ACDIS and AHIMA’s 2026 Query Guidelines Finally Address AI-Written Queries

On September 1, 2026, the Association of Clinical Documentation Integrity Specialists (ACDIS) and the American Health Information Management Association (AHIMA) released the 2026 update to their Guidelines for Achieving a Compliant Query Practice — the reference document that shapes how CDI specialists and coders word physician queries without crossing into leading. For the first time, the guidelines explicitly address queries that originate from technology systems rather than a person, a gap that had left programs deploying AI-assisted query tools without a shared compliance standard to point to.

The distinction matters more than it might sound. Query compliance sits at the center of how a CDI program defends its work: a query that leads a physician toward a specific diagnosis can taint the documentation it produced, and by extension the code, the DRG, and any risk score built on top of it. Until this update, programs experimenting with software that suggests or drafts query language had no industry-recognized rulebook telling them where the line sat.

A draft built for how documentation actually happens now

The update wasn’t rushed. ACDIS released a draft in April 2026 and opened it for public comment; nearly 200 individuals and organizations submitted feedback before the final version went live this week. The organizations frame the revision around three shifts: care delivered across inpatient, outpatient, ambulatory, and professional-fee settings; documentation workflows that increasingly run through software; and query practices that no longer assume a human always drafts the question.

Rana Abuzir, AHIMA’s director of health information practice advancement, put the stakes plainly: query practices need to stay “clear and trustworthy” as more of the documentation process runs through technology. That’s a notable acknowledgment from the two bodies that effectively set the industry standard — they’re not warning programs away from automation, they’re telling them to bring it inside the compliance framework instead of treating it as a workaround.

What actually changed

The 2026 update keeps the guidelines’ long-standing core test — a compliant query presents clinical indicators from the record and asks an open, non-leading question — but extends that test to any query a system generates or drafts on a coder’s or CDI specialist’s behalf. The framework applies the same standard regardless of whether a person or a piece of software produced the first draft of the question.

Provider judgment stays the anchor

The guidelines are explicit that a query, AI-assisted or not, cannot presuppose a diagnosis or suggest an answer. The physician’s independent clinical judgment remains the deciding factor — automation can surface the indicators and propose query language, but it cannot narrow the provider’s options or imply which response the record is “supposed” to support.

Where agentic tools fit — and where they don’t

Read closely, the update is less about banning AI from query workflows and more about defining the boundary: a system can flag a documentation gap, pull the supporting clinical indicators, and draft candidate query language. What it can’t do is finalize or send a query without a qualified human reviewing it against the same non-leading standard every manually written query has always had to meet. That review step is now the compliance line, not the presence of automation itself.

That’s a meaningfully different posture than treating AI-assisted queries as an edge case to be handled by internal policy alone. By writing the boundary into the same guidelines document that governs manually written queries, ACDIS and AHIMA are telling programs that a system-drafted query isn’t a separate category with its own, looser rules — it’s held to the identical non-leading standard, with an added expectation that a qualified person reviews it before it ever reaches the provider.

What CDI and coding teams should do with this now

Programs already piloting AI-assisted query drafting — or evaluating vendors that offer it — have a few concrete items to work through before the guidance settles into practice:

• Re-check any AI-drafted query templates against the updated non-leading standard, not just the prior version’s language.
• Confirm the workflow keeps a documented human review step between a system-drafted query and the one that reaches the provider.
• Update internal query policy and physician-facing documentation to state explicitly that some queries may originate from a technology system.
• Register for ACDIS’s free webinar on September 30, 2026, which offers continuing education credit on the update.

Why this matters beyond the query itself

Query compliance has always been an audit exposure point — a leading query can undermine the diagnosis it helped capture, with downstream effects on DRG assignment, HCC risk scores, and RADV or payer audit defensibility. As more health systems adopt agentic AI in CDI and coding workflows, that exposure doesn’t disappear just because a machine drafted the language; if anything, auditors reviewing a case are now more likely to ask who — or what — wrote the query and whether it was reviewed before it went to the provider. Having a named, dated industry standard to cite is exactly the kind of documentation a compliance program wants on hand when that question comes up.

That review-and-audit-trail requirement is precisely the discipline that Medikode’s automated medical coding platform is built around: pairing AI-assisted documentation and coding suggestions with the traceable human review step that guidelines like this one now require.