TxDOT Has Saved 22,000 Staff Hours a Year With AI. Here's How.

The Texas Department of Transportation has saved an estimated 22,000 staff hours a year by automating invoice workflows — and that’s one of more than 30 completed or active AI initiatives the agency was running as of January 2026, according to GovTech reporting.

The number makes TxDOT an outlier in state government, and a useful benchmark now that Texas law requires every agency to catch up.

What TxDOT has built

TxDOT has identified more than 200 AI use cases internally, with another 20-plus projects in development. It has deployed Microsoft 365 Copilot to more than 940 staff. Its Enterprise Data Platform connects 51 data sources to feed AI-driven analysis.

Completed deployments include invoice-processing automation (the source of the 22,000-hour figure), onboarding automation, and traffic-incident detection. On deck: a statewide incident-detection rollout, predictive-maintenance expansion, AI-assisted permitting reviews, and mandatory annual AI training starting this year.

The agency’s CIO, Anh Selissen, sits on the state’s Public Sector AI Systems Advisory Board — a sign that TxDOT’s work is feeding broader Texas government strategy rather than sitting siloed in one department.

The governance rule that makes it work

One policy underpins the whole operation: a “Human-Led, AI-Supported” standard that requires human validation of every AI-assisted output before it takes effect. It’s simple enough to state in a sentence and consequential enough to head off the failure mode — automated decisions with no review — that tends to do the most institutional damage.

That principle now has statutory backing. Three laws signed June 20, 2025 and effective September 1, 2025 set new floors for AI governance across all Texas state agencies:

• SB 1964 requires the Texas Department of Information Resources (DIR) to maintain an inventory of state AI systems, develop an ethics code, and mandate disclosure and risk assessments for high-stakes “Heightened Scrutiny AI Systems.”
• HB 2818 creates a dedicated Artificial Intelligence Division within DIR to help agencies modernize legacy systems using generative AI.
• HB 3512 makes AI and cybersecurity training mandatory for state and local government employees, with DIR certifying approved programs.

TxDOT was already operating ahead of those requirements. Most agencies are now building toward them.

The benchmark worth tracking

Our review turned up no other Texas agency with a public AI deployment profile as detailed as TxDOT’s. That gap reflects where most of state government sits right now: the statutory framework is in place, the inventory and training mandates are live, and the AI Division at DIR has its mandate — but the operational depth TxDOT has accumulated takes years to replicate.

The 22,000-hour figure is the concrete number to watch. If even a handful of agencies match TxDOT’s invoice-automation savings in their own back-office workflows, the aggregate effect on state capacity would be hard to ignore. The September 2025 legislative package was built to make that possible; whether it happens is now an execution question, not a legal one.