She used to spend an hour clearing one criminal record. Her AI app does it in three minutes.

The man rolled up to a Salt Lake County expungement clinic with a rolling suitcase stuffed with paper records: years of arrests and court filings he had carried in by hand. Noella Sudbury, the former public defender running the clinic, looked at the pile and said the only sane thing: "This is nuts because all of this could live in a digital format."

That suitcase is the reason Sudbury built Rasa Legal, an AI app that determines whether a person is eligible to have an old arrest or conviction cleared or "expunged," which means legally sealing a case so it no longer surfaces on employer background checks. What used to take a lawyer up to an hour of piecing together paper records now takes the app about three minutes.

According to an NPR feature republished by KAZU, Sudbury started running in-person expungement clinics in Salt Lake County in 2018. Demand was so high that she and her fellow lawyers routinely worked until 9 p.m. and then had to turn people away. "We were hitting a wall," she told NPR. "The lawyers had to go home."

Drug arrests and retail theft are the most common offenses driving record-based hiring rejections, the same NPR reporting found. People whose records are cleared recidivate at lower rates than the general public, NPR added, citing unnamed research; the underlying study is not pinpointed in the available coverage.

Rasa's app works the way an experienced attorney would. It pulls court records, walks the user through the offenses on their record, and checks each one against the state's eligibility rules. If a case qualifies, the app drafts the petition. The whole loop, which used to eat a lawyer's evening, now runs on a phone while the user is still at the clinic table.

The startup says more than 25,000 people have run its eligibility check and that more than 5,000 records have been sealed or expunged with its help. Those numbers are self-reported on the company's homepage, not independently audited, and they should be read as company claims rather than confirmed outcomes. The service is live in three states: Arizona, Pennsylvania, and Utah, with more on the roadmap.

Headlines this year have been full of AI-driven layoffs across entry-level white-collar work, from paralegals to junior accountants to customer support. The same technology reshaping those jobs is also the lever letting a chronically understaffed legal-aid clinic serve more of the people who keep showing up.

Rasa's pitch is not that AI replaces lawyers. Sudbury still runs the clinics, and the company charges for the app and petition help, not for legal advice. The argument is narrower: the bottleneck in clearing an old record was never funding or awareness. It was the hour a lawyer had to spend reconstructing a paper file. Compress that to three minutes, and the bottleneck moves downstream, to clinic capacity and lawyer time, not to intake.

That is why Sudbury named the underlying principle. "There's a presumption of innocence that I think we all hold as Americans," she told NPR. "But in the application of a background check, it's almost like there's a presumption of a record."

Rasa's three-state footprint is small. The company says it is adding states, and Sudbury is still scheduling clinic days. The next test is whether the same per-case time collapse holds up once a second state auditor starts running the numbers.