Concis Labs' AI Strategy: Why Clean Medicaid Data Comes First

People often ask how Concis produces results that are both accurate and genuinely useful. The answer comes down to one thing: the quality of the data underneath it.

It starts with the data

Concis tracks how every state publicly interprets Medicaid criteria. We record meetings across more than eight types of state Medicaid board, and we gather PDLs, agendas, meeting minutes, and other key documents from all 50 states.

That gives us a record stretching back more than 20 years, and we monitor updates continuously so you don't have to.

Built with domain experts

But raw data is only the starting point. We talk directly with our customers to understand exactly what they need.

To make sure the structure and thinking behind our results hold up, we work with seasoned Medicaid consultants whose industry insight validates that what we produce is accurate and tells the right story. Once we understand a customer's goals, we confirm with some of the top consultants in the field that the output is both correct and complete.

Every answer is cited, not guessed

Every meeting, document, and PDL is indexed and linked to the rest of the information in our system. So when our intelligence layer answers a request, it cites specific meetings — even ones from years ago — rather than leaving a person to guess whether a fact was hallucinated.

We take this seriously. No false information should ever reach an end user.

Human verification is the safeguard

Our safeguard is a human verification step. Medication names, speaker names, rosters, and other details pulled from transcripts and documents are checked and correctly annotated. Anything questionable gets flagged and resolved right away.

This costs us extra time and resources, but it is worth it. Clean data and reliable summaries are the bedrock our intelligence layer depends on.