athenahealth Express Coding: Autonomous Ambulatory Coding Arrives

For years, autonomous coding progress in healthcare has been lopsided. Inpatient DRG coding got most of the early investment and attention. Ambulatory practices — the high-volume, lower-margin settings where the majority of US healthcare encounters occur — largely kept waiting. On June 3, 2026, athenahealth moved that conversation forward with the announcement of Express Coding: an AI coder built directly into athenaOne that automates medical coding for ambulatory encounters, with broad availability set for July 2026.

What athenahealth Announced

athenahealth’s June 3 announcement covered more than 80 AI features across athenaOne, but the coding-specific headline was Express Coding. Currently in beta with more than 500 clinicians, Express Coding automates more than 51% of the beta group’s charges and fully automates coding for nearly one-third of their claims — meaning those claims flow directly to billing with zero coder review required.

The company was explicit that this is not computer-assisted coding (CAC). In athenahealth’s words, Express Coding is “an AI coder designed to deliver consistent, reliable coding support that can match, and in some cases exceed, the performance of human coders.” That distinction matters. CAC surfaces suggested codes for a human to review and approve. Express Coding assigns codes and routes claims without waiting for that confirmation step — on eligible charges.

Alongside Express Coding, athenahealth reported that its existing Denial Resolution Automation tool drives a 30% increase in recovered payments on coding-related denials compared to manual corrections, while insurance-related denials are down 16% across accounts where the feature is deployed.

Why Ambulatory Has Been the Harder Problem

Inpatient DRG coding was automated first for a structural reason: the case volume per provider is lower, the documentation format is more standardized, and each encounter resolves to a principal diagnosis that drives most payment logic. The surface area for automation is manageable.

Ambulatory is genuinely harder. A busy primary care practice generates 40 to 60 encounters per day per provider. Each encounter requires selecting the correct E/M level based on medical decision-making complexity or documented time, attaching diagnosis codes with the appropriate specificity, applying modifier logic, and navigating payer-specific bundling rules that vary by contract. Specialty practices add procedure codes, CPT add-ons, bilateral modifiers, and site-of-service requirements on top of that baseline.

The volume is high, the margin per encounter is thin, and the rule complexity is substantial. That is why most ambulatory automation has historically stopped at “computer-assisted” — surfacing suggestions for coders to accept or override. Express Coding’s beta data suggests the technology has crossed a meaningful threshold: AI can now handle a material share of ambulatory charges without the confirmation step.

How EHR-Native Coding Differs from Bolt-On Platforms

Most medical coding AI today operates outside the EHR. A separate platform ingests clinical documentation via HL7 feed or API, runs its models, and returns suggested codes that surface inside the EHR’s coding workflow. Coders often end up working across two systems — one for documentation review and one for code selection — with the friction and latency that implies.

EHR-native coding is structurally different. The coding model runs within the same environment where clinical documentation lives. Express Coding has access to the full athenaOne chart context — problem lists, medication history, previous encounter patterns, payer-specific rules already captured in the system — rather than only the discrete fields that a third-party HL7 feed delivers. athenahealth’s models have also been trained on its own large ambulatory population, which means the system has been exposed to the payer rules and documentation patterns that its specific customers produce.

The operational benefit is reduced friction. There is no data latency from a third-party integration. Code assignment and billing queue management happen in the same workflow a practice already uses. When a claim is queried or audited, the documentation and code assignment live in the same record. And because the system is continuously learning from the outcomes of athenaOne’s broad network, model performance improves across the entire installed base as it matures.

What the Beta Numbers Mean for Coding Teams

The specific figures from athenahealth’s June 3 announcement are worth breaking down:

• 51%+ charge automation rate: More than half of the beta group’s charges are coded by AI without human review
• ~1/3 of claims fully automated: These flow directly to billing with no coder touch at any point
• 30% improvement in recovered payments on coding-related denials via AI-generated claim corrections
• 16% reduction in insurance-related denials across deployed accounts
• Broad availability July 2026 for athenaOne customers

A 51% automation rate does not mean coders lose half their work. It means coders can redirect their attention to the 49% of charges that require human judgment: complex multi-system encounters, new patient workups, visits with incomplete or ambiguous documentation, or charges where the AI’s confidence falls below threshold. That reallocation — autonomous handling of the routine, human review on the complex — is the model most high-performing RCM organizations are targeting in 2026.

For coding managers, the more important figure may be the claim automation rate, not just the charge automation rate. When one-third of claims reach billing without any coder involvement, that compresses coding lag and reduces unbilled days outstanding — one of the highest-impact metrics in ambulatory revenue cycle performance.

What Coders Need to Watch Going into July 2026

Express Coding’s broad release is weeks away. For coding and RCM teams at athenaOne practices, preparation should start now on a few fronts.

First, audit trail visibility. AI-assigned codes must be auditable to meet compliance standards. Teams should confirm that Express Coding’s documentation of code assignment rationale is sufficient for payer audit purposes, particularly for high-specificity diagnoses, E/M levels at 99215 or above, and encounters with multiple comorbidities.

Second, payer-level performance monitoring. An AI model that performs at 51% automation on average may perform quite differently across specific commercial contracts. Identifying which payers generate the highest post-automation denial rates will reveal where human review still adds the most value — and which payer rules may need to be flagged for model refinement.

Third, workflow redesign for freed capacity. Coders who are no longer reviewing routine charges need an updated job scope. The highest-value transition is toward auditing AI-coded claims, training the system through feedback loops, and handling the clinically complex cases that require judgment beyond what any model currently delivers reliably. Practices that treat this purely as a headcount reduction opportunity tend to realize less revenue benefit than those that reinvest coder capacity in quality assurance and exception management.

Finally, CDI tracking. If EHR-native coding is drawing on richer chart context — problem lists, prior codes, medication signals — it may reduce the volume of documentation queries issued to clinicians. Practices should track query rates before and after implementation to see whether autonomous coding is driving documentation quality upstream, which is the more durable revenue integrity benefit.

Autonomous Ambulatory Coding Is No Longer a Future State

The athenahealth Express Coding beta shows that for ambulatory practices on a major EHR platform, autonomous coding is already live and producing measurable results. The July 2026 general availability will extend that to thousands of practices. The industry question is no longer whether AI can handle ambulatory coding at scale — it is how organizations build the governance, audit infrastructure, and coder workflows that allow autonomous systems to operate safely and improve continuously.

If your team is preparing for the shift to autonomous coding, Medikode’s automated medical coding platform is designed to support exactly this transition — combining AI-driven code assignment with the compliance visibility and audit trails that practices need as autonomous coding becomes the standard of care for the revenue cycle.