What changed
The June 2026 Federal IDR Operations final rule (Federal Register 2026-11140) finalizes new disclosure requirements for group health plans and issuers in the No Surprises Act IDR process, including standardized machine-readable denial/payment reason codes (CARCs/RARCs) in out-of-network remittances [FACT: federalregister.gov 2026-11140]. Simultaneously, sub-35M-parameter OCR (PP-OCRv6) makes cheap self-hosted extraction of scanned EOBs practical [FACT: huggingface.co/blog/PaddlePaddle/pp-ocrv6], and frontier-model cost-performance keeps improving [FACT: openai.com/index/gpt-5-6]. Payer remittances become a stable parsing target for the first time.
Why now
Before this rule, OON denial reasons were inconsistent free-text or payer-specific codes, so IDR-eligibility screening was manual expert work. The rule creates a fresh, standardized data surface with a compliance effective date, meaning every payer's remittances converge on one schema at roughly the same time — a window where a fast solo builder can ship a parser before incumbents retool. HYPOTHESIS: incumbents' denial-management modules will take quarters to fully exploit the new codes.
Converging signals
(1) Regulation: mandatory CARC/RARC disclosure in OON remittances per the IDR final rule [FACT]. (2) AI capability: 1.5M–34.5M-parameter multilingual OCR runs on minimal CPU infrastructure, enabling HIPAA-friendly self-hosted document extraction with no cloud OCR vendor [FACT]. (3) AI economics: better reasoning per dollar makes per-claim classification nearly free [FACT of the claim; performance in this workflow is HYPOTHESIS].
Customer pain
OON-heavy provider groups (emergency medicine, anesthesia, radiology, air ambulance) and the billing companies that serve them must decide, claim by claim, whether a payment is IDR-eligible and then act within strict statutory windows (open negotiation, then IDR initiation) or forfeit the money. Missed deadlines are pure revenue loss. HYPOTHESIS grounded in the rule's structure: the rule exists precisely because this process is high-volume and disputed; the input provided no direct complaint/PAIN evidence, so pain intensity beyond the mandate itself is inferred.
Who pays
Primary: small/mid medical billing companies (RCM firms) serving OON-heavy specialties — they eat the cost of manual triage today and sell recovery as their value. Secondary: independent physician groups' in-house billing managers. These buyers are reachable without enterprise procurement (owner-operated firms, HBMA/AAPC communities). HYPOTHESIS: willingness to pay is inferred from the contingency-fee IDR-services market, not from evidence in this input.
Solved today
Manual review of EOBs/835s by billing staff; outsourcing to specialist NSA/IDR services that typically charge contingency percentages; big RCM suites (Waystar-class) with generic denial-management modules not yet tuned to the new mandatory code semantics. [HYPOTHESIS: from background knowledge; no competitor evidence was provided in input.]
Why current solutions are bad
Manual triage doesn't scale and misses deadlines; contingency vendors take a large slice of recoveries and mainly want big-dollar disputes, leaving small claims unworked; incumbent RCM denial modules are payer-agnostic and slow to adopt a brand-new code mandate. The new standardized codes remove the main technical moat incumbents had (payer-specific mappings).
Proposed product
A web app + watched-folder ingester: drop in 835 files or scanned EOBs → on-device OCR + CARC/RARC parser → per-claim IDR-eligibility classification (state vs federal, NSA-covered service type) → deadline clock per claim (open-negotiation and IDR initiation windows) → auto-generated open-negotiation notice and pre-filled Federal IDR portal submission package. Mirrors the founder's proven ELDT shape: a federal process with a portal, deadlines, and per-transaction monetization — with the caveat (FACT vs shape) that the rule forces PAYERS to disclose codes; the provider buyer is not legally forced to buy, but is money-motivated and deadline-bound.
MVP version
CLI/web tool that ingests standard X12 835 remittance files, extracts the mandated CARC/RARC codes, outputs a ranked 'IDR-eligible, deadline in N days, estimated recoverable $' worklist plus a generated open-negotiation letter. Skip OCR of scanned EOBs in v1 (835s are already machine-readable); add OCR in v2. No portal write-automation in v1 — human files with a pre-assembled packet.
