Healthcare integrations are a world of pain
Today we’re launching Asteroid Integrate: hundreds of prebuilt integrations to the EHRs, payer portals, and government systems that will never have good APIs.
Healthcare does have real integration rails. HL7 and FHIR work, EDI moves claims every day, and the top of the market publishes APIs worth building on. Underneath that, most of the systems an ops team actually touches offer one way in: a login page. So companies hire BPOs to fill the forms, duct-tape RPA onto portals that break, or wait a year in an integration queue that never moves.
AI has widened the gap. Voice agents are answering patient calls. Scribes are writing notes. Fax automations are reading referrals. Almost all of it dead-ends at the same login page, where a person still has to take over and re-type what the AI just captured.
A year automating healthcare’s hardest portals
For the past year we have gone after the systems most vendors decline. Not the friendly ones with a sandbox and a developer portal: the payer portal behind an IP allowlist, the state system that only opens inside Citrix, the EHR that rotates credentials every 30 days and kills the session after five minutes of thinking.
| 700,000+ | executions run in production |
| 250+ | agents live across 40+ systems |
| 6 years | of manual work saved |
What makes this work hard is everything around the form.
A hospice technology company needed patient records out of the referral portals that hospitals and skilled nursing facilities actually use. The records only ever exist on screen, there is no export, and every health system runs its own portal with its own login. A value-based oncology company needed care tasks executed inside its own systems at a scale its team couldn’t click through. A UK voice-AI receptionist could capture a patient’s request perfectly on the phone, then had nowhere to put it, because every GP practice runs a different online-triage system.
An ACA enrollment operation moved deals from a phone call into a marketplace platform field by field, plan by plan. A patient intake team drove one of the most stubborn ambulatory EHRs in production through referral creation, document upload, and scheduling, by hand, every time.
Underneath all of it sit the conditions that make this work unglamorous. Citrix-published apps that exist only as pixels. VPNs and IP allowlists. MFA on every login and credentials that rotate on a schedule. Session timeouts measured in minutes. And interfaces that quietly lie, like the EHR calendar that shows a booked slot as empty to anything that isn’t actually looking at the screen.
The volumes are real. One enrollment deployment runs ~26,500 portal executions a month. Another pulls ~40,000 patient records a month across eleven EHRs. A third processes 16,000+ inbound referrals. The companies differ, the queue names differ, and the motion underneath never does.
Why the standards never reached these systems
Healthcare has a name for the fix: interoperability. HL7, FHIR, and a generation of integration engines exist precisely so systems can exchange data. At the top of the market, they work. Major EHRs publish FHIR endpoints, the largest payers expose claims and coverage APIs, and regulation keeps pushing both forward.
The limit is structural: a standard only helps the systems that adopt it. The long tail of software that runs healthcare operations (the payer portal, the licensing board, the credentialing hub, the enrollment queue, the county registry) might never expose a functioning API. Not this roadmap, not the next one. There is no incentive and no budget. So “integration,” in practice, becomes a person with thirty browser tabs.
Our opinion, after a year inside these systems: the last mile of healthcare interoperability is an access problem, and standards work can’t immediately solve it. The intelligence to do this work has existed for a while. What was missing was a dependable way into systems that only offer a login page.
There have only ever been four ways in:
| Approach | Works when | Breaks because |
|---|---|---|
| Standards & APIs (FHIR, HL7, EDI) | The other system implements the standard | Most operational portals never will; no API, no roadmap |
| Custom integrations / RPA scripts | You have engineers and the target never changes | One UI change snaps the script; every system is a new project |
| Offshore ops teams | Always; a person can use any interface | Cost scales with volume, throughput doesn’t, and the work burns people out |
| Internal scripts | A developer owns and babysits them | The babysitting never ends, and it is never anyone’s real job |
This is the status quo Asteroid replaces.
What’s in the catalog
The catalog is the list of work healthcare ops teams do by hand every day. Out of the box:
- EHR extraction and write-back
- Scheduling and prior authorization
- Eligibility, claims, and denials
- Credentialing, exclusion screening, and provider enrollment
- ACA, ICHRA, and Medicare enrollment
Web portals, Citrix environments, and Windows desktop apps are all in scope: if a person can operate it, we can connect you to it.
You don’t build from scratch. Take a template, customize it to your operations in minutes with Astro, our AI agent builder, then plug it into your product via API or MCP. Examples are available in the workflow library, and to get full access to the catalog,
How to get started
Book a 30-minute scoping call - Customize the templates. Sign up, take what matches your operations, adjust it in Astro, and connect it to your product via API or MCP.
- Go live in days, not months. Run it with approval gates on the steps that deserve a human.
Book a demo and watch the workflow you staff by hand run as an agent instead.
What changes for your ops team
If you own healthcare integration operations, you already know the math this changes: every new client, every new payer, every new state means another set of portal logins someone on your team has to learn, or a six-month project to build the integration yourself.
Start with the workflow that hurts most. Browse the library, or book a demo and watch your own workflow run live. The deeper background on this problem space is in our healthcare workflow automation guide.
We started Asteroid because we believe AI should be deployed safely and responsibly on the work that matters most. Integrate is that belief pointed at healthcare operations: agents handling patient scheduling, eligibility checks, benefits enrollment, and claims, with humans approving the decisions that deserve a human.
Over the coming months we will keep adding integrations and partnering across the industry to make these connections smoother and more reliable. If the system your team dreads logging into isn’t in the catalog yet, tell us and it will be.
David & Joe, co-founders, Asteroid