The national RCM/EOB AI lane is occupied by one credible venture-backed player naming both ABA and SUD, and the ICBD flag has gone quiet

The question

Verbatim: "Beyond the Tampa-Bay-only Form D scan, is there a funded RCM/EOB AI startup nationally that already targets the SUD/behavioral-health/autism-adjacent verticals the family co-founder venture is scoping -- and does ICBD Holdings' Curative AI (flagged in a separate 2026-05 brief) compete directly with that angle?"

Context: [[2026-09-23-tampa-ecosystem-map-v0]] found no funded Tampa RCM/EOB AI startup and explicitly refused to call that proof of absence. This is the national next step, scored against the specific angle the family venture sketched, not against generic healthcare AI.

What we already know (from the vault)

• The Tampa negative was flagged as unproven, not as a clear lane. [[2026-09-23-tampa-ecosystem-map-v0]] section (d) states "This pass found no funded Tampa RCM/EOB AI startup... A search miss is not proof of absence," with scope limited to the 60-day Tampa and Florida-healthcare Form D runs ([[2026-09-23-formd-tampa-60d]]) plus 2025-2026 web research. Nearest local work it found was Greenway Health's RCM line, Tampa General's Palantir deployment, and Moffitt's coding/denial proof of concept with NVIDIA, Oracle and Deloitte.
• The angle being scoped is provider-side EOB/denial automation for independent practices, in the family's own verticals. Per the family-venture thread (project_life_sciences_career_exploration, 2026-09-16): co-founders Michael Holzum (ABA Centers exec) and Dr. Christian Whidden (OMFS resident, ~3 years from finishing), with the twins and Michelle secondary; 3-6 year horizon tied to Christian's residency and to capital accumulation. The founder's own sketch: EOB to stage to PARSE_DOCUMENT to Cortex Search to a Cortex Agent on a continuous trigger, and he called the stack "commoditized."
• The RCM/EOB idea was a diagnostic case study, not a commitment. Same-day correction 2026-09-16: Michael's shuttered play "never even reached a demo," and the founder himself stepped back with "not sure that's the issue we want to tackle together." This brief therefore tests whether a lane is occupied, not whether a decision is live.
• The 2026-05 ICBD brief already called the overlap a competitive one. [[2026-05-03-icbd-holdings-curative-ai-research]] verdict: "Exact Billing Solutions is ICBD's existing RCM company, serving SUD / mental health / autism -- exactly the verticals an AI-rebuilt SMB RCM would target. Curative AI is explicitly building the AI RCM layer. ICBD is not a buyer in this category -- they are a competitor compounding internally." That brief also flagged the $20M-revenue claim as likely intercompany and weakly validated.
• The family network is the demand-side asset, and it is provider-side. user_extended_family_map plus [[2026-07-21-family-medical-careers-scenario]]: ophthalmology clinic, dental, OMFS, ABA/rehab. One payer-side node (Stacey Whidden, dental HMO). This is who a venture here would sell to first.

