United StatesPublic sector2023 – ongoingCovered California
Automated verification of eligibility documents
Covered California's pilot validated 84% of eligibility documents automatically, up from 28-30% with its old system
Problem
Covered California received about 50,000 documents a month in 56 classifications, and staff checked 71% of them by hand. It could take weeks to find that someone had sent the wrong form or wrong information. That created extra work for staff and for residents.
Approach
In 2023 the agency ran a small pilot of Google Cloud's Document AI with Deloitte, which extracts data from documents with machine learning. The team used the tool out of the box, without extra training, to set a benchmark. A launch was planned for June 2024, ahead of the October open enrollment.
Results
Takeaway
An untrained, off-the-shelf document tool already beat the agency's old automated checks by a wide margin. The agency still had to train it further to reach its own target.
Categories
- Size
- Large enterprise
- Industry
- Public sector
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Predictive analytics and ML
- Adoption
- Off-the-shelf tool or platform
- Evidence
- Vendor case study
Sources
Source summaries
Google Cloud's customer story describes how Covered California piloted Document AI with Deloitte to verify identity and income documents. The pilot validated 80-96% of documents depending on type, 84% on average, against 28-30% for the old system. The agency handles about 50,000 documents a month and aims for more than 95% automated verification. The page carries no date; the Internet Archive first saved it on January 14, 2025.
Full details are in the original sources.