Automate the work, not the judgment.
After 15 years of building healthcare automation systems, one principle holds above all others. The efficiency case for automation is real and proven. The question is not whether to automate. The question is which steps to automate and which to protect. This article walks through four high-risk automation scenarios (duplicate member merges, prior auth denials, fax transcription, and HCC chart review), the exception queue and confidence threshold architecture that makes automation safe, and five questions every executive should ask before any workflow automation goes live.
Healthcare has data standards. None of them are standard.
We built X12, HL7, FHIR R4, and a dozen other frameworks for healthcare data interoperability. Every major organization agreed to use them. Then each one implemented them differently. A standard that everyone implements differently is not a standard. It is a suggestion with a formal name. This article traces why the companion guide problem persists, what real standardization would actually require, and why the barrier is not technology. It never was.
Claims data is never current. Your value-based contract assumes it is.
Every Monday morning, someone at a health plan or provider group pulls a utilization report and makes decisions from it. Most of them do not know they are looking at a picture of the world from 90 days ago. Claims do not travel in real time. They move through a long pipeline (billing, clearinghouse, adjudication, remittance, data warehouse) and by the time a claim shows up in a report, the world has moved on. This article walks through why the lag exists, what it costs VBC operations in concrete terms, why it hits value-based contracts harder than fee-for-service, the IBNR concept that actuaries account for but most ops teams never do, and how to build a lag-aware VBC operation that makes decisions on real data, not a 90-day-old version of it.
V28 did not break your risk scores. Your documentation did.
V28 went fully live January 1, 2026. Seven months in, most plans and provider groups are still operating with the wrong map. RAF scores shifted, revenue looks different than expected, and the coding team swears nothing changed. Something did change. It just did not happen on their end. This article covers what HCC categories are, how RAF scores are built from them, what specifically changed in V28, and what quality coding and documentation looks like under the new model, in plain language, from first principles.
Your AI Problem Is Probably a Data Problem
Most organizations come to us thinking they have a workflow problem. After digging in, it's almost always a data problem wearing a workflow dress. AI is generating a lot of excitement in healthcare operations right now, and rightfully so. But there's a gap between what AI promises and what it actually delivers when the data feeding it hasn't been addressed first. Before any automation conversation begins, we look at three things: whether the data is accurate, whether it's complete, and whether it's timely. Skipping any one of those questions doesn't slow down the problem. It accelerates it.