Navina
The deepest patient context we have audited — 360M documents reconciled into one portrait — running on an engine that only ever adds diagnosis codes, in the year the DOJ priced add-only chart review at $117.7M.
Element scores
Strengths
Navina wins on context, and context is the hard part of clinical AI. It reconciles claims, health-information-exchange feeds and unstructured EHR text into one longitudinal patient view, and its own pilot showed why that matters: consolidated documentation scored 94.8 against 80.1 for ambient-only capture, with a 42.5-point gap on completeness. Adoption evidence is unusually solid for this category — 86% weekly active providers, 20,000+ clinicians, and a Summit Health rollout that onboarded more than 1,000 providers, coders and quality staff in seven weeks. The Phyx Primary Care study across 120 PCPs at 19 practices reported a 40% drop in chart-review time on complex visits, 32% less burnout, RAF up 0.153 and 92% of physicians saying they trust the HCC suggestions. Every suggestion carries its source document, which is the single most useful design decision in the product: a clinician can check the evidence in one click, and 73–84% of suggestions get accepted. It won #1 Best in KLAS for Clinician Digital Workflow in 2025 at 94.6, and integrates properly — SMART-on-FHIR on Epic, bidirectional on athenahealth, plus eClinicalWorks, NextGen and Elation.
Honest dings
Navina surfaces conditions and adds codes. Nowhere in its public material does it describe surfacing codes that should be removed — and in March 2026 the DOJ settled a $117.7M False Claims Act case in which $87.2M turned on exactly that: an add-only chart-review programme that submitted codes and failed to delete unsupported ones. A certified coder on the AAPC forum describes the failure mode in the wild: "a lot of unnecessary codes being accepted by our providers that don't need to be in the chart," naming E66.812/813 companion codes documented nowhere — the same obesity code family that appeared in that settlement. Add the January 2026 Senate Judiciary staff report naming AI risk-adjustment tools directly, OIG's February 2026 guidance flagging add-only chart review as a suspect practice, and CMS running concurrent RADV audits from PY2020 forward, and a one-directional engine in a two-directional enforcement climate is a real exposure — sharpened by the incentive shape, since Navina bills PMPM while advertising RAF lift. Transparency is the other half. The site says "proprietary AI" and "600+ AI algorithms"; the only model ever named is Claude Sonnet 4, disclosed inside a preprint PDF. There are no model cards, no published evals, no bias-audit results, no trust portal, no HITRUST, no CHAI membership, and the newest dated SOC 2 and ISO 27001 announcements are from 2024 while the footer badges carry no dates. The homepage still displays a "Best in KLAS" badge without a year: that award is 2025, and the 2026 Clinician Digital Workflow title went to TransformativeMed. Buyers on KLAS report missing medication adherence in the quality module, support friction across the Israel time zone, customer-success handoffs, an implementation where the vendor "seemed to be figuring things out as they went," instability, missed diagnoses, and evidence linked to the wrong document with no way to unlink it. There is no developer surface at all — no SDK, no docs site, no MCP.
Sources (20) — every claim traceable
Every audit lists the research it rests on — transparency and traceability are the product. Tools evolve: each audit is a snapshot of its audit date, and re-audits supersede older versions (kept below for reference).
