Three tiers bracketed by audit volume. Flat monthly fee, no recovery share, no per-claim charge. AKS safe-harbor-aligned for any US pilots. Published in the executed IMA / BAA for Canadian clinics so counsel can sign off without further review.
For solo practitioners and small clinics.
For multi-physician practices and PCN member clinics.
For PCN central offices and multi-clinic networks.
Every claim goes through the auditor before it leaves your desk — not after AHCIP denies it. You save the revenue. You don't share the savings with us.
Temperature 0, seed pinned, model version pinned. Re-run the same input and you get the same findings. The audit trail is hash-chained so a privacy officer can verify the full chain at any time.
We never proxy your claims through our LLM account. You bring your own provider (OpenAI, Anthropic, Gemini, or any Ollama-compatible endpoint) and we call it directly from the auditor. Your data doesn't train anyone else's model.
We don't take referral fees from any EHR or clearinghouse. The auditor runs against the 837P payload regardless of which EMR produced it (Telus PS Suite, OSCAR, Accuro, Med Access, or any FHIR-capable system).
See /security for the full posture and /legal/privacy for the data handling agreement.
Per-claim pricing punishes the clinics that use us most aggressively — the clinics with the highest denial rates that need the most help. Flat-fee pricing means we make the same amount whether you run 100 or 10,000 audits, so the incentive to under-audit is gone. Your biller audits every claim, not just the ones that look risky.
Tiers are 1,000 / 3,000 / 3,000+ audits per month. We count audited claims, not claims you upload — if the auditor runs on a 1,200-claim month on the Solo tier, we'll reach out before the overage to talk about the upgrade. We don't auto-upgrade you to the next tier.
You bring your own LLM API key (OpenAI, Anthropic, Gemini, or any OpenAI-compatible local endpoint via Ollama). The auditor runs against whichever provider you've configured; if your provider is unavailable, the audit fails closed (the biller sees "auditor unavailable, claim not audited") rather than falling back to a weaker model silently.
Temperature 0 + seed pinned in the prompt + model version pinned in the API call. The auditor emits the same JSON output for the same input across runs. Re-run a prior encounter and the findings match byte-for-byte. The how-it-works page has the full reproducibility contract.
OHIP and MSP are on the 2027 roadmap. We sign HIA Information Manager Agreements today for Alberta clinics. PHIPA and HIPAA templates are on request for pilots in Ontario and the US.