Clearing AI
EDI built the modern healthcare and B2B economy. It is also the reason every integration takes a quarter, every ERP migration runs over, and every consultant invoicing $300/hr knows the same five companies will keep calling. Clearing AI is the structured-translation layer between what enterprises emit and what modern systems can consume.
01 · The problem
EDI is the dark matter of US enterprise software
Most enterprise transactions still move on EDI. Every modernization project runs into a translation layer staffed by retirement-age specialists charging $300/hr, with project timelines measured in quarters. The translation layer is a tax that compounds every year a company tries to move to the cloud.
02 · The thesis
EDI is a translation problem, not an integration problem
The reason EDI projects run long is not technical complexity. It is that every implementation is a bespoke negotiation between two enterprises about field semantics that should already be canonical.
Treat the canonicalization as the product. Build a structured-data layer that ingests any EDI dialect, emits any API shape, and tracks lineage at the field level. The translation becomes infrastructure, not a project.
03 · The product
What it does
Canonical schema layer
Maintain a single internal representation across X12, EDIFACT, HL7, and modern API shapes.
Lineage at the field level
Every transformation carries a cryptographic audit so compliance can verify any output against any input.
Self-mapping ingestion
Onboard a new partner from sample documents in days, not quarters, using LLM-assisted schema inference with human approval.
Drop-in API egress
Emit OpenAPI-conformant payloads on top of the canonical layer so downstream systems consume modern shapes.
04 · Why now
The timing case
- 1
Cloud-native ERPs are forcing every mid-market and enterprise to re-platform their EDI integrations on a deadline they cannot extend.
- 2
The specialist talent pool is retiring faster than it is being replaced. The clock is structural.
- 3
Foundation models materially reduce the cost of inferring schema from sample documents, which is where 80% of an integration consultant's time has historically gone.
05 · Why I see it
The view from inside the work
I lived inside the EDI pipelines that move most of US healthcare's B2B settlement. The market underestimates how addressable this is once you treat canonicalization as the product, not the integration.
06 · Comparable references
What's already in the market, and where the gap is
An honest read on the adjacent landscape. Not every comparable is a competitor. Some are partners. Some are the market the venture displaces.
07 · Key risks
What could break the thesis
Operator-grade pre-mortem. Surfaced because the buyers and partners worth talking to will surface them anyway.
EDI is sticky. Buyers tolerate it because switching costs are real.
Sell as overlay first, replacement second. The first contract pays back on a single integration project.
Foundation-model accuracy on schema inference is not 100%.
Human approval in the onboarding loop, with confidence scoring. The product converges to fully automated as the model improves, not before.
Enterprise procurement cycles are long.
Land via developer/integration teams on per-project pricing. Expand to platform contract on year two.
08 · Proof of motion
What I've already shipped on this thesis
The artifacts that turn this from an essay into something with traction. Published work, working-group seats, operator scars.
- Canonicalization Beats WorkflowPillar essay on why this category compounds. Clearing AI is one of three instances.essay
- Clearinghouse data-layer operator track recordI ran the EDI pipelines that move most of US healthcare's B2B settlement. Lived through the consultant-tax tax cycle from inside.filing
09 · Questions partners ask
The next three follow-ups
Pre-empted because the buyers and partners worth talking to will surface them anyway.
Hasn't Boomi or Cleo been doing this for 20 years?
Yes. They sell the tool. The tool requires a consultant to map every new partner, every time. The ongoing cost compounds with the number of partners. Clearing AI sells the canonical schema layer that makes per-partner mapping a configuration, not a project. The vendors above kept selling the tool because they couldn't get past the schema. The shift this time is that foundation models can infer schema from sample documents in days, not quarters.
Schema inference at 100% accuracy isn't real. How do you handle the gap?
We don't claim 100%. We claim confidence-scored inference with a human approval step inside the onboarding loop. The product converges toward fully automated as the model improves, but it's deployed as a co-pilot from day one. The buyer gets day-one value at consultant-grade accuracy. The accuracy curve is the moat as it improves.
What's the wedge into a market with established incumbents?
Cloud-native ERPs forcing mid-market re-platforms on a deadline. Every NetSuite, Intacct, or Workday Financials migration runs into a translation layer. We land as the integration layer for the migration, on per-project pricing. The platform contract follows on year two when they've seen the schema layer compound across their partner network.
Status
Clearing AI is a published essay, not a stealth company. I am running Finexio. The thesis is here so the right operator or investor can find it and we can talk.
Of the eight ventures I've published, two are in discovery and I expect to operate one of them after Finexio. The rest, including this one, are pattern recognition I want in the open. If you read this and want to start it yourself, that is the outcome I'm hoping for.
Other concepts
Healthcare Adjudication
Apex Adjudication
AI claims-resolution engine for healthcare payers. Resolves manual denials at API speed.
Healthcare Settlement
Ledger Health
Cryptographic settlement and audit trails for multi-party healthcare reimbursement.
Healthcare ML
Vault Healthcare
Compliant data enclaves for collaborative healthcare ML training.