Orange County, Calif., United Way using AI to cut costs, improve services for those in need
Over the last six years, the Orange County, California, branch of the global nonprofit United Way has undergone a technology modernization effort, with the organization now running a full tech team and managing more than 3,000 social services programs, and their providers. It’s also using artificial intelligence to better tailor services for those in need.
The upgrade efforts, which were jumpstarted in 2020 by the COVID-19 pandemic, have transformed the Southern California’s social service organization from a traditional charity into a more sophisticated technology service provider. With a full technology team capable of application development, robust data analytics work and large language model training, the nonprofit its elevating service delivery for those at risk of homelessness and currently experiencing it.
The organization has spent the last several years also improving how its Homeless Management Information System, or HMIS, shares data. The Orange County United Way’s system, which all care providers that receive federal funds to combat homelessness must use to share data with the U.S. Department of Housing and Urban Development, is used by more than 1,000 social services providers across the state to coordinate care.
The Orange County United Way tech department tapped into the wealth of data local care providers share about those receiving services, and using AI, to predict potential care needs. It’s allowing the organization and its network of care providers to offer preventative care, which one leader of the United Way called the “holy grail” of social services.
Chris Ticknor, chief transformation officer of the Orange County United Way, said before the pandemic and upgrading the HMIS over the last several years, the organization’s tech focus was limited to building and using mobile or web applications for corporate donation campaigns. Ticknor said once the organization began receiving pandemic relief monies to distribute to those in need, it first built an application to help facilitate and administer the emergency financial and rental relief programs, along with a housing navigation app.
“In building that application, we took on the change management and also some of the skills and expertise in technology to go, ‘Oh my goodness, we’re actually doing something pretty innovative,'” Ticknor said.
After expanding on that existing app software, other nearby counties and cities asked for the organization’s help distributing benefits they’d received. The application they built allowed the Orange County United Way to distribute more than $91 million to approximately 10,000 families through 2022. In 2023, the Orange County United Way acquired the 211 OC service, a helpline for connecting people to service or care providers. That acquisition also came with a web and mobile application that allowed users to search for services they needed across dozens of different providers.
“We started investing different technologies into that solution, including a data warehouse that does data movement and also data analytics. … And now we’ve gotten to a state that’s been such a success — we started out with about 31 providers when we inherited 211 OC to today, where we have 100 plus providers using this solution,” Ticknor added.
The organization created a closed loop referral system, in which patients are referred between providers, and referring providers receive feedback on outcomes. The referral system connects to several different programs, databases, 211 OC, the benefits applications and community-based providers associated with the United Way. In addition to improving coordination, the system also reduces the number of platforms or webpages health care and service providers must navigate.
The referral system also taps into the organization’s HMIS platform. Rose Buenaventura, vice president of development operations at the Orange County United Way, said that merging systems and data enables the use of large language models and predictive analytics that enable the group to predict if someone will end up in the HMIS system.
“If somebody is calling, let’s say they need transportation to a hospital because their vehicle broke down. Then, later on they call in to say, ‘Hey, I need utility payment assistance.’ And then let’s say later on, they’re saying, ‘I need rental assistance.’ If those three pattern sets are a determining factor of homelessness, we can now sift through an entire database of who’s actually done those things and determine the likelihood this person will end up homeless, which means we can infuse preventative resources,” Buenaventura said.
AI is also saving the organization money. Buenaventura said the average cost to get someone in housing is about $100,000 annually, while homelessness preventative costs are less than $2,000 per person.
“It’s a lot of information to sift through, so what we’re trying to do is get this AI to basically recognize what this person’s asking about, and then find out if the [community-based organizations] have the capacity, the capability and the programs to best meet that individual’s need,” Ticknor said.