30-day build
Read the final rule and the current Federal IDR portal process end-to-end; build the 835/CARC/RARC parser and eligibility ruleset; validate against 3–5 real remittance samples obtained from 2–3 design-partner billing firms recruited via HBMA/LinkedIn/billing-manager Facebook groups; sign BAAs (self-hosted stack keeps PHI local).
60-day build
Ship the triage worklist + deadline tracker to design partners free; instrument how many eligible-but-unworked dollars it surfaces per firm (that number is the sales pitch); add open-negotiation-notice generation; publish 2–3 SEO/LinkedIn explainers on the new mandatory codes to capture the compliance-date search wave.
90-day revenue plan
Convert design partners to paid: $299–$799/mo per billing firm + $15–$40 per dispute package assembled. Target 5–10 paying billing firms by day 120–150. Expansion: per-filing automation into the Federal IDR portal (the ELDT playbook) once volume justifies it.
Distribution path
Direct demonstrated-value outreach to billing-company owners (send them a free triage report on their own sanitized 835 batch); HBMA/AAPC communities; SEO on the new rule's code requirements timed to the compliance date; partnerships with NSA consultants who want tooling. No ad spend, no marketplace.
Pricing hypothesis
Hybrid: platform fee $299–$799/mo per billing org (unlimited claims triage) + per-dispute-package fee $15–$40. Anchors against contingency vendors taking 10–30% of recoveries — the tool is 10–100x cheaper per recovered dollar. [Contingency norms are HYPOTHESIS.]
Technical difficulty
Moderate. X12 835 parsing is well-trodden; CARC/RARC mapping is table-driven; the hard 10% is correct NSA-coverage/eligibility logic (state vs federal IDR, service categories) and deadline math — exactly the systems-thinking work the founder is good at. OCR path (v2) is proven cheap [FACT: PP-OCRv6]. LLM used only for messy-document normalization, not for the eligibility ruling (keep that deterministic and auditable).
Legal / regulatory risk
Moderate and manageable: PHI handling requires HIPAA compliance and BAAs — self-hosted OCR/parsing keeps PHI off third-party clouds, which the sub-35M OCR specifically enables. The tool advises on process eligibility, not medical or legal judgments; keep outputs as 'eligibility screening' with human confirmation to avoid unauthorized-practice claims. Not FDA/clinical territory.
Platform dependency
Low. Inputs are industry-standard 835 files and the customer's own documents; the Federal IDR portal is a government system (stable, but any future write-automation must respect its terms — same risk profile the founder already manages with the FMCSA TPR).
Founder fit
Strong (8). Matches the proven edge: read a federal rule, find who must act under deadline, build the submission/automation layer, charge per transaction. Honest deviation from the ELDT template: the FORCED party here is the payer (disclosure), while the paying customer (provider/billing firm) acts voluntarily for revenue recovery — demand is deadline-driven and money-driven rather than compliance-compelled. Healthcare billing is a new domain for him, but the buyer (small billing-firm operators) matches his demonstrated-value, no-relationship-sales channel. The accumulated lesson that portal-mandate opportunities fit best (confidence 0.80) applies partially for this reason.
Breakout potential
Good: wedge is OON/IDR triage; expansion is general denial-code intelligence, state-IDR variants, payer-behavior analytics sold back to provider groups, and per-filing portal automation. Each expansion sells to the same buyer.
Final recommendation
PURSUE as a validation sprint, not a full build. The convergence is real and time-boxed: a brand-new standardized data surface, cheap self-hostable extraction, and a deadline-driven buyer with money at stake. It survives the kill attempts because the buyer is reachable without enterprise procurement, adjacent spend plausibly exists (contingency IDR services), and the founder has shipped this exact federal-portal shape before. The two honest risks — payer-not-provider is the forced party, and PHI trust cold-start — are both testable within 60 days for under a few thousand dollars.
Next action
This week: pull the final rule text (2026-11140) and extract the exact CARC/RARC requirements and compliance dates; then message 15 owner-operators of billing companies serving ER/anesthesia/radiology groups offering a free 'IDR-eligible dollars you're leaving on the table' analysis of one sanitized 835 batch. Two yeses = build the MVP; zero yeses in 3 weeks = kill.