Behavioral-health billing is bundled: of 24 ABA/SUD EHR vendors, one publishes an offer to run revenue cycle on a rival system of record

Sensitivity note: this brief names a family member's private holding company and an abandoned internal project. Local vault only, per the living-person rule in [[hcls/network|HCLS network]].

The question

Verbatim: "Is behavioral-health billing AI being sold as a feature of the system of record rather than as standalone software? Test it: do any ABA/SUD EHR vendors sell their billing AI to practices running a COMPETITOR's EHR?"

Context: this is the structural test promoted out of [[2026-10-01-national-rcm-eob-ai-behavioral-health-scan]]. If no electronic health record (EHR) vendor will sell its billing artificial intelligence (AI) onto a rival EHR, distribution is the moat and a standalone entrant is structurally disadvantaged regardless of model quality. Acronyms used throughout: ABA = Applied Behavior Analysis; SUD = Substance Use Disorder; RCM = revenue cycle management; EMR = electronic medical record (used interchangeably with EHR by these vendors); PMS = practice management system; EOB = explanation of benefits; 835 = the X12 electronic remittance advice file a payer returns against a claim; CARC/RARC = claim and remittance adjustment reason codes; MSO = managed service organization; ONC = the federal Office of the National Coordinator for Health IT; HCLS = healthcare and life sciences.

What we already know (from the vault)

• The parent scan [[2026-10-01-national-rcm-eob-ai-behavioral-health-scan]] asserted, as internal synthesis with no external citation, that "billing automation is being sold as a feature of the system of record, not as a standalone," on two data points only: RethinkBH's BillAI announcement (confirmed) and Raven Health's RCM launch (unverified). It flagged the cross-EHR test as an explicit open follow-up and enumerated no EHR vendor set. This brief supplies both.
• That scan named the likely moat as payer-specific 835 heterogeneity, CARC/RARC rules, PMS integration and trust thresholds rather than model quality. That moat read survives this pass. Its separate observation that it found no pricing data does not survive: this pass found five published price points, listed under open follow-ups.
• [[2026-06-30-every-ai-strategy-bets-explicit]] supplies the frame that makes the question load-bearing: products connecting AI to "messy external systems (EHR workflows, systems of record)" survive capability growth, while absorbable scaffolding does not. The unresolved half is whether the integration survives the incumbent, not the model.
• [[hcls/problem-log|HCLS problem log]] holds the originating pain: independent practices losing staff time reconciling EOB and 835 remittances, logged 2026-09-16 at medium confidence, sourced to Michael Holzum's holding company shutting down its own AI RCM effort before a demo.
• The family venture is an exploration on a 3-6 year horizon tied to Christian Whidden's residency, not a committed build, and the founder already stepped back from RCM as "the" idea on 2026-09-16. The network in [[hcls/network|HCLS network]] is almost entirely provider-side.

What the web says

Evidence standard applied: every verdict traces to a page retrieved 2026-10-03. Lines like "integrates with any EHR" with no named system were rejected as evidence. Quotes and redirects marked (self-verified) I re-extracted myself by curl rather than trusting a sub-agent paraphrase; two sub-agent claims failed to reproduce and were corrected or dropped as a result.

The enumerated vendor set and the test result. Twenty-four independent ABA, SUD and behavioral-health EHR/practice-management vendors, so the ratio below is reconstructible:

