US Federal News Bureau

How TSA’s AI Agent Resolves 96% Traveler Questions Without Human Help

avatar

Written by: Tathagata Sen

Updated 6:54 AM EDT, September 15, 2026

post detail image

The Transportation Security Administration’s (TSA) AI agent Ace now handles about 100,000 conversations a month and resolves 96% of routine questions without a human stepping in. Salesforce, the organization that built the tool, disclosed those performance figures in a September 14 release, according to a MeriTalk report.

The agent, which went live this summer, is part of TSA’s broader push to modernize customer service and free staff for complex cases. It appears as a live chat widget on TSA’s official website (TSA.gov). Salesforce said Ace draws on a centralized repository of traveler interactions, built from two other TSA applications, to answer questions based on agency guidance. 

TSA’s modernization effort began with those two: the Customer Service Managers App, which brought together traveler questions from email, phone, and social media into one place, and the TSA Cares App, which connects travelers with medical conditions or other needs to support teams.

What the Numbers Show

Salesforce says traveler interactions used to cost more than $4 each. Ace has lowered that cost by more than 90%. Across TSA’s broader modernization work, the agency projects $11 million in savings over three years.

TSA is also using automation for internal work beyond traveler questions. 

Salesforce says automating about 70,000 internal data-sharing transactions each year, work that used to take about 10 minutes each, will save more than 11,660 staff hours annually. 

TSA previously needed the equivalent of 12 full-time employees to handle excess workloads and backlogs, which Ace and related tools are now designed to absorb.

The Real Lesson: Data Comes First

For chief data officers (CDOs), the most useful part of this story is what TSA had to build before Ace could work. Consolidating scattered traveler data into one system came first. The AI layer came after, and it works because the data underneath it was already organized.

That’s the same lesson AI governance frameworks keep pointing to: an AI tool is only as good as the data it can actually reach. TSA’s results, a 96% resolution rate and real cost savings, are a working example of what that groundwork can deliver once it’s done.

Related Stories

September 17, 2026  |  In Person

Chicago Leadership Summit

Renaissance Chicago Downtown Hotel

Similar Topics
Artificial Intelligence
Data Management
Diversity
Testimonials
background imagebackground image
Community Network

Join Our Community

starElevate Your Personal Brand

starShape the Data Leadership Agenda

starBuild a Lasting Network

starExchange Knowledge & Experience

starStay Updated & Future-Ready

logo
Social media icon
Social media icon
Social media icon
Social media icon
About