9 production skills covering the full per-case workflow
Evidence-map gate — no unsupported clinical claims ship
Plan-type aware: commercial, MA, Medicaid, ERISA
PHI redaction gate for anonymized case studies

Per-case workflow

Nine skills. One case. One packet out the door.

Each skill owns one step of the denial-appeal and prior-auth workflow. They share a single `case.json`, a single evidence store, and a single per-case audit log. Start a case with a denial letter and a few chart documents — end with a reviewer-ready packet.

MSK specialty overlay (lomn-msk)

Spine / pain / ortho 3 tools

First pilot specialty overlay on top of the LOMN builder. Enforces conservative-care documentation, pain/function score trend analysis, and imaging-to-symptom concordance. Designed against the payer-policy split on spine procedures.

  • Conservative care validator (PT, NSAIDs, ESI, duration)
  • ODI / NDI / VAS score-trend check
  • MRI/CT concordance with reported symptoms

Packet checklist & index

Attachments 2 tools

Required and optional attachment list driven by denial category, specialty, and plan type. Produces the packet manifest and the claim-to-source index bound to the evidence map.

  • PAF-style packet composition by denial type
  • DMEPOS overlay (face-to-face encounter, detailed written order)
  • MA IRE / ERISA claim-file / Medicaid fair-hearing branches

Submission companion sheet

Routing 2 tools

Routing cover sheet — fax vs. portal vs. mail, expedited flag, auth number, dates of service, attachment list. Plan-type aware: commercial, MA IRE / Part D IRE, ERISA voluntary external review, Medicaid MCO vs. state fair-hearing.

  • Pattern-only routing — customer fax table injected at pilot onboarding
  • Expedited vs. standard timeline handling
  • on_behalf_of_member rendering for member appeals

Peer-to-peer prep

Physician brief 3 tools

One-page brief for scheduled peer-to-peer calls: case summary, denial reason, key clinical facts, prior-therapy timeline, likely reviewer questions with suggested answers.

  • Built from the shared case.json — no re-entry
  • Likely-questions generator per denial category
  • Completeness check before the call

PHI redaction

Case-study gate 3 tools

Pre-publication PHI scrubber for anonymized case studies. Scans, applies DAY+n date anchoring, and audits the output against a HIPAA Safe Harbor checklist. The gate you want before posting a teardown.

  • Safe Harbor identifier coverage
  • DAY+n relative date preservation
  • Audit-check blocks accidental PHI leaks

How it compares

Why specialty clinics and RCM firms choose Rakenne

Templates in Word break on payer-specific criteria. Generic chatbots hallucinate clinical content no reviewer will trust. Rakenne runs the packet like an operational workflow — evidence-first, completeness-gated, submission-ready.

Word templates + portals
Generic AI (ChatGPT)
Evidence-backed clinical claims
Word templates + portals Reviewer hand-cites every claim to the chart — slow or skipped
Generic AI (ChatGPT) No linkage to source documents — hallucinations look plausible
Denial-reason rebuttal
Word templates + portals Generic template paragraphs — often rebuts the wrong reason
Generic AI (ChatGPT) Responds to the vibe of the denial, not the specific cited policy
Packet completeness
Word templates + portals Paper checklist; missing attachments caught after mailing
Generic AI (ChatGPT) No structural awareness of what a complete packet needs
Plan-type routing (MA / ERISA / Medicaid)
Word templates + portals Tribal knowledge in a spreadsheet that rots
Generic AI (ChatGPT) No concept of MA IRE, ERISA external review, Medicaid fair hearing
HIPAA posture
Word templates + portals PHI lives in Word files and shared drives
Generic AI (ChatGPT) PHI pasted into a consumer chat product — unacceptable for most buyers
Autonomous submission
Word templates + portals Manual fax/portal every time
Generic AI (ChatGPT) Some tools will submit on your behalf — a liability landmine

See it in action

The evidence map blocks export until every claim is cited

The single most effective trust control for this buyer is refusing to export a letter that contains an unsupported clinical claim. It catches hallucinations and it catches rushed drafting.

A step-therapy appeal, walked through the gate

In this example, the agent drafts a step-therapy appeal, maps each clinical claim to the chart, detects two uncited claims, pulls them from the record, and re-validates before export.

Rakenne Agent
Agent
Appeal letter draft complete for case PA-2026-0412 (step-therapy denial, Cigna commercial, adalimumab).
Agent
Running clinical_claim_extract_candidates on the draft…
Agent

Found 11 clinical claims. 9 cited. 2 uncited:

  • “Patient failed methotrexate 20 mg weekly for 14 weeks with no improvement in DAS28.”
  • “Contraindication to sulfasalazine due to documented G6PD deficiency.”
Agent
Searching chart notes for supporting excerpts…
Agent

Proposed citations:

  • Claim 1 → chart/rheum-2026-02-14.pdf p. 2, span “MTX 20 mg QW × 14 wk, DAS28 unchanged 5.1 → 5.0”
  • Claim 2 → labs/cbc-panel-2025-11-03.pdf p. 1, span “G6PD activity 2.1 U/gHb (low)”
Agent
Running clinical_claim_check_draft again…
Agent
PASS — 11/11 claims cited. Running appeal_completeness_check + contradiction_checker
Agent
PASS — denial-reason rebuttal present, preferred-therapy failure section complete, no contradictions. Export gate clear. Packet ready for your review.