• Exactly one vendor in that set states plainly that its EHR is not a prerequisite. Netsmart's RCM FAQ: "Am I required to have a Netsmart electronic health record (EHR) to use Netsmart revenue cycle technology? No - Netsmart revenue cycle solutions are EHR-agnostic. You do not have to have a Netsmart electronic health record to use our collections automation or claims management solutions" (self-verified, https://ntst.com/Solutions/By-Capability/Revenue-Cycle-Management). High confidence that the claim is published. Lower confidence that it is commercially live at scale: the page names no rival EHR and no cross-EHR customer, and Netsmart does not label the product as AI.
• Two vendors put the EHR prerequisite in writing on a pricing page. Ensora Health (formerly Therapy Brands): "Ensora RCM. A billing team that works inside your EHR. Add-on managed billing for Fusion and TheraNest users" (self-verified, https://ensorahealth.com/pricing/rcm/), both Ensora's own EHRs. BestNotes: "This API pricing is exclusive to BestNotes customers managed under an existing Billing Services CollaborateMD account" (self-verified, https://www.bestnotes.com/cmd-pricing/). High confidence.
• The two category leaders describe billing only as an in-platform surface, including their AI products. CentralReach: "Access RCM experts who will manage all billing processes within CentralReach" (self-verified, https://centralreach.com/products/billmax/); ClaimCheckAI is "embedded directly into the CentralReach platform" and ClaimAcceleratorAI resolves denials "inside CR Enterprise." Kipu: "By integrating with Kipu EMR, teams save time and reduce human-error by pushing patient demographics and charges seamlessly to the billing software" (self-verified, https://www.kipuhealth.com/products/rcm/). High confidence on the quotes. Neither page states the EHR is required, and Kipu leaves one door ajar with "from the Kipu EMR or bulk importing treatments," naming no system. For both leaders this is "the published material describes it only as bundled," not "verified not sold cross-EHR."
• Important calibration limit on the test as posed. The question asks about billing AI. Only three vendors in the 24 have an identified AI billing product at all: CentralReach (two), RethinkBH (BillAI), and Raven Health (claimed). That subset is too small to carry a ratio, and RethinkBH's integrations page 404'd. The bundling pattern above is therefore established for billing and RCM generally, and is only suggestive about AI billing specifically. Medium confidence, and this is the weakest joint in the brief.
• The cross-EHR layer that does exist is occupied by independents who do not own an EHR, and it sits beside billing rather than inside it. Nirvana, an eligibility and benefits AI layer, publishes a named-platform list of roughly 25 systems plus "+71 more platforms," including four behavioral-health systems across three parents: Netsmart, Kipu Health, CareLogic and Qualifacts, the last two being one parent (self-verified, https://www.meetnirvana.com/ehr). Candid Health names Qualifacts OnCall. Plutus Health, an outsourced RCM firm, names CentralReach. Waystar's supported practice-management list names Credible. High confidence on the named lists. Low confidence that any of these is a billing AI product rather than eligibility checking, claims plumbing, or labor-based services.
• The most behavioral-health-native independent publishes no integration names of its own. Camber has dedicated ABA and SUD vertical pages and about 15 named ABA customers including Intercare, ACES ABA and Gracent, with a case-study claim of ">80% reduction in initial denials rate" (vendor-reported, unaudited). Its integrations page 404s. One third-party page, Hi Rasmus's integrations list, names Camber. So Camber's cross-EHR reach is asserted by a partner, not documented by Camber.
• Two re-bundling counter-signals. Apex EDI, formerly an independent clearinghouse serving many practice-management systems, now 301s to "Ensora Clearinghouse, also known as APEX EDI" inside an EHR owner (self-verified). Thoughtful AI, named in the parent scan as an ABA/autism RCM vendor, now 301s to smarterdx.com carrying utm_campaign=site-sunset (self-verified). High confidence on both redirects. Low confidence on what the Thoughtful AI redirect means commercially: no acquisition or shutdown statement was retrieved, and SmarterDx's site is hospital-oriented with no ABA mention.
• Unplanned finding: the vendor set is more consolidated than the parent scan implied. Ten formerly independent brands now redirect into four parents (all self-verified by captured 301/302): datafinch.com, webaba.com, theranest.com, fusionwebclinic.com, codemetro.com and apexedi.com into ensorahealth.com; threadlearning.com into edu.centralreach.com; totalaba.com into rethinktotaltherapy.com; credibleinc.com and insynchcs.com into qualifacts.com. High confidence on those ten. Four further consolidations reported by a sub-agent did NOT reproduce on my own curl and I am not relying on them: procentive.com did not resolve for me, echogroupinc.com redirects to an unrelated electrical supplier, and tenelevengroup.com, carelogic.com and lightningstep.com each return 200 on their own domain. So "roughly six roll-ups" is the right order of magnitude but is not established at that precision.

