1. Tell us about yourself, your background, and what inspired you to focus on AI in Health & Human Services.

 

I’m the founder and CEO of Health Roads. We build purpose-built infrastructure for California’s Medicaid transformation—CalAIM—across revenue cycle, interoperability, and care coordination. My background is in health information exchange: I spent six years on the Alameda County Social Health Information Exchange and today serve as Senior Technical Advisory Director for Sacramento County’s SHIE program. So I came to AI from the data side, not the other way around.

What pulled me in was watching social care organizations carry the same convoluted Medicaid rules their medical-provider peers had decades to master—but with a fraction of the staff and experience to do it. The industry’s reflex is to use AI to eliminate the human. I saw the real need was the opposite: augment the human where the need is genuine. And I saw how poorly equipped large software companies are to do this, because they don’t carry the lived experience of this sector. That’s what set me on a mission to democratize AI for the people actually doing the work.

 

2. Healthcare and human services organizations generate massive amounts of data. Why do so many organizations still struggle to turn that data into actionable intelligence?

The problem was never volume—it’s that the data is fragmented across formats and systems that were never designed to talk to each other. A single member shows up as five different identities across an HL7 feed, an 837 claim, an eligibility file, and a CBO’s intake spreadsheet. Until you resolve that into one trusted record, every downstream analytic is built on sand.

Most organizations skip the unglamorous work—entity resolution, governance, a real golden record—and jump straight to dashboards. That’s exactly why the dashboards don’t drive action.

 

3. In your view, what are the most impactful use cases of AI in Health & Human Services today, and where are you seeing the greatest measurable outcomes?

 

The highest-impact work right now is in claims and revenue cycle intelligence. Pre-submission fraud, abuse, and waste detection catches problems before a claim ever leaves the building. Denial prediction and automated denial analysis turn a reactive, manual grind into something proactive. And extracting structured social-need signals out of unstructured clinical data turns SDOH from a checkbox into something operational.

The common thread is that these are narrow, high-frequency, high-cost workflows where a small accuracy gain compounds fast. On the revenue cycle side we’re seeing collection rates in the high 90s—near-complete—today. I’d add an honest caveat: we’re still early, and as we scale into greater complexity those numbers will be tested. But the direction is clear and the gains are real.

 

4. What are the key takeaways attendees can expect from your session, “AI in Health & Human Services — From Data Noise to Action Intelligence”?

Three things. First, the bottleneck isn’t the AI—it’s the data foundation underneath it, and I’ll show why. Second, domain expertise is the real moat: a general-purpose model doesn’t know what a 277CA rejection means or how an ECM bundle gets billed, and that knowledge is what converts noise into action. Third, this is achievable for organizations of any size—not just large health systems.

 

5. Looking ahead, what AI trends do you believe will have the biggest impact on healthcare, public health, and human services over the next 12–24 months?

 

Two. The hype will keep pushing horizontal, general-purpose AI—but the durable value in Health & Human Services will come from vertically specialized systems that actually understand Medicaid policy, billing rules, and interoperability standards. 

And we’ll move from AI that summarizes to AI that acts: agentic workflows that don’t just flag a denial but draft the appeal, don’t just surface an SDOH signal but route it to a referral. Whoever pairs that with a clean, governed data foundation wins.