AKASA vs Fathom
Two Autonomous Medical Coding vendors, side by side. Facts from public sources; judgments are ours.
At a glance
Derived from public facts · a rough scale, not a ranking
| AKASA | Fathom | |
|---|---|---|
| Pricing model | Enterprise contract (custom) · Subscription sized by transaction volume | Not published · custom quote based on coding volume |
| Speed to go live | 60-90 days typical; longer multi-facility | 4 to 6 months, EHR integration and validation |
| Automation model | Autonomous agents · GenAI automation with human review | Autonomous agents · autonomous coding, human review fallback |
| Built for | Mid-size groups, Enterprise systems | Enterprise systems, Billing companies |
| Security posture | HITRUST, SOC 2 Type II, HIPAA | HITRUST, SOC 2 Type II, HIPAA |
| Company maturity | 8 yrs (est. 2018) | 10 yrs (est. 2016) |
| Financial backing | $205M · Series B | $61M+ · Series B |
| Named customers | 2 named | 2 named |
| Published results | Specific numbers public | Specific numbers public |
| Documented integrations | 3 listed | 3 listed |
| Third-party validation | None found | None found |
Bottom line
- Pick AKASA if you run a mid-size or large health system, ideally on Epic, and want generative AI working claims, auths, and coding in-house instead of outsourcing staff.
- Pick Fathom if you code high chart volumes and want most encounters coded autonomously, and can fund a multi-month EHR integration.
AKASA
Generative AI for coding and revenue cycle operations
- Founded
- 2018
- HQ
- South San Francisco, CA
- Stage
- Series B
- Raised
- $205M
What it does
- Generative AI medical coding trained on clinical documentation
- Clinical documentation integrity (CDI) review at scale
- Automates prior auth status and claims follow-up work
- LLMs fine-tuned on customer clinical and financial data
- Surfaces missed codes and documentation gaps pre-bill
Where it's strong
- Cleveland Clinic co-developed and is now deploying its GenAI CDI product across all US locations, a rare tier-one clinical validation.
- Deep pockets ($205M raised) and deployment across 650+ hospitals reduce vendor-viability risk.
- Focus on mid-revenue-cycle (coding plus CDI) fits health systems that want one vendor for both.
What buyers should weigh
- The company pivoted from RPA-style automation to generative AI, so ask which product generation you are actually buying.
- Flagship proof points are large academic systems; fit and pricing for smaller hospitals is less proven.
- Last disclosed raise was 2022, so probe current burn and roadmap funding.
Named customers
Cleveland Clinic · Duke University Health System
Integrations
Fathom
High-volume autonomous coding across specialties
- Founded
- 2016
- HQ
- San Francisco, CA
- Stage
- Series B
- Raised
- $61M+
What it does
- Codes encounters autonomously with deep learning and NLP
- Automates 90%+ of coding volume in many deployments
- Covers ED, radiology, primary care, and other specialties
- Routes low-confidence charts to human coders
- Improves HCC/RAF capture for value-based contracts
- Reduces coding cost, denials, and days to bill
Where it's strong
- Highest published automation rates in the autonomous coding market, with customer-verified results like Your Health's 95.5% automation at 98.3% accuracy.
- Epic Toolbox listing and multi-specialty deployment model shorten implementation for health systems.
- Strategic backing from CVS Health Ventures and clinical investors like Cedars-Sinai signals enterprise credibility.
What buyers should weigh
- Narrowly focused on coding, so you still need separate vendors for the rest of the revenue cycle.
- Automation rates vary a lot by specialty and documentation quality; your mix may not hit headline numbers.
- Total disclosed funding is modest relative to peers, worth probing on enterprise support depth.
Named customers
ApolloMD · Your Health
Integrations
Compare against the rest of Autonomous Medical Coding
Deciding between these two?
First Pass tracks Autonomous Medical Coding every week: funding, launches, and what changed since this page was written.