L.A. County’s public defender uses AI to improve client management

The Los Angeles County Public Defender’s office recently began using an AI-powered client case management system that officials said will help public defenders better defend their clients.

Amazon Web Services, the architect of the AI tool in the new management and document-processing system, said in a case study published Wednesday it expects the technology will help the 109-year-old office reduce its manual data entry from documents by up to 85%. The new system also makes all of the documents searchable and proactively alerts defenders to filings from law enforcement, courts and the district attorney’s office.

The new system upgrades the office’s previous data management system, which required the office’s 700 public defenders and their staff to manually scan in the paper files it received. That dated system, according to office Chief Information Officer Mohammed Al Rawi, was comprised of 23 legacy systems dating back to the 1960s, and it was mainly used for archival purposes.

“Technology back then was not looked at as a strategic tool that will drive operations and efficiency — it was looked at as a glorified office equipment,” Al Rawi told StateScoop. “But by accumulating data this way, you’re really dehumanizing — unintentionally — the business process. We as public defenders, we don’t represent cases. We represent clients. So we needed a system that is client-centered, not case-centered.”

‘Client-centered’ system

Al Rawi said the first step to obtaining a client-centered system was to consolidate the information stored in those legacy systems onto a single application in the cloud. That included 160 million records that Al Rawi said were “reverse engineered” from case-centric to client-centric — meaning they instead told the story of the person and the case’s impact on the person, not the case or the “story of the calendar.”

But, Al Rawi said, making all of that digitized data usable is where Amazon Web Services came in, offering solutions to two problems. The first was that most of the documents were digitized as PDFs, which Al Rawi said were not searchable or categorizable. Attorneys and legal staff would have to read them to find information.

The second problem was managing a huge number of paper documents received from arresting agencies and the District Attorney’s office — sometimes thousands of documents in a single day. Al Rawi said he considered hiring data-entry staff, which would have required 40 dedicated personnel just to keep up with the workload. However, he said, that would have been too costly.

Al Rawi said he then suggested AI, but the reaction he got was not exactly enthusiastic. He said public defenders were used to paper and were already having to adjust to using digital files in the courtroom. So he settled on a pilot project to explore using machine learning for processing documents and sorting data into the correct client files. That’s when AWS came into the picture.

“So [AWS] put their AI ML Lab — which is the same group that created Alexa, which is pretty exciting — to build that solution for us,” Al Rawi said. “And we started using the ML to train the AI on the hundreds of types of documents, and it started maturing and this is when things started to become clear that this is going to be a good solution.”

AWS developed a custom pipeline that ingests scanned PDFs, classifies useful pages, extracts agency-specific information and allows the Public Defender’s office staff to verify results through a user interface. Then, the ML solution was connected to the office’s client case management system.

“When we did the first presentation to a group of attorneys, you should have seen their faces,” Al Rawi said. “It identified the summary sheet, got the relevant charges and maximum sentences. It went and found witness information and created witness records. … Everyone was leaning forward to the computer with their mouth open looking at this. I was like, this is insane.”

Beyond case management

After three years of work, the new system is helping public defenders assist their clients, Al Rawi said. He described one instance in which a public defender used the system to help a defendant avoid an arrest on a bench warrant after being proactively alerted to the warrant’s entry into the system.

“He knew about his client’s probation violation, because the document was digitized and the digital image was actually processed, and he was alerted,” Al Rawi said. “So he quickly called his client and reached out with his paralegal, and the client to the probation officer, did the progress report show up to court, and the judge was not expecting to see the client there.”

The judge moved the probation matter off the court’s calendar and the client avoided an arrest, he said.

“You hear a lot when you talk to technology, executives about efficiency, and cost saving and environment impact,” Al Rawi. “All of that is amazing. But here, this is way beyond — this is keeping families together. This is avoiding people from falling into the criminal legal system, falling into homelessness and unemployment.”