Convergences and contradictions

• Convergence, now with evidence under it. The parent brief's uncited synthesis holds for billing and RCM broadly. Of 24 independent ABA and SUD EHR vendors, one publishes an EHR-agnostic RCM offer, one prices an optional link to a third-party RCM or PMS, one did so historically and no longer does, two put the prerequisite in writing, and the remainder describe billing as a module of their own platform or publish nothing either way. The falsifiable test resolves much closer to "no EHR vendor offers billing onto a rival EHR" than to "several do."
• Contradiction with the clean version of that read. The layer is not sealed, and one vendor in the set is an outright exception. Independents do sell across named behavioral-health EHRs, but the functions they sell are eligibility and benefits verification, claims plumbing, and labor-based RCM services rather than the denial and remittance automation the EHR vendors are bundling. The unbundled slice is real, narrow, and adjacent to the contested function rather than inside it. Medium confidence, bounded by how few integration pages exist at all in this category.
• An asymmetry worth naming, on one vendor's evidence. Kipu's EMR page says billable claim information can go "directly to Kipu RCM, or other billing solutions." For Kipu, openness runs outward from the system of record, not inward to its billing product: it will let a third-party biller consume its data, while none of its retrieved pages offers Kipu RCM to a non-Kipu-EMR practice. I have this shape documented only for Kipu and should not generalize it to the set, but it is what you would expect if distribution rather than billing capability is the asset being protected.
• Two vendor-page framings that look like cross-EHR evidence and are not. AlohaABA's integrations page names Ensora Health, Hi Rasmus, Motivity and Twyll (Twyll marked "COMING SOON"), and Hi Rasmus's names Camber, Lumary, Aloha ABA, Therapy PMS and Boost (both self-verified). Both are inbound data integrations into the publishing vendor's own platform, not an offer to run the publisher's billing on a rival's system of record. They fail the test as posed, which is why both are marked NOT-THE-TEST above. Naming this explicitly because it is the easiest false positive in the category.

Synthesis for RDCO

The named test comes back close to a no, and the practical consequence is that a standalone behavioral-health billing product aimed at practices has to beat a bundle it cannot unbundle. One vendor in 24 publishes an EHR-agnostic RCM offer, and that one backs it with no named rival EHR and no named cross-EHR customer. Everyone else describes billing as something their staff or their software does inside their own platform, and two say so on a pricing page. For a practice owner, the incumbent's billing module is a line item on a contract already signed, with no integration project, no second vendor security review, and no data-mapping risk. A standalone has to win that comparison on a budget line the incumbent controls, and model quality is not the variable that decides it. I hold this at medium-high confidence for billing and RCM generally, and only medium for billing AI specifically, because just three of the 24 vendors have an identified AI billing product and one of those pages 404'd. A great deal of this category's commercial reality is also gated behind sales calls, and "no page I retrieved offers cross-EHR billing" is materially weaker than "cross-EHR billing is not offered."

The more useful observation is where the opening might be, and it points at a buyer rather than a feature. Several of the cross-EHR examples in this pass involve a vendor selling across many systems of record (Nirvana across roughly 25 named platforms, Netsmart explicitly EHR-agnostic, Plutus naming CentralReach, Kipu's RCM testimonial coming from a third-party billing firm). What I did not establish is the matching claim on the buyer side: that billing companies, MSOs and acquisition-grown provider groups actually run several EHRs each and therefore cannot be served by any one incumbent's bundle. That is a plausible hypothesis, not a finding. No page in this pass stated how many EHRs any such buyer operates. I hold the beachhead argument at low confidence until that number is established, and it is cheap to establish: Michael Holzum's holding company runs an ABA group plus rehab assets and a captive biller (Exact Billing Solutions), and already tried and abandoned an internal AI RCM effort. One conversation settles whether the captive biller touches more than one EHR. If it does, the hypothesis has a design partner. If it does not, the hypothesis is weak and this thread should be scoped differently.

For the venture decision on a 3-6 year horizon, the consolidation finding matters more than any single vendor's answer. Ten formerly independent ABA and SUD brands now redirect into four parents, Avea was absorbed into Kipu RCM, Apex EDI into Ensora Clearinghouse, and Thoughtful AI's standalone site is retired. The direction of travel looks like tightening rather than loosening, which would mean a 2029 start date faces a more consolidated bundle than today's. I hold that forward call at low-medium confidence: it extrapolates from one pass of redirect evidence, and I did not verify most of the acquisition history at primary source. A reported 2025 Roper Technologies acquisition of CentralReach surfaced only in secondary summaries and I am not relying on it. The honest version is that two exit shapes look more plausible than winning the practice-software market: sell a function to the roll-ups as a component, or own a multi-EHR buyer segment the roll-ups structurally cannot address. It also raises a fair question about whether the EOB and 835 reconciliation pain in the problem log is a venture or a feature CentralReach ships in 2027.

I would not read this brief as a kill on the RCM thread. It is a kill on one shape of it: a general-purpose billing AI sold to independent practices on the strength of its model. The narrow reading is that the function is bundled, and that whether the buyer segment is also bundled is still an open question rather than a settled opening. The family network the founder actually has is provider-side and operator-side, which fits investigating the buyer-side question better than it fits building the product.

