On October 2, 2026, healthcare automation developer AKASA announced the commercial launch of its autonomous mid-cycle AI platform, expanding its machine learning capabilities into acute inpatient medical coding and clinical documentation integrity (CDI). The release targets one of the most operationally demanding administrative functions in hospital administration, addressing clinical documentation processing bottlenecks that have historically slowed acute care hospital billing.
By targeting the mid-cycle—the bridge between direct clinical care delivery and back-end claims submission—AKASA claims its platform addresses the most complex operational bottleneck in acute care and dramatically compresses accounts receivable (A/R) days. The commercial rollout follows an operating period during which AKASA’s processed inpatient volume grew nearly 6x year-over-year, scaling across partner health systems that collectively represent more than $180 billion in aggregate net patient revenue (NPR). Today, AKASA models process encounters representing approximately 1 in every 10 U.S. inpatient hospital discharges.
What Changed in Revenue Cycle Mid-Cycle Operations
Under traditional hospital revenue cycle workflows, inpatient medical coding and CDI require intensive manual review by certified professionals. Human medical coders typically take 30 to 60 minutes per inpatient chart, reflecting the extensive documentation generated during acute admissions, surgical operations, pharmacology reviews, and multi-specialty consultations. Because of high chart volume and labor constraints, hospital accounts often remain unreviewed for three to four days post-discharge while awaiting coder assignment.
With the launch of its autonomous mid-cycle AI system, AKASA states that its platform completes end-to-end autonomous coding of inpatient encounters in under 90 seconds post-discharge. This shifts chart processing from a multi-day administrative queue into an automated process completed shortly after patient discharge.
AKASA claims that its engine stops billing discrepancies at the chart level before claims reach a clearinghouse or payer adjudication engine. By resolving documentation omissions, coding conflicts, and specificity requirements immediately at discharge, the platform aims to prevent post-billing claim edits, denials, and prolonged administrative re-work.
Technical Architecture and Generative Model Tuning
Acute inpatient stays present complex natural language challenges, frequently involving hundreds of pages of unstructured progress notes, laboratory outputs, nursing assessments, and surgical summaries. To address these demands, AKASA’s platform utilizes custom-tuned generative models tailored to each health system’s historical case-mix index, localized clinical criteria, documentation patterns, and specialty nuances.
Rather than applying generalized language models across distinct clinical environments, AKASA’s models are trained and calibrated on the specific documentation behaviors and case mixes of individual health systems. A health system with an advanced regional trauma center, for instance, maintains distinct diagnostic distribution patterns and localized clinical criteria compared to a specialized orthopedic hospital. AKASA’s architecture incorporates these parameters to interpret documentation patterns accurately across varied clinical specialties.
AKASA claims its platform is designed to fully code acute, multi-morbid inpatient stays across all clinical specialties without human intervention. This capability is structured to handle multi-layered clinical conditions where primary and secondary diagnoses interact to determine billing complexity and severity weighting.
Independent Evaluations and Performance Metrics
To measure operational accuracy against standard human benchmarks, the platform underwent third-party, blinded evaluations. These assessments tested inpatient encounters representing 65% to 85% of total hospital inpatient volume, examining high-frequency clinical cohorts as well as complex, multi-diagnosis cases.
The blinded evaluations showed the AI matched or exceeded human coders across key clinical and reimbursement classifications, specifically:
- Medicare Severity Diagnosis Related Group (MS-DRG) Assignment: Accurate assignment of baseline inpatient payment categories determining facility reimbursement.
- Principal Diagnosis Selection: Correct identification of the primary condition established after study to be chiefly responsible for admission.
- Present-on-Admission (POA) Indicator Capture: Comprehensive tagging of conditions existing at the time the order for inpatient admission occurs versus hospital-acquired conditions.
- Quality and Severity-of-Illness (SOI) Documentation Capture: Documentation of secondary diagnoses, comorbidities, and complications impacting risk-adjustment scores and clinical severity ratings.
By matching or exceeding human performance across these specific dimensions, the platform demonstrated the technical capacity to evaluate complex clinical histories without omitting relevant comorbidity indicators.
