Airports test patience daily. Long lines. Confusing rules. Questions that pile up. The Transportation Security Administration now turns to an artificial intelligence agent named Ace to ease some of that friction.
Built by Salesforce on its Government Cloud, Ace went live during the peak summer travel period. It fields routine inquiries about liquids in carry-ons, medical devices at checkpoints, and what to expect during screening. Results came fast. The agent manages roughly 100,000 conversations each month. It resolves 96% of them without handing off to a human. One hundred thousand. That’s scale.
But numbers alone don’t capture the shift. Nearly three million people move through U.S. airports every day. A simple question shouldn’t derail the entire experience. Kendall Collins, CEO of Missionforce and Government Cloud at Salesforce, put it plainly. “Nearly three million people travel through U.S. airports every day, and a simple question shouldn’t stand between them and a smooth trip.”
The Mechanics Behind Ace
Ace draws from the TSA website and more than 550 knowledge articles. Travelers reach it through tsa.gov or the myTSA app. Basic queries receive direct answers. Complex ones transfer to live staff, complete with an AI-generated summary that speeds up the handoff. Paul Tatum, executive vice president of global public sector solutions at Salesforce, described the process to Nextgov/FCW. The system pulls accurate information on rules, such as whether liquids are permitted at specific airports like Dulles.
This isn’t a standalone chatbot. The TSA pursued a phased modernization under its Customer Experience Branch. First came two foundational applications. They unified scattered data, automated manual workflows, and built a central repository of traveler interactions. Ace sits on top of that structure, running inside a FedRAMP-authorized environment with added security from Salesforce Shield. The approach avoids the risks of bolting AI onto outdated systems.
Cost figures tell part of the story. Consumer interactions once exceeded $4 each. Ace cut those expenses by over 90% while holding service quality steady. Broader applications in the modernization effort project $11 million in labor savings across three years. Internal data-sharing transactions that used to take 10 minutes apiece now run automatically. The agency expects to reclaim more than 11,660 staff hours annually. Those hours shift to security priorities instead of repetitive admin work. Previously, excess workloads demanded the equivalent of 12 full-time employees.
Tatum emphasized the larger picture in the Salesforce announcement. “The TSA is leading with trusted AI to meet the mission. They are delivering a secure, data-driven, and AI-first foundation that scales operations, lowers costs, and protects public trust, even during the most stressful travel times.” Collins echoed the point. “TSA’s new Ace AI agent gives travelers fast, accurate answers before they reach the checkpoint — from what they can pack to how they can prepare — while helping TSA officers stay focused on their essential security mission.”
The deployment fits a pattern. Other travel organizations experiment with similar technology. Heathrow Airport in the UK deploys its own Agentforce-powered agent, Hallie, which resolves 90% of queries and trims digital contact times by up to 40%. Private travel platforms like Engine use agents to handle half their chat volume without human input. Yet the TSA case stands out. It operates in a regulated government setting where accuracy, security, and public trust carry extra weight. One error on prohibited items could create real problems. So far, the resolution rate suggests the grounding in official knowledge articles holds.
Recent moves at Salesforce add context. The company unveiled expansions to its Agentforce platform just days before the TSA announcement, including new job-ready agents and deeper integrations with cloud providers. On September 15, Salesforce detailed AIforce, a layer that extends platform capabilities to interfaces beyond its own apps. It also introduced Koa, a reasoning model trained on decades of CRM data and built with NVIDIA. These releases signal broader ambitions. Government deployments like Ace serve as proving grounds for technology that must meet strict compliance standards.
Critics of public-sector AI raise valid concerns. Data privacy. Potential bias in responses. Over-reliance that erodes human oversight. The TSA and Salesforce structured Ace to escalate when needed and to generate summaries that keep humans in the loop. Still, success depends on continuous monitoring. The 4% of cases that require escalation will reveal where the agent falls short. Those interactions should feed back into improvements.
Travelers already benefit. No more digging through dense web pages while rushing to a gate. Answers arrive in plain language before problems compound at the checkpoint. Officers spend less time repeating basic instructions. The hours saved accumulate. Over time, that could translate to shorter lines or more attention on actual threats.
Implementation details matter. Ace didn’t appear overnight. The foundational applications came first. Data unification preceded the agent. This sequence reduced the usual pitfalls of flashy AI pilots that fail in production. Collins highlighted the long-term value for taxpayers and staff alike. The projections on hours and dollars provide concrete metrics in an industry often short on them.
Look ahead. If Ace maintains its performance through holiday peaks and beyond, expect other agencies to study the model. The combination of high volume, regulatory constraints, and measurable outcomes makes it a reference case. For an industry that moves millions daily, even modest gains in efficiency compound quickly. Ace won’t eliminate the stress of modern air travel. But it removes one needless source of it. And that counts.
View Comments (0)