Why this is in the vault

This closes the explicit cross-EHR follow-up left open by [[2026-10-01-national-rcm-eob-ai-behavioral-health-scan]] and replaces its uncited "billing is a feature of the system of record" synthesis with a first-party-cited, enumerated vendor test, so the family RCM/EOB AI exploration can be scoped against a specific next question (how many EHRs a behavioral-health billing company or MSO actually runs) instead of against the practice market the incumbents already own. It also corrects the vendor-count assumption that pass worked from: ten formerly independent brands now redirect into four parents, which changes who an eventual product would be sold to or acquired by.

Open follow-ups

• Establish how many distinct EHRs a behavioral-health billing company or MSO actually operates against, starting with Exact Billing Solutions inside Michael Holzum's holding company. The entire multi-EHR beachhead argument rests on that number being greater than one, and nothing in this pass established it.
• Get Michael Holzum on the record about why the ICBD/ABA Centers AI RCM effort died, which is still open from the 2026-09-16 problem-log entry.
• Retrieve rcm.insynchcs.com, which failed with an SSL handshake error on two attempts. It is the likeliest single place Qualifacts would publish a cross-EHR RCM claim, so the Qualifacts bundled-only verdict has a hole in it.
• Open the Ensora ONC real-world-testing filings at ensorahealth.com/onc/{echo,teneleven}/. Certification filings are the second likeliest place a named rival system appears, and they were not opened. They would also settle whether Echo Group and TenEleven are in fact Ensora brands, which my own curl did not confirm.
• Verify what happened to Thoughtful AI: acquisition, asset sale, or shutdown. Its retirement is the strongest available natural experiment on whether a standalone ABA RCM AI survives against the bundle.
• Price the bundle properly. The only published prices found this pass were Ensora RCM ("by quote"), Theralytics (4.5-5% of receivables), BestNotes ($50/month for the CollaborateMD API connection), SimplePractice (Insurance Navigator at $1.79/claim), and on the independent side Enter Health ($2,500-$5,000/month plus 5-6% of collections). No price was found for CentralReach, Kipu, Qualifacts or any AI billing product. Without the incumbent's price, a standalone's pricing power is unknowable.
• Re-run the test against AI billing products specifically once more exist. Three identified products across 24 vendors is too thin to support a ratio, and this is the brief's weakest joint.
• Verify CentralReach's and Kipu's acquisition histories at primary source, including the reported Roper Technologies transaction, which this brief does not rely on.

Related

• [[2026-10-01-national-rcm-eob-ai-behavioral-health-scan]]
• [[2026-06-30-every-ai-strategy-bets-explicit]]
• [[2026-09-30-mercor-lessons-and-snowflake-dataset-monetization]]
• [[hcls/problem-log|HCLS problem log]]
• [[hcls/network|HCLS network]]
• [[2026-09-23-tampa-ecosystem-map-v0]]

Sources

Vault:

• ~/rdco-vault/06-reference/research/2026-10-01-national-rcm-eob-ai-behavioral-health-scan.md
• ~/rdco-vault/06-reference/2026-06-30-every-ai-strategy-bets-explicit.md
• ~/rdco-vault/01-projects/hcls/problem-log.md
• ~/rdco-vault/01-projects/hcls/network.md
• ~/rdco-vault/01-projects/hcls/index.md

Web, all retrieved 2026-10-03. (self-verified) = re-extracted directly by the author of this brief.

EHR vendor pages bearing on the test:

