Akasa AI review (Oct 2026): autonomous inpatient coding, CDI Optimizer, AI Advisor, and GenAI for revenue cycle

What this Akasa page covers

This page is for an inpatient coding manager, CDI director, or revenue-cycle leader at a hospital or health system. It is not coding, billing, or compliance advice.

Akasa’s 2 October 2026 release says its customers account for more than $180 billion in net patient revenue and roughly 1 in 10 U.S. inpatient discharges, that its autonomous technology is designed to fully code complex inpatient cases across all specialties, and that outpatient facility encounters are coming soon. Akasa says it trains a custom model per health system on that system’s documentation, case mix, and coding decisions. All of those are Akasa’s statements.

Ground rules: the coding manager sets every autonomy threshold and approves each increase after a sample; coders validate Coding Optimizer suggestions; CDI specialists decide every physician query; a named person approves every appeal or auth letter AI Advisor helps draft; and Akasa’s accuracy and speed figures are its claims, not our measurement.

Major features (with how someone uses them)

Examples at a glance: each feature has a concrete example from a real job plus a short Try this you can run this week.

1. Autonomous inpatient coding (primary)

What it is: Akasa’s autonomous inpatient coding reads the complete chart — discharge summaries, operative and progress notes, consults, labs, imaging, medications — and assigns the full code set, including principal diagnosis, MS-DRG, and present-on-admission indicators. Akasa says a blinded third-party evaluation on encounters representing 65% to 85% of inpatient volume found it matched or exceeded expert coders on key accuracy measures, and that coding can finish in under 90 seconds after discharge versus 30–60 minutes for a coder. Those are Akasa’s results, not ours.

“Matched or exceeded” on average still allows individual misses. Compare against bills you already trust before letting any encounter go out uncoded by a person.

Concrete example (inpatient coding manager at a regional health system) A coding manager has Akasa code twenty closed inpatient encounters that already have a final coded bill. Seventeen match on principal diagnosis and MS-DRG. Two differ on a secondary diagnosis that changes severity. One assigns a POA indicator the documentation doesn’t support. She logs the three, asks Akasa to explain each rationale, and keeps the threshold where it is.

Try this

Primary workflow:

Have Akasa code twenty closed inpatient encounters that already have a final coded bill. Have the coding manager compare every code and rationale with the final bill before raising the autonomy threshold.

2. Evidence and rationale behind every code

What it is: Akasa says every code links to the exact chart language that supports it, with the rationale and its confidence level, so “validation takes seconds, and the audit trail is already built” (Akasa). The same evidence view appears in Coding Optimizer and CDI Optimizer.

A quote from the chart proves the words exist, not that they meet coding guidelines. Coders still judge whether the documentation supports the code.

Concrete example (coding auditor) An auditor reviews the evidence links on twenty-five autonomously coded encounters. Twenty-two quotes support the codes. Two quote a provider’s differential, not a confirmed diagnosis. One quotes a copied-forward note from a prior stay. She sends all three back for review and flags copy-forward text for CDI.

Try this

Judge the quote:

Have a coding auditor open the evidence link for every code on twenty-five encounters and flag any quote that is a differential or copied-forward text.

3. Autonomy thresholds, service-line rules, and audit sampling

What it is: Akasa says “you define thresholds, service line rules, payer mix, and audit sampling” and that it “codes only what you’ve approved it to code” (autonomous coding page). Its launch release says Akasa will work with each health system on a deployment plan to scale autonomous volume.

Raising a threshold is a compliance decision, not a software setting. Write down who can change it and what sample size justifies each step.

Concrete example (coding manager at a three-hospital system) A coding manager starts autonomy on two lower-complexity service lines only, with a 20% audit sample. After a month the sample shows two POA indicator errors on one line. She holds that line at the current threshold, keeps the other, and logs who approved each change.

Try this

Raise in steps:

Have the coding manager set autonomy on one or two service lines with a written audit sample, and approve each threshold change in writing after reviewing that sample.

4. Coding Optimizer prebill review

What it is: Coding Optimizer reviews already-coded inpatient encounters before billing for missed diagnoses, quality indicators, and denial or compliance risks, and presents suggestions with justifications and direct quotes for coders to validate and approve. Akasa cites a CMI gain and revenue impact for one $1B NPR health system — its own case figures.

Suggestions that add codes deserve the same scrutiny as suggestions that remove them. Watch for one-way patterns.

Concrete example (coding supervisor) A coding supervisor tracks two weeks of Coding Optimizer suggestions. Coders accept 60%. She notices nearly all accepted suggestions add severity and none remove unsupported codes. She asks compliance to review a sample of accepted additions before the quarterly audit.

Try this

Watch the direction:

Track two weeks of Coding Optimizer suggestions. Have the coding supervisor count additions versus removals and send a sample of accepted additions to compliance.

