow automation for local government. That has already expanded into homestead and compliance programs, communication management, and adjacent offices like clerks and recorders who handle the permanent indexing of documents. The direction is clear: keep going deeper where the need is urgent, and expand carefully into neighboring functions where the same automation principles apply.

The through line of the episode was refreshing: public sector innovation is not about flashy tech demos. It is about learning the messy reality, integrating into legacy systems, reducing tedious work, and earning trust one workflow at a time. If the next decade of assessment technology is going to be meaningful, it will look a lot like what Travis described: practical automation, strong human oversight, and an obsession with the day to day work that keeps local government running.