• https://ntst.com/Solutions/By-Capability/Revenue-Cycle-Management - Netsmart FAQ, "EHR-agnostic" (self-verified)
• https://ntst.com/-/media/pdfs/billing-revenue-cycle/rcm-services-brochure.pdf - "Uses organization's existing EHR"
• https://ensorahealth.com/pricing/rcm/ - "Add-on managed billing for Fusion and TheraNest users" (self-verified)
• https://www.bestnotes.com/cmd-pricing/ - "exclusive to BestNotes customers" (self-verified)
• https://www.qualifacts.com/behavioral-health-software/revenue-cycle-management-services/ - "within your Qualifacts EHR" (self-verified)
• https://centralreach.com/products/billmax/ - "within CentralReach" (self-verified)
• https://centralreach.com/products/claimcheckai/ , https://centralreach.com/products/claimacceleratorai/ - AI claims products, in-platform framing
• https://centralreach.com/centralreach-extends-billing-offering-with-acquisition-of-bronco-billing-now-named-cr-billmax-services/ - Bronco Billing acquisition, Dec 2018
• https://centralreach.com/about/partners/ - 7 named partners, none an EHR
• https://www.kipuhealth.com/products/rcm/ - "By integrating with Kipu EMR"; "or bulk importing treatments" (self-verified)
• https://www.kipuhealth.com/news/kipu-acquires-avea-combining-the-best-clinical-and-revenue-cycle-solutions-for-behavioral-health/ - Avea acquisition, Sept 2021
• https://www.kipuhealth.com/partners-and-integrations/certified-partners/ - 11 named partners, none an EHR
• https://www.opusehr.com/links/professional-billing-services - optional HL7 link to a third-party RCM or PMS, $2,500 setup
• https://www.lightningstep.com/blog/easy-billing-rcm - 2021 page, "stand-alone, web-based" LightningRCM sold to billing companies
• https://www.ritten.io/rcm , https://sunwavehealth.com/products/rcm/ , https://helloalleva.com/alleva-billing/ , https://streamlinehealthcare.com/solutions/revenue-cycle-management/ , https://www.valant.io/services/revenue-cycle-management/ , https://www.behavehealth.com/rcm , https://www.simplepractice.com/pricing/ , https://ravenhealth.com/key-features-managed-billing-services , https://theralytics.net , https://motivity.net/integrations , https://www.measurepm.com/aba-billing-software/ , https://www.artemisaba.com , https://portiapro.com , https://welligent.com/solutions/secure-integrations/ - bundled-only or no-evidence verdicts
• https://www.alohaaba.com/integrations - names Ensora Health, HiRasmus, Motivity, Twyll ("COMING SOON") (self-verified)
• https://hirasmus.com/integrations/ - names Camber, Lumary, Aloha ABA, Therapy PMS, Boost
• https://eleos.health/ai-frequently-asked-questions/ - names 12 rival EHRs, but for its documentation product, not its RCM product

Independent (non-EHR-owning) vendors:

• https://www.meetnirvana.com/ehr - ~25 named platforms plus "+71 more"; behavioral-health names Netsmart, Kipu Health, CareLogic, Qualifacts (self-verified)
• https://candidhealth.com/integrations - names Qualifacts OnCall among EMR partners
• https://www.plutushealthinc.com/ - names CentralReach; "Base + performance kicker"
• https://www.waystar.com/practice-management/ - names Credible, Net Health
• https://www.camber.health/ , /solutions/by-vertical/aba , /solutions/by-vertical/sud , /customers/stories/intercare - ABA and SUD verticals, ~15 named customers, ">80% reduction in initial denials rate" (vendor-reported), no integrations page
• https://enter.health/pricing - published price card, $2,500-$5,000/month plus 5-6% of collections

Consolidation evidence, captured 301/302 (self-verified):

• datafinch.com, webaba.com, theranest.com, fusionwebclinic.com, codemetro.com, apexedi.com -> ensorahealth.com product pages
• threadlearning.com -> edu.centralreach.com ; totalaba.com -> rethinktotaltherapy.com
• credibleinc.com -> qualifacts.com/credible-ehr/ ; insynchcs.com -> qualifacts.com
• www.thoughtful.ai -> smarterdx.com with utm_campaign=site-sunset
• https://ensorahealth.com/news/therapy-brands-is-now-ensora-health/ - Ensora rebrand, Apr 2025

Reported by a sub-agent but NOT reproduced on my own curl, and therefore not relied on: procentive.com into Ensora (did not resolve for me), echogroupinc.com into Ensora (redirects to an unrelated electrical supplier), tenelevengroup.com, carelogic.com and lightningstep.com into their parents (each returns 200 on its own domain), aveasolutions.com into a "Kipu RCM" page (did not resolve; www.aveaoffice.com serves a login portal).

Not reached, so no verdict drawn: rcm.insynchcs.com (SSL handshake failure), rethinkbehavioralhealth.com/integrations/ (404), centralreach.com/pricing/ and kipuhealth.com/pricing/ (both 404), camber.health/integrations (404), aveasolutions.com historical snapshots (no Wayback records returned), lumary.com and fortahealth.com (redirects not followed), janushealth.com (connection refused), Infinx and AKASA integration pages (404).