5. CDI Optimizer documentation gaps

What it is: CDI Optimizer reviews every inpatient encounter post-discharge and prebill, prioritizes likely documentation gaps by DRG and quality impact, and shows the clinical evidence for each query opportunity. CDI specialists validate suggestions, engage providers, and update cases.

A query is a question to a physician, not a request for a particular answer. Keep queries compliant and non-leading.

Concrete example (CDI director at an academic medical center) A CDI director reviews fifteen query opportunities CDI Optimizer ranks highest. Eleven are worth a query. Three are already addressed in a late progress note. One would lead the physician toward a specific diagnosis. Her team rewrites it as an open-ended question.

Try this

Keep queries open:

Have the CDI director review the fifteen highest-ranked CDI Optimizer opportunities and rewrite any leading query as an open-ended question before it goes to a physician.

6. AI Advisor for prior auth, denials, and peer-to-peer prep

What it is: AI Advisor is a GenAI research assistant for non-clinical revenue-cycle staff: it searches the chart, summarizes records, gives cited answers, and helps draft letters for prior authorization, denials, peer-to-peer prep, and clinical validation. Akasa also lists Auth Status and Claim Status automation.

A drafted appeal letter goes to a payer in your organization’s name. A named person reads it against the chart before it is sent.

Concrete example (appeals specialist) An appeals specialist uses AI Advisor to draft eight medical-necessity appeal letters. Six cite the right notes. One cites an imaging report from a different encounter. One misses the payer’s stated denial reason. She fixes both and has the denials lead approve all eight.

Try this

Approve every letter:

Draft eight appeal letters with AI Advisor. Have the denials lead check each citation and the payer’s denial reason against the chart before any letter is sent.

Pricing context (October 2026)

Akasa does not publish a price list as of 19 Oct 2026. For its Prebill Optimization Suite it describes performance-based pricing — “no upfront fees,” and Akasa “only invoices after measurable financial improvement is realized” — and Coding Optimizer mentions “zero-risk pricing.” We found no stated pricing model for autonomous coding or AI Advisor, so treat those as quote-only.

Ask how “measurable financial improvement” is defined and audited, whether autonomous coding is priced per encounter or by outcome, and whether outpatient facility encounters will be priced separately when they arrive.

Uncertainty: We did not test Akasa. Autonomous inpatient coding launched on 2 October 2026; outpatient facility encounters are described as coming soon. Accuracy comparisons, the under-90-second figure, customer counts, and revenue figures are Akasa’s claims. Akasa says its technology is designed to code with no human intervention — whether your organization allows that is a compliance decision for your coding manager and compliance officer.

When it’s useful (honest take)

Useful when you run inpatient coding at a hospital or health system, have a coding backlog or DNFC pressure, and a coding manager will set thresholds and sample output before raising them.

Less useful when you are an ambulatory practice (see eClinicalWorks AI or athenahealth AI), your gap is claim resubmission and payer responses (Waystar AI), or you want coding inside your EHR vendor’s own AI (Epic Systems EHR AI Penny).

Related options (educational)

FAQ

What did Akasa launch in October 2026?

On 2 October 2026 Akasa launched an autonomous AI platform for the mid-cycle covering inpatient medical coding and CDI, adding to its prebill Coding Optimizer and CDI Optimizer. Outpatient facility encounters are described as coming soon.

Does Akasa code without coders?

Akasa says its technology is designed to fully code complex inpatient cases with no human intervention, within thresholds the health system sets. Your coding manager should decide thresholds and keep audit samples.

How much does Akasa cost in October 2026?

There is no public price list. Akasa describes performance-based pricing for its prebill suite (no upfront fees; invoices after measurable financial improvement). Treat autonomous coding and AI Advisor as quote-only.

How do coders check Akasa’s codes?

Akasa links every code to the supporting chart language with a rationale and a confidence level. Coders and auditors should still judge whether the documentation meets coding guidelines.

Who should check Akasa output?

The coding manager for thresholds and samples; coders for Coding Optimizer suggestions; CDI specialists for queries; the denials lead for any AI Advisor appeal or auth letter.

Optional next step (if seats overlap)

If you pay for Akasa, an RCM agent platform, your EHR vendor’s coding AI, and a CDI tool that all review the same encounters, that overlap is worth mapping. Map the overlap with the $19 AI toolkit audit or the $9 AI toolkit spreadsheet. Email Audit $19 or Toolkit spreadsheet $9.

Checkout is live via Stripe. After you pay, we email delivery (spreadsheet) or intake next steps (audit).

Related: athenahealth AI · Epic Systems EHR AI · Waystar AI · Oracle Health AI · Abridge AI · Microsoft Dragon Copilot AI · Suki AI · athenahealth vs Epic Systems EHR · eClinicalWorks AI · Applied Epic AI (insurance AMS — different product) · Microsoft 365 Copilot · ChatGPT for finance · Claude · Three checks before another AI seat · How-to: Set Akasa’s autonomy threshold for inpatient coding, and have the coding manager review every code below it · ops hub