Field notes on AI governance in community health.
Health centers lose Medicaid revenue in three distinct ways that no single department sees. The discipline that finds it — revenue integrity — exists everywhere in healthcare except the safety net, where it is needed most.
Ron Diver
Founder
Community health organizations operate at negative margins while the revenue they already earned sits uncollected. The invisible margin — denials, eligibility lapses, wraparound variance, and PPS underpayment — is large, structural, and measurable from data your organization already produces.
A practical framework for evaluating AI vendor claims in healthcare — what to demand before signing, and the red flags most community health organizations miss.
Why vendor validation data is necessary but not sufficient for clinical AI — and what local validation actually looks like at community health organizations.
Clinical AI models can show strong aggregate accuracy while systematically failing the patients safety-net providers exist to serve. What CMOs and quality directors need to demand from vendors — and monitor internally.
Existing patient safety frameworks weren't built for AI failures. Most community health organizations don't have an AI incident response plan. Here's what one looks like — and why you need it before something breaks.
Most health centers either have no AI governance or a borrowed template they never use. The real challenge is calibration — knowing which AI deployments need full review and which need a fast lane.
AI creates compliance questions HIPAA never anticipated — PHI in training data, BAA gaps with AI vendors, fabricated clinical content. Here's what your compliance team should be asking right now.
The absence of a single healthcare AI regulation is not the absence of regulation. Requirements are forming across FDA, HIPAA, CMS, ONC, and state legislatures — and organizations without governance now will pay to retrofit it later.
Most health system AI strategies are slide decks full of buzzwords. Here's what a real strategy looks like — and why the hardest part is saying no.
A Nature Medicine study found ChatGPT Health undertriaged 52% of emergencies. But the deeper failure — the one nobody is measuring — is what happens after any AI system makes a clinical recommendation and no one tracks the result.
Everyone says clinicians need to learn AI. Almost nobody specifies what that actually means — and the gap is where careers get stuck.
AI is commoditizing routine clinical work. That's not a threat to expertise — it's the economic force that makes expertise worth more.
The people who will determine what AI does to healthcare careers aren't building algorithms. They're writing billing codes at CMS.
37 questions across five domains. Scored, benchmarked, and followed by a free facilitated debrief with your leadership team.