What the web says

• Camber is the direct hit, and it names both target verticals in its own copy. Its site headlines "RCM Billing Software for Specialty Care" and "Healthcare revenue infrastructure, rebuilt from first principles," and the verticals it names are "ABA," "PT," "ENT and Allergy," "Cardiology," and "SUD" (camber.health). Stated buyers include "growing clinics," "large group practices," and "multi-state providers and PE-backed platforms." Its own page names backers Craft, ACME, a16z and Y Combinator but states no funding amount and no round name (primary source for investor identity; amount unknown). Metrics on its own page: "2,699,889" claims processed to date, "$2.5B+ in live claims data across specialty care," "-7pp" initial denial rate and "+3pp" net collections rate within a year. A secondary content site adds "90,000 patients across 42 states" and a late-2020 launch (acuity.news) -- unconfirmed, not on the company's own page.
• Behave Health is the only company in this scan whose funding I verified from a primary filing, and it sells the exact workflow. Its site headlines "All-in-one Software for Behavioral Health," tagline "CRM, EHR, RCM, and ERP built specifically for behavioral health organizations," and its RCM description names "insurance verification," "claims submission," "ERA/EOB processing," "denial management," and "payment posting" (behavehealth.com). It names SUD settings (inpatient detox, residential, outpatient, recovery residences, drug courts) and mental health; it does not name autism or ABA. Claims "Hundreds of Facilities," "2000+ Recovery Facilities," "50 States Served."
• Behave Health's Form D is real but seven years stale. SEC Form D accession 0001789714-19-000001, filed 2019-10-18: entity "Behave Health Corp.," San Francisco CA, incorporated 2015, industry group "Other Health Care," Rule 506(b), total offering amount $4,432,258, total amount sold $4,432,258, remaining $0, date of first sale 2019-10-16 (SEC primary doc). It is the company's only EDGAR filing of any type (SEC submissions). No later round is visible in EDGAR.
• Incumbent ABA software vendors are adding AI billing modules, which changes the competitive shape more than any startup does. Rethink Behavioral Health (Rethink Autism Inc / RethinkFirst) announced "BillAI," an AI billing system for ABA, in April 2026 (PRNewswire release). Raven Health, an ABA data-collection vendor, was reported in September 2026 to have launched an AI-powered RCM platform for ABA providers (breakingnewsaba.com) -- surfaced via search result only, page not fetched, treat the launch and its scope as unverified.
• A whole services tier markets AI-assisted SUD and ABA billing, with no verified venture funding. Plutus Health, Coronis Health, Nexus io, AnnexMed and ASP RCM Solutions all publish ABA or SUD billing-service pages. These are outsourced billing shops, not software companies, and the pages surfaced are search-engine lead-generation content; no funding claim of theirs was verified.
• The denial-rate premise the venture rests on is widely asserted and thinly sourced. Secondary content pages put SUD claim denials at ~25% and ABA initial denials at 15-30%, "3-5x the medical average" (aihealthfinance.com, rethinkbehavioralhealth.com). Both are vendor marketing. Unverified; no payer or regulator source was reached this pass.

Scored landscape

Company self-descriptions are quoted, not paraphrased. "Source type" refers to the funding evidence only.

Does ICBD Holdings' Curative AI compete directly with this angle?

No -- not on the evidence available as of 2026-10-01. The 2026-05 flag should be downgraded from "competitor compounding internally" to "a vacated internal build in an occupied vertical." Three pieces of evidence, in order of weight:

1. The buyer never matched. Even at its most active, the ICBD RCM stack was captive: Exact Billing Solutions bills ABA Centers' own claims, Curative AI was incubated against that flow, and the 2026-05 brief itself recorded "Customer 1: Almost certainly ABA Centers itself... External customer logos not public." The family venture's buyer is the independent practice. A captive biller and a software vendor selling to third parties can share a vertical and a workflow and still never meet in a deal.
2. The product's public surface is gone. curativeai.com and www.curativeai.com both return HTTP 404 (checked directly, 2026-10-01), while icbd.com and exactbilling.com both return HTTP 200 on the same check. A dead apex with live siblings is a meaningful signal about where the group is putting its web presence. It is not proof of shutdown: migration, rebrand, or folding the product under icbd.com would produce the same 404.
3. An independent signal points the same way. The founder's own channel report (2026-09-16, via Michael Holzum) is that the holding company's RCM AI play was shuttered after "endless requirements gathering and excuses," never reaching a demo. That is a family-channel account, not a public filing, but it is independent of the 404 and agrees with it.

Two corrections to the prior brief follow. First, its exit-thesis logic ("they already own the asset") weakens if the asset was never built. Second, the secondary $9.8M self-funded figure and the earlier $20M-contracts figure cannot both be straightforwardly true; neither is confirmed, and the earlier brief's own skepticism flag on intercompany revenue now looks correct.

The competitive threat in this lane is Camber, not ICBD. That is the substantive revision.