Sector Context: Surging Investment in Autonomous Mid-Cycle AI
The introduction of AKASA’s inpatient capabilities coincides with an acceleration in capital expenditure focused on autonomous healthcare revenue operations. According to a September 2026 report from Capstone Partners, spending on autonomous coding and billing grew 125% year-over-year to $450 million.
This growth in spending illustrates an industry-wide transition from traditional computer-assisted coding (CAC) tools—which rely on rules-based keyword matching and still require complete human validation—toward fully autonomous systems capable of end-to-end chart completion. As hospital margins face pressure from labor shortages and administrative overhead, mid-cycle automation has emerged as an active area of strategic investment.
Leadership within the sector has emphasized the systemic necessity of automation in maintaining administrative throughput. Commenting on the operational dynamics driving adoption, industry leadership noted: “The incredible demand for healthcare is finally being addressed by advancements in AI. Multiple parts of the healthcare ecosystem will need to scale up, with documentation and coding being critical components,”
Reflecting on the milestone of automating acute inpatient stays, AKASA highlighted the longstanding difficulty of solving mid-cycle workflows: “For years, an autonomous mid-cycle has been a holy grail in our industry. Today, AKASA is making it real.”
Early Adopters and Health System Deployments
Health systems across the United States have begun integrating AKASA’s mid-cycle technologies into their operational environments. Cleveland Clinic has deployed AKASA’s prebill review and is exploring its autonomous mid-cycle solutions to further streamline inpatient workflows.
Similarly, Nebraska Methodist Health System is cited as an early adopter of the platform, leveraging the technology to process acute inpatient documentation. The participation of established health systems reflects a growing willingness among health system leadership to test and adopt generative automation in core financial and compliance workflows.
The broader revenue cycle technology landscape features active engagements and technology touchpoints involving prominent health enterprise partners and tech vendors, including Ensemble Health Partners, Penelope Health, and Oracle Health, alongside industry executives such as Malinka Walaliyadde.
Business and Operational Implications for Acute Care
The ability to compress inpatient chart processing from several days to under 90 seconds introduces concrete operational shifts for hospital finance departments:
Accounts Receivable (A/R) Compression
When accounts wait three to four days post-discharge just to be opened by human coders, gross days revenue outstanding (GDRO) and overall A/R days stretch outward. By reducing coding turnaround time to under 90 seconds post-discharge, facilities can prepare claims for clearinghouse delivery almost immediately upon clinical discharge, significantly shortening cash-flow cycles.
Preemptive Error Mitigation
AKASA claims that stopping billing discrepancies at the chart level prevents clean-claim rejections downstream. Errors in MS-DRG selection or missing POA flags often trigger complex payer denials that take weeks or months to resolve through appeals. Capturing documentation nuances before claims are generated mitigates this administrative friction.
Optimization of Coding Staff
With human medical coders historically spending 30 to 60 minutes per chart, skilled HIM (Health Information Management) professionals have often faced severe volume pressure. Automating chart completion across encounters that account for 65% to 85% of hospital inpatient volume enables organizations to reassign human coding resources toward edge cases, unique surgical specialties, regulatory audits, and secondary clinical reviews.
Limitations, Uncertainties, and What to Watch Next
While the commercial launch addresses inpatient coding and CDI, several limitations and key milestones remain relevant for healthcare administrators evaluating the technology:
Outpatient Encounter Automation
Inpatient stays represent acute, high-value claims, but outpatient facility encounters constitute an immense portion of daily hospital operations. AKASA has indicated that outpatient facility encounter automation is scheduled to “follow shortly,” though specific timeline details for that release have not been finalized.
Third-Party Testing Transparency
Although blinded comparative performance evaluations established that the AI matched or exceeded human coders across MS-DRG selection, POA capture, and severity tracking, the exact names or identities of the independent third parties that conducted these blinded comparative evaluations are not specified in the initial release data.
Ongoing Customization and Governance
Because AKASA relies on custom-tuned generative models aligned with historical case-mix index data, local clinical criteria, and specialized documentation patterns, initial deployment across a new health system requires fine-tuning. Tracking how rapidly these models adapt to shifting clinical protocols, emerging payer requirements, and varied regional health networks will remain an important indicator of scalability as autonomous mid-cycle AI systems expand across the acute care landscape.