DebtBook Pushes AI Into US Public Finance to Fight Fraud in a $7.9 Trillion Sector

DebtBook's New AI Layer Targets a $7.9 Trillion Public-Finance Back Office

US public finance may sit on US$7.9 trillion in cash and investments, but much of the day-to-day treasury work behind that figure still runs on disconnected spreadsheets and manual bank-file routines. Tyler Traudt, founder of Charlotte-based software company DebtBook, argues that gap is not just inefficient — it is a fraud risk. Recent cases in Aurora, Illinois, and Arlington, Massachusetts show how easily attackers can redirect public money when account-change procedures depend on a phone call and a trusted employee.

DebtBook's answer is Insights, an AI layer launched on 25 June and integrated into its existing treasury-management software. The tool produces a daily report on cash, debt and investments, flags items such as excess cash or upcoming bond payments, and includes an assistant called Marty for simple queries. DebtBook says it deliberately keeps the underlying calculations in conventional, audited logic rather than letting a large language model generate financial numbers.

The company, founded in 2019 after Traudt left Citigroup's public-finance investment banking group, says it now serves more than 2,100 state and local governments, school districts, hospitals, water utilities and other public-sector organizations. It has raised more than US$22 million, with Boston-based Elephant Partners as lead investor in its Series A rounds. DebtBook is private, so its reported growth and cash-flow claims remain unaudited.

Why DebtBook Is Betting on Fragmented State and Local Treasury Operations

The Fraud Cases Show the System Is the Vulnerability

The April 2026 loss in Aurora — about US$1.1 million after an employee shared account details with a caller posing as a bank representative — and the US$446,000 diverted from Arlington through four consecutive payments to a fake vendor show the weakness is not a single bad actor but a flawed process. The FBI's Internet Crime Complaint Center recorded US$3.05 billion in business email compromise fraud losses in 2025. For public finance teams, the practical problem is fragmentation: one city may spread funds across ten banks and fifty accounts, which makes unauthorized changes harder to spot.

Insights Splits AI Assistance From the Money Math

DebtBook's product decision reflects a real concern among public-sector buyers: an AI model that could produce a plausible but wrong debt or cash figure would be unacceptable under audit. Brendan Mannion, the company's vice president of product, said trust must be paramount. Insights therefore uses AI to retrieve information, prepare summaries and flag attention-worthy items, while the calculations that support those answers run through conventional, auditable logic. This is a measured approach in a sector where even payment transfers have been described as 'click and pray'.

The Aging Public-Finance Workforce Raises the Stakes

Research from the GFOA cited in the story says 60 percent of public finance employees are over 45, and officials list talent retention as a top concern. That demographic reality means manual processes and institutional knowledge are concentrated in a group nearing retirement. For DebtBook and similar vendors, the replacement cycle is not only about fraud prevention but also about capturing workflows before experienced staff leave.

The Unverified Numbers Behind the Growth Story

Traudt estimates the largest 1,000 public entities hold more than US$3 trillion in cash and investments, and DebtBook claims a 30 to 40 percent growth rate and positive cash flow. Because the company is private, none of those figures can be independently verified. The broader Census figure of US$7.9 trillion gives the opportunity real scale, but the company's specific market share and financial performance remain founder estimates until audited disclosures appear.

What Public Finance Leaders and Software Buyers Should Do Next

For public-sector finance teams and software buyers evaluating AI-based treasury tools:

• Map your account and vendor-change exposure first. DebtBook cites cities operating across ten banks and fifty accounts; the Aurora and Arlington frauds succeeded through account and vendor redirection, so identify every point where a phone call or email can change payment instructions.
• Require deterministic, audited logic for any software's cash, debt or investment figures. DebtBook's Insights explicitly avoids letting AI generate financial calculations; make that separation a written requirement in procurement and security reviews.
• Treat vendor growth and market numbers as unaudited unless proven otherwise. DebtBook is private and its 30 to 40 percent growth, positive cash flow and US$3 trillion top-1,000 estimate are founder claims, not audited disclosures.
• Use the 60 percent over-45 workforce statistic as a succession-planning trigger: ask whether your current Excel-based processes can function when senior staff retire, and test any new tool with the next generation of treasury staff before contracting.