Convergences and contradictions

• Convergence: the Tampa negative survives, and is now better characterized. Nothing in this national scan contradicts [[2026-09-23-tampa-ecosystem-map-v0]]. No direct player surfaced with a verified Tampa HQ. But the Tampa brief's "Ray's read" that "Tampa has room for a local RCM/EOB AI startup" now reads as geographically true and strategically weak: this is software sold nationally, so a Tampa gap is not a market gap.
• Contradiction: the founder's "commoditized stack" read and a16z's behavior point opposite ways. The founder's 2026-09-16 sketch treated parse-plus-retrieval as commodity. Camber's own site shows a company with 2.7M claims processed and a16z/YC/Craft/ACME behind it, in exactly ABA and SUD. Either the moat is in the parts the stack sketch omitted (payer-specific 835 heterogeneity, CARC/RARC rules, PMS integration, trust thresholds) or top-tier investors are funding a commodity. The first is far likelier, and it was the pushback Ray already filed on 2026-09-16.
• Contradiction, unresolved: ICBD's numbers. $9.8M self-funded (secondary, 2026) versus "$20M signed customer contracts for 2025 revenue" (2026-05 brief, LinkedIn-sourced). Both unverified. Do not cite either as fact.

Synthesis for RDCO

What my evidence supports: the lane is occupied at the top by one credible venture-backed entrant. It does not support "crowded" or "contested." The honest count is one direct software competitor with blue-chip investors and no disclosed round size (Camber), one direct SUD/EOB competitor whose last verified capital was $4.4M in 2019 (Behave Health), two installed ABA vendors bolting AI billing onto existing distribution (RethinkBH confirmed, Raven Health unverified), one captive in-group biller that does not sell to this buyer (Exact Billing), and a tier of outsourced billing shops. That is an occupied lane with a clear leader, not a knife fight. The distinction matters because "crowded" would argue against entry while "one leader plus sleepy incumbents" is the normal shape of a market a focused team can still enter.

The specific thing that changed is who the competitor is. The 2026-05 brief pointed the founder at ICBD as the thing compounding against him. That reading no longer holds up: the product site is a 404 and the family channel says the build was shuttered. The thing actually compounding is Camber, and the uncomfortable detail is that Camber's own homepage names ABA and SUD as two of its five verticals and claims $2.5B+ of live claims data. Claims data is the asset that compounds in this category; a 2029-start team begins at zero against a firm that already has it. If the family venture ever revisits RCM/EOB, the honest framing is not "is anyone doing this" but "what do we have that a four-year-head-start claims corpus does not," and the only credible answer on the table is the provider-side family network as a distribution wedge, not the pipeline design.

The second thing that changed is where the defensible entry sits. Two of the four direct-vertical players found this pass are not RCM startups at all; they are practice-management vendors adding billing AI from an installed base (RethinkBH, probably Raven Health). That is the pattern in this vertical: billing automation is being sold as a feature of the system of record, not as a standalone. A standalone EOB/denial product sold to independent practices has to beat both the incumbent's bundle and the labor-arbitrage billing shops on the same budget line. That is a harder shape than the 2026-09-16 sketch assumed, and it is the kind of structural fact worth putting in front of Michael, who has sat on the provider side of exactly this purchase.

Methodological note that is itself a finding: SEC Form D is nearly blind to this cohort. EDGAR full-text search returned zero Form D hits for Camber Health, Thoughtful Automation, Finni Health, Raven Health and Nirvana Health, all of which have publicly reported rounds. The only Form D I found in the whole scan was Behave Health's, from 2019. Form D covers exempt offerings only, is filed under legal entity names that often differ from brand names, and is skipped outright by companies relying on Section 4(a)(2) placements. The founder's standing primary-source preference is right on principle, and this pass shows that for venture-stage healthcare software the practical primary source is the company's own site plus state filings, not EDGAR. The Tampa Form D scan's negative should be read with that in mind: it was testing with an instrument that mostly cannot see this cohort.

