OpenEvidence Just Built AI Coding Into the Clinical Note. Here's What It Means for RCM.

OpenEvidence—used by hundreds of thousands of U.S. physicians—just launched Coding Intelligence™: automatic ICD-10 diagnoses, E/M level recommendations with MDM rationale, and CPT suggestions, all generated the moment a note is finished. No extra steps. No separate workflow. It's embedded in documentation.

What Launched

On March 26, 2026, OpenEvidence announced Coding Intelligence™ as a live feature inside its Visits product. Here's what it actually does:

Free Daily RCM Intelligence

Denial trends, payer policy moves, vendor intel — delivered every morning. Free.

Why This Is a Material RCM Development

The traditional coding workflow runs downstream of clinical documentation. A physician finishes a note, a coder—internal or outsourced—reviews it, assigns codes, and the claim goes out. Lag time, human variation, and undercoding are endemic to that model. Studies consistently show physician-led practices leave 10–15% of reimbursable complexity on the table through undercoding alone.

What OpenEvidence is doing is collapsing the documentation-to-coding gap to near-zero. The coding intelligence runs at note completion—not the next day, not after a coder queue review, not during a periodic audit. It runs when the clinical detail is freshest and most complete, and it writes the MDM rationale directly into the note before the physician closes the encounter.

That's a meaningful architecture shift. The coding isn't happening after the fact. It's happening as part of documentation.

The MDM Problem This Solves

Medical decision-making documentation has been the persistent weak point of E/M coding since the 2021 AMA guidelines overhaul. Physicians understand the clinical complexity of what they did. Articulating it in code-supporting language—in real time, at the end of a long clinic day—is where the breakdown happens. Generating that rationale automatically from the note removes the translation burden entirely.

The Implications for Revenue Cycle Teams

This creates both opportunity and adjustment requirements for RCM operations:

Coding quality at the source improves

If physicians are submitting claims with AI-generated code suggestions that are already sequenced and rationale-supported, the back-end coder's job shifts. Less time correcting undercoding, more time on complex exception cases and payer-specific rule compliance. That's a better use of a skilled coder. The volume of routine rework should decrease; the complexity of what remains will increase.

Audit and compliance exposure changes

Automatically generated MDM rationale that is written directly into the note creates a documentation trail that didn't previously exist for many practices. That's good for audit defense. It's also a new surface area: if the AI rationale is consistently formulaic or doesn't reflect individual encounter nuance, RAC auditors and payers will eventually develop pattern recognition for it. Compliance teams need to understand what AI-generated documentation looks like in the record and audit it accordingly.

Denial patterns may shift

Undercoding denials are less common than overcoding-related takebacks, but missed procedure codes generate their own revenue leakage. As AI surfaces uncommon CPT codes that were previously missed by habit-based coding, expect some friction with payers who haven't seen those codes billed at the same frequency from a given practice. Front-end authorization requirements for flagged codes need to stay current.

Vendor consolidation pressure increases

OpenEvidence is a free tool with hundreds of thousands of verified physician users. It doesn't charge for Coding Intelligence. This is a strategic land-grab—put coding intelligence in the hands of the clinical decision-making tool physicians already use, and make it free, so the workflow dependency deepens before any monetization conversation begins. Mid-market coding software vendors who charge for CPT suggestion tools should be watching this carefully. The competitive ground is shifting.

The Broader Pattern

This is the third major category of AI-embedded coding to reach the market in the past 18 months—after ambient documentation tools like Nuance DAX and Suki AI, and after scribing tools that clean up physician-dictated notes. Each wave moves the intervention point earlier in the revenue cycle. Ambient capture moves coding to real-time. OpenEvidence moves it to note-completion. The trajectory is toward zero-lag coding that requires no separate downstream workflow.

That trajectory has real consequences for how RCM departments are staffed and how coding contracts are structured. Organizations that are still pricing their coding operations as if human review of every claim is the default should be stress-testing that assumption now, not in three years when the workflow has already moved.

The question for billing operations isn't whether AI coding tools work. It's whether your workflows are built to extract the value from them—and whether your compliance infrastructure is keeping pace with the documentation those tools are generating.