- navina.ai — Official homepage: exact tagline, 20,000+ clinicians, 86% weekly active providers, 3.5M+ lives, 600+ AI algorithms, undated Best in KLAS badge (accessed 2026-08-28)
- navina.ai/core-technology — Official: five-stage architecture, Patient Portrait, suspected-conditions engine, 360M documents ingested, AI-ethics posture, multimodal ingestion including audio (accessed 2026-08-28)
- navina.ai/products/clinician-copilot — Official Copilot scope: 2-minute chart review, own ambient scribe with MEAT/E-M support, one-click orders, referrals and patient letters (accessed 2026-08-28)
- navina.ai/products/risk-adjustment — Official risk-adjustment scope: suspected conditions from claims plus HIE plus unstructured EHR, evidence-linked HCCs, 77% acceptance and 43% newly-identified-condition claims (accessed 2026-08-28)
- medrxiv.org/content/10.1101/2025.11.07.25339620v1 — medRxiv preprint 09/11/2025, n=354 encounters: consolidated 94.8 vs ambient-only 80.1 composite, 42.5-point completeness gap; 8 of 10 authors are Navina employees; LLM-run scoring with 20% physician validation; full PDF names Claude Sonnet 4 as the note-generation model (accessed 2026-08-28)
- fiercehealthcare.com/ai-and-machine-learning/ai-scr… — Independent coverage of the study, 25/11/2025, explicitly noting it has not been peer reviewed (accessed 2026-08-28)
- hitconsultant.net/2025/05/07/ai-assistant-shows-pro… — Non-sponsored coverage of the Phyx Primary Care study: 120 PCPs across 19 practices, 40% less chart-review time, 32% burnout reduction, 92% trust in HCC suggestions (accessed 2026-08-28)
- klasresearch.com/comments/navina/252868 — Verified KLAS buyer comments Aug 2025–Aug 2026: PMPM on attributed population, missing medication adherence, Israel support lag, CSM handoff, Chrome-extension vulnerability concern, instability, missed diagnoses, wrong document links with no unlink (accessed 2026-08-28)
- klasresearch.com/best-in-klas-ranking/clinician-dig… — 2026 Best in KLAS Clinician Digital Workflow winner is TransformativeMed, not Navina — the badge still shown on navina.ai is the 2025 award (accessed 2026-08-28)
- navina.ai/news/navinas-ai-copilot-ranks-1-in-2025-b… — Official announcement of the 2025 Best in KLAS win at 94.6, second consecutive year of KLAS honors (accessed 2026-08-28)
- aapc.com/discuss/threads/navina-ai-tool-for-capturi… — Certified coder reporting unsupported companion codes (E66.812/813) surfaced by Navina and accepted by providers with no documentation behind them (accessed 2026-08-28)
- raapidinc.com/blogs/radv-audit-guidelines — DOJ False Claims Act settlement 11/03/2026, $117.7M total, of which $87.2M turned on an add-only chart-review programme that submitted codes without deleting unsupported ones, plus $30.5M on false morbid-obesity codes (accessed 2026-08-28)
- americanbar.org/groups/health_law/resources/health-… — Senate Judiciary majority staff report 12/01/2026 naming potential misuse of AI tools in risk adjustment; CMS RADV audit schedule from PY2020 onward announced 04/03/2026 (accessed 2026-08-28)
- doctorsmanagement.com/blog/medicare-advantage-compl… — OIG February 2026 guidance naming add-only chart reviews as a suspect practice; warning that coding intensification must follow documentation rather than financial targets (accessed 2026-08-28)
- navina.ai/news/navina-and-summit-health-partner-to-… — Official 17/03/2026: Summit Health implementation completed in seven weeks, 1,000+ providers, coders and quality staff onboarded (accessed 2026-08-28)
- prnewswire.com/news-releases/clinical-ai-leader-nav… — Official Series C, $55M led by Goldman Sachs Alternatives, 25/03/2025; $100M raised across four rounds; 10,000 clinicians and 3M patients at that date (accessed 2026-08-28)
- navina.ai/news/successful-soc-2-type-ii-hipaa-audit — Most recent dated SOC 2 Type II and HIPAA audit announcement, 14/03/2024 — no later dated certification announcement exists on the site (accessed 2026-08-28)
- prnewswire.com/news-releases/nabla-and-navina-annou… — Nabla ambient-AI partnership, 09/07/2025: Navina's engine processes ambient conversation data alongside historical records (accessed 2026-08-28)
- rfp.wiki/specialty-industries/healthcare-life-scien… — Analyst aggregator, last verified 20/07/2026: no list price, billing unit undisclosed, one G2 review limiting peer triangulation, encounter submission and retrospective chart retrieval outside the public product footprint (accessed 2026-08-28)
- navina.ai/about-us — Company facts: founded 2018 by Ronen Lavi (CEO, ex-Unit 8200) and Shay Perera (CTO), leadership and advisory board including Bruce Broussard and former AAFP CEO Douglas Henley (accessed 2026-08-28)