Why this is in the vault

This resolves one of two open flags on the family co-founder venture thread (project_life_sciences_career_exploration): whether the RCM/EOB angle is an empty lane. It is not empty, and the named competitor is Camber, not ICBD -- which is a direct correction to the competitive verdict in [[2026-05-03-icbd-holdings-curative-ai-research]] and to the "Tampa has room" read in [[2026-09-23-tampa-ecosystem-map-v0]] section (d). It also supplies the second question to put to Michael Holzum, alongside the already-open one about why the ICBD build failed.

Method and audit trail

So the negative is checkable rather than assertable.

Vault docs read (6): 01-projects/network-map/2026-09-23-tampa-ecosystem-map-v0.md, 06-reference/2026-05-03-icbd-holdings-curative-ai-research.md, and the agent-memory records project_life_sciences_career_exploration, user_extended_family_map, project_income_seat_gap_400k (referenced), user_career_commitment_shape (referenced).

QMD queries (3 of 5 used), collection rdco-vault: (1) lex "RCM EOB AI venture Michael Christian co-founder residency" + vec on the family RCM/EOB venture; (2) lex "revenue cycle management denial 835 EOB claims automation RCM AI" + vec on AI RCM competitive landscape; (3) directory listings for cross-link verification.

SEC EDGAR company-name search (browse-edgar, type=D, all dates), 10 names: Thoughtful Automation, Finni Health, Nirvana Health, Eleos Health, Behave Health, Ritten, Camber, Thoughtful, Nirvana, Raven Health. Hits: Behave Health Corp. (1 Form D, 2019). "Raven Health" and "Finni" returned "No matching companies." Camber, Thoughtful and Nirvana matched only unrelated funds and operating companies.

SEC EDGAR full-text search (efts.sec.gov), forms=D and all-forms, 5 names: "Camber Health," "Thoughtful Automation," "Finni Health," "Raven Health," "Nirvana Health." All five returned 0 Form D hits.

Direct HTTP status checks, 2026-10-01: curativeai.com → 404; www.curativeai.com → 404; exactbilling.com → 200; icbd.com → 200.

WebSearch (3 of 3 used), verbatim: "AI revenue cycle management startup ABA autism behavioral health claims denials funding 2026"; "substance use disorder addiction treatment billing AI startup raises seed Series A 2026 revenue cycle"; "Curative AI Exact Billing Solutions ICBD revenue cycle autism ABA SUD customers 2026."

WebFetch (3 of 3 used): camber.health (200, read); curativeai.com (404, no content); behavehealth.com (200, read).

Blind spots this method leaves:

• Bootstrapped and revenue-funded companies file no Form D at all. Exact Billing Solutions is the in-scan example.
• Form D covers exempt offerings only, and the brand-name-to-legal-entity mismatch defeats name search. Any competitor incorporated under an unrelated name is invisible to everything I ran.
• Stealth companies have no website and no filing, so they are structurally unfindable by this method.
• No state business filings were searched this pass (Florida Sunbiz, Delaware, Washington). That was the third primary leg the backlog item called for and it is undone.
• Search results skewed heavily to SEO lead-generation blogs. A competitor with no content-marketing budget would not have surfaced.
• One of three WebFetch slots was consumed by the Curative AI 404, so Exact Billing's own copy, Camber's customer and pricing pages, Thoughtful AI (ABA/autism RCM agents), Finni Health and Nirvana Health went uncharacterized.
• No payer-side or regulator source was reached, so every denial-rate figure here is vendor marketing.