Denial and appeal coverage

Every denial type, every appeal pathway

The skill set covers the denial categories and appeal pathways that account for the bulk of specialty-clinic and RCM workload. Specialty overlays are added as pilots bring real volume.

Denial categories

  • Not medically necessary
  • Experimental / investigational
  • Step therapy / formulary
  • Lack of prior authorization
  • Out-of-network
  • Timely filing
  • Post-service reimbursement dispute
  • Service termination (NOMNC)

Appeal pathways

  • Clinical (medical-necessity) appeal
  • Post-service claim reconsideration
  • Medicare Advantage Level-2 IRE
  • Part D expedited redetermination
  • ERISA external review (IRO)
  • Medicaid fair hearing (state + MCO)
  • BFCC-QIO expedited determination (NOMNC)
  • DMEPOS prior authorization

First pilot specialties

  • Musculoskeletal / spine / pain (lomn-msk shipped)
  • Oncology (next)
  • Rheumatology (next)
  • GI biologics
  • Behavioral health
  • DMEPOS / sleep / respiratory

Reference corpus

  • CMS Integrated Denial Notice (MA)
  • CMS NOMNC / DENC
  • CMS DMEPOS PA process
  • Anthem Provider Clinical Appeal Request
  • UnitedHealthcare claim reconsideration
  • Patient Advocate Foundation appeal guide
  • Genentech LOMN & appeal templates
  • Novartis RHAPSIDO PA + appeals kit
  • HL7 Da Vinci PAS / DTR / CRD

Go deeper

The nine skills in detail

Each skill page includes the canonical conversation, sample output, tool reference, and getting-started instructions. Published and kept in sync with the shipped skills.

FAQ

Common questions from RCM and utilization-management teams

Independent specialty clinics, specialty groups, and the niche RCM and denial-management firms that serve them. Typically 5–50 providers, with a dedicated prior-auth / denials team doing 50+ prior auths or 20+ appeals per week. Hospital systems have longer security cycles and heavier integration asks — not the right first customer for this wedge. If that’s you, we will tell you so.

No. There is no autonomous submission in v1. The product assembles a packet — appeal letter, LOMN, evidence map, packet index, submission cover sheet — and your team submits through your existing fax, portal, or mail channel. Human approval is required before export.

Each project lives in an isolated per-tenant workspace. Prompts, documents, and outputs are processed only to deliver the service and are never used to train foundation models. Every case has an append-only audit log. A dedicated redaction-tool skill gates any anonymized case study before publication. A BAA is available to design-partner pilots.

Not for v1. The workflow is upload-in, export-out. Your team uploads the denial letter, payer criteria, and the chart excerpts you’d attach anyway. The product does not touch the EHR. EHR integration is deferred until the workflow ROI is validated — this is a deliberate scoping choice so we don’t disappear into a six-month integration project.

The core nine-skill workflow is specialty-agnostic. The first shipped specialty overlay is musculoskeletal (lomn-msk) — spine, pain, orthopedics. Oncology and rheumatology are the likely next two, chosen after pilot volume tells us where the denial workload is heaviest. Pilots drive the specialty roster, not a marketing roadmap.

Pilots run $3,000–$6,000 for 30–45 days of live-case work, including workspace setup, one specialty configuration, and founder-led support. We do not run free unlimited pilots. Pricing is deliberately a real operational commitment so we both take it seriously. Post-pilot pricing is a platform fee plus case volume — starting at $1,500/month base (up to 100 cases) for clinics and $3,000–$5,000/month for RCM / service-firm multi-client workspaces.

Per case: time to draft, first-pass evidence-map cleanliness, appeal turnaround, resubmission rate, and — where we can attribute it — overturn rate and recovered revenue. Process metrics come first; outcome metrics become reliable after enough volume. Every metric comes from the per-case audit log, not self-report.

Yes. The RCM / service-firm plan supports multi-client workspaces — one workspace per end-client, with separate case isolation, separate audit logs, and shared routing tables. This is the highest-leverage buyer for the product and the second of the three design-partner segments we are actively recruiting.

Five design-partner pilot slots. One specialty focus. Paid.

We are opening five paid design-partner pilots across independent specialty clinics, RCM firms serving specialty practices, and boutique denial-management services. Pilots run 30–45 days on live backlog cases with founder-led onboarding. If you're doing 20+ appeals a week and you want a real operational tool instead of another template pack — let's talk.

Ready to let your expertise drive the workflow?

Stop wrestling with rigid templates and generic chatbots. Describe your process, let the agent handle the rest.

Get Started Free — No Sign-Up