Agents and AI are on every vendor’s banner these days, Availity’s Sean Barrett observed — yet getting payer data into the workflows those agents run remains the hard part.
During a featured session sponsored by Availity at Becker’s 11th Annual IT + Revenue Cycle Conference: The Future of AI & Digital Health on Sept. 15, Availity’s chief product officer and two leaders from an early-adopter billing company discussed what changed once payer connectivity was exposed as APIs. The panelists were:
- Sean Barrett, chief product officer at Availity
- Becca Goare, MPA, CHBME, senior vice president of operations at MSN Healthcare Solutions
- Matt Dietz, director of business operations at MSN Healthcare Solutions
Note: Quotes have been edited for length and clarity.
1. The starting point was an operational problem, not a technology
Ms. Goare said MSN did not set out to do AI. The company bills roughly 4 million claims a month, and claim status work was consuming staff. “We started with the problem of not enough people, and then AI was the approach that we needed to use to get to the solution,” she said. Agents now check hundreds of claims at a time and return the results, which she said took out about half the work in front of her team.
2. Exposing the network as APIs changes who can build on it
Mr. Barrett said Availity has spent the past year turning portal-only capabilities into APIs, offered through a new platform, Availity Extend, that pairs developer documentation and a sandbox with a Model Context Protocol (MCP) layer for agents. Availity has had a coverage API for seven or eight years, he noted; what was missing until now was documentation and a realistic sandbox. Mr. Dietz, whose team did the builds, said the MCP layer means users no longer need to know what an MCP tool is. MSN built its agents in-house.
3. Closer to the source means fewer hops and far more volume
Mr. Barrett said each hop in the data path adds days to accounts receivable. Ms. Goare said MSN had the technology but not the payer connectivity — its insurance database holds roughly 38,000 active plans, Mr. Dietz noted — and building that connectivity independently would have been too large an undertaking. The change, Mr. Dietz said, is one of scale: an employee who does nothing but check claim status gets through 80 a day; now the team can check 80 in a minute.
4. The first win came from a use case the vendor never imagined
Ms. Goare said MSN’s initial test ran 130 claim status requests against a new client’s backlog and immediately showed the claims were paid but remittances were routing to an old lockbox address — something the team would otherwise have found a week or two later. MSN fixed it in a day, she said. Mr. Dietz cited a second use case: call center reps checking eligibility and claim status during patient calls.
Ms. Goare cautioned: “I wouldn’t confuse the technology with thinking, well, hey, now I have this product, so now I do not need these people.” Mr. Dietz said the human review element has to stay.
The strategic implication
APIs alone are not the differentiator — context is, Mr. Barrett argued; an agent cannot answer a question it lacks the information to answer. Mr. Dietz’s advice: test, fail and keep trying. Mr. Barrett’s: start small, build one solution, keep building.