Open follow-ups

• Verify Camber's actual funding: round sizes, dates and HQ, from its own announcements or Delaware/state filings rather than content sites, and check whether any Form D exists under a different legal entity name.
• Verify or kill the Raven Health September 2026 RCM launch from the company's own site, and determine whether it is a built product or an announcement.
• Characterize the three companies this pass named but never read: Thoughtful AI (markets AI RCM agents for ABA/autism), Finni Health (autism provider enablement including billing), Nirvana Health (behavioral-health benefits and eligibility verification). All three had zero Form D hits, so site-first.
• Run the state-filings leg the backlog item asked for: Florida Sunbiz, Delaware and Washington entity searches for RCM/behavioral-health software registrants, to test whether the EDGAR blindness is covered by state records.
• Confirm Curative AI's status beyond the 404: check ICBD's own site and careers page for whether Curative AI is still listed as an operating company, and get Michael Holzum's account of what happened to it on the record alongside the already-open "why did it fail" question.
• Test the structural read that billing AI in behavioral health is being sold as a feature of the system of record rather than standalone, by checking whether any of the named ABA/SUD EHR vendors sell their billing AI to practices that use a competitor's EHR.
• Source the ABA and SUD denial-rate claims (15-30% and ~25%) to a payer, state Medicaid, or peer-reviewed source instead of vendor marketing, since the whole venture premise rests on that spread being real.

Related

• [[2026-09-23-tampa-ecosystem-map-v0]] — the Tampa-only scan this brief extends; section (d) holds the RCM/EOB negative and the "a search miss is not proof of absence" flag
• [[2026-05-03-icbd-holdings-curative-ai-research]] — the ICBD/Curative AI brief whose competitive verdict this brief downgrades
• [[2026-09-23-formd-tampa-60d]] — the Form D run underlying the Tampa negative, and the instrument this brief shows is nearly blind to venture-stage healthcare software
• [[2026-07-21-family-medical-careers-scenario]] — the family provider-side network that would be the venture's distribution wedge
• [[2026-07-02-hipaa-ambient-ai-independent-medical-practices]] — adjacent prior scan of healthcare AI sold into independent practices, including how fast scribing commoditized
• [[2026-08-30-shortlist-refresh-scored]] — the acquisition funnel's home health / RCM lane, the other place RCM shows up in RDCO thinking

Sources

Vault

• ~/rdco-vault/01-projects/network-map/2026-09-23-tampa-ecosystem-map-v0.md
• ~/rdco-vault/06-reference/2026-05-03-icbd-holdings-curative-ai-research.md
• ~/rdco-vault/01-projects/network-map/2026-09-23-formd-tampa-60d.md
• ~/rdco-vault/01-projects/life/2026-07-21-family-medical-careers-scenario.md
• Agent memory: project_life_sciences_career_exploration, user_extended_family_map

Primary

• Camber homepage — https://www.camber.health/ (self-description, named verticals, named investors, claims metrics)
• Behave Health homepage — https://behavehealth.com/ (self-description, ERA/EOB and denial-management copy)
• SEC Form D, Behave Health Corp., filed 2019-10-18 — https://www.sec.gov/Archives/edgar/data/1789714/000178971419000001/primary_doc.xml
• SEC EDGAR submissions index, CIK 0001789714 — https://data.sec.gov/submissions/CIK0001789714.json
• SEC EDGAR company search and full-text search, queries listed under Method (negative results)
• Direct HTTP status checks on curativeai.com, www.curativeai.com, exactbilling.com, icbd.com, 2026-10-01
• RethinkBH "BillAI" announcement, April 2026 — https://www.prnewswire.com/news-releases/rethinkbh-unveils-billai-an-intelligent-billing-system-built-to-reduce-burden-and-tackle-abas-most-complex-challenges-302748305.html

Secondary / unconfirmed

• Camber profile (claims volume, patient count, states, launch year) — https://acuity.news/technology/camber-ai-aba-billing-revenue-cycle-management/
• Curative AI $9.8M self-funded figure — https://breakingnewsaba.com/featured/curativeai-aba-centers
• Raven Health RCM launch (search-result snippet, page not fetched) — https://breakingnewsaba.com/press-release/raven-health-launches-ai-powered-revenue
• SUD denial-rate figure (vendor marketing) — https://aihealthfinance.com/substance-use-disorder-billing-guide/
• ABA denial-rate figure (vendor marketing) — https://www.rethinkbehavioralhealth.com/resources/ai-reduce-billing-denials-aba-therapy/