Asteroid
Also known as: Asteroid AI, Asteroid Agents, Asteroid Astro, asteroid.ai
Healthcare portal integration platform whose supervised browser and computer use agents run prior authorization, eligibility, credentialing, claims, scheduling and enrollment across payer portals, EHRs and desktop apps, with approval checkpoints and every run recorded. YC W25.
Asteroid, from Entropy Systems, Inc., is a healthcare portal integration platform. It offers one connector for EHRs, payer portals, registries and desktop apps that have no good API. Workflows are graphs of agent nodes and transitions rather than ReAct loops. Each node reads the screen and acts in a Browser, Linux or Windows sandbox, and a proven path can be turned into a Playwright script that replays in milliseconds with no model cost. Astro, the agent builder, explores a portal with the user, builds the workflow, analyzes past executions and writes improvements back to a draft.
A Workflow Library ships pre built workflows across ten categories, from OIG and DEA checks and eligibility to prior authorization on payer, UM and ePA portals, provider and Medicare enrollment, EHR extraction and write back, claims and denials, and referral intake.
Workflows run on cron schedules or from the API, MCP and Slack, push events to webhooks and Slack, hold credentials in an organization vault, and give each agent profile its own inbox for login codes. Operators watch agents live, pause them and require approval before sensitive steps, and every action is recorded and replayable. Asteroid holds SOC 2, signs BAAs for HIPAA, offers role based access control and hosts data in the United States. YC W25.
Vendor details
Canonical URL
https://asteroid.ai
Category
Browser / computer-use agent
Funding status
Y Combinator W25; used in production across hundreds of startups and enterprises with 5,000+ production agents built, concentrated in healthcare and insurance
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
Asteroid operates EHRs, payer portals, registries and desktop apps through the screen, with credentials in an organization vault and VPN access. It calls APIs such as SAM.gov where they exist. Workflows start from the API, MCP and Slack and send events to webhooks and Slack. Developers get a REST API, an SDK and an MCP server at mcp.asteroid.ai.
In practice
A prior authorization team has an agent log into payer portals with credentials from the organization vault, answer the payer's clinical questionnaire strictly from the chart material supplied for the case, and check pended requests again on a schedule.
A credentialing team runs Workflow Library templates such as OIG Exclusion Check and DEA Registration Verification, with agents calling SAM.gov directly and writing the results back to the EHR or CRM.
An operations lead watches agents work live in a Windows sandbox and pauses them before sensitive steps, then replays the screen recording and step by step timeline to see why a node failed while Astro writes a fix into a draft version.
Sources & related URLs
Agentic Index coverage score
11.5 / 14 capabilities · 82%
| Integrations & Tool Calling | Full |
|---|---|
|
Agents act in EHRs, payer portals and desktop apps with credentials held in an organization vault, reach systems through VPNs, call APIs such as SAM.gov directly, and write results back to EHRs and CRMs; each agent profile owns an inbox for login codes and mail. Sourceasteroid.ai/llms.txtread 2026-09-27 |
|
| Workflow Orchestration | Full |
|
Workflows are graphs of agent nodes and transitions with failure paths, loop backs, handoffs, per node Playwright scripts and batch executions, built in the platform or by Astro from a described process. Sourcedocs.asteroid.ai/llms.txtread 2026-09-27 |
|
| Knowledge Grounding & RAG | Partial |
|
Prior authorization workflows answer payer clinical questionnaires strictly from the chart material and documentation supplied for each case, and workflows read attached files. That context is assembled for each run, and Asteroid keeps no standing index over the customer's knowledge. Sourceasteroid.ai/llms.txtread 2026-09-27 |
|
| Human Oversight & Guardrails | Full |
|
Operators watch agents live, pause them and require approval before sensitive steps; workflows carry clinician in the loop checkpoints, a human checkpoint before enrollment submission and a person approving every binder payment, and nodes can ask the user a question. Sourceasteroid.ai/securityread 2026-09-27 |
|
| Security, Identity & Governance | Full |
|
Customers manage role based access control and least privilege with granular permissions, and portal credentials are encrypted, revocable and scoped per agent. Asteroid holds SOC 2, with the report in the Trust Center, alongside HIPAA with BAAs and third party penetration testing. Its security page calls the SOC 2 report Type II, while its llms.txt summary says Type I. Sourceasteroid.ai/securityread 2026-09-27 |
|
| Observability & Auditability | Full |
|
Every action an agent takes is recorded and replayable, with screen recordings, a step by step timeline, the page seen at each step and the reasoning behind each action and transition, plus an activity log with node durations. Sourcedocs.asteroid.ai/build/test-and-publishread 2026-09-27 |
|
| Memory & State Persistence | Partial |
|
An agent's own files include memory kept beside its nodes in the workflow file tree, and the workspace folder carries state across steps and loop backs inside one run only. Asteroid does not say how agent memory is scoped or how long it lasts. Agent profiles carry browser state, not memory. Sourcedocs.asteroid.ai/llms.txtread 2026-09-27 |
|
| Deployment & Data Residency | Not documented |
|
Data is hosted in the United States on Asteroid's cloud, the only region offered, and there is no VPC, on premise or self hosted option. The Browser, Linux and Windows sandboxes and VPN access into customer networks are where agents run and how they connect, and none of them places the platform in the customer's own infrastructure. Sourceasteroid.ai/securityread 2026-09-27 |
|
| Prebuilt Agents, Templates & Packs | Full |
|
A Workflow Library offers pre built workflow templates across ten categories, mirrored in the docs catalog, such as OIG Exclusion Check, DEA Registration Verification, Eligibility and Benefits Check, Drug Prior Authorization on ePA Portals and Medicare Commission Reconciliation. Each template handles one job. Sourceasteroid.ai/workflow-libraryread 2026-09-27 |
|
| Triggers & Channel Coverage | Full |
|
Workflows run on a repeating cron schedule or a single booked time, start from the API, MCP and Slack, and push execution events to webhooks and Slack with signed payloads; pended prior authorizations are re-checked on schedule. Sourcedocs.asteroid.ai/operate/schedulesread 2026-09-27 |
|
| Model Flexibility & Routing | Partial |
|
Each node picks a model from Asteroid's own tiers, asteroid-fast, asteroid-balanced and asteroid-max. The changelog names Anthropic, OpenAI and Google models behind prompt tuning and a provider grouped picker, but customers get no list of providers to choose from, so Asteroid decides which providers serve each tier. Sourcedocs.asteroid.ai/llms-full.txtread 2026-09-27 |
|
| APIs, SDKs & MCP Extensibility | Full |
|
A REST API and SDK deploy, execute and manage agents, with a workflow file representation for version control, and an MCP server at mcp.asteroid.ai builds, executes and debugs agents from a coding assistant. Sourcedocs.asteroid.ai/llms.txtread 2026-09-27 |
|
| Testing, Debugging & Optimization | Full |
|
Changes run as a draft version before publishing, with version compare and a warning before schema breaking publishes. Astro reads past executions' timelines to name slow or failing nodes and writes improvements back to a draft, so the deployed agent's behavior feeds its next version. Sourcedocs.asteroid.ai/operate/improveread 2026-09-27 |
|
| Browser & Computer Use | Full |
|
Agent nodes with browser_use and computer_use capabilities read the screen and act in payer portals, EHRs and desktop apps inside Browser, Linux or Windows sandboxes the vendor runs, with Playwright scripts replaying proven paths. Sourcedocs.asteroid.ai/concepts/environmentsread 2026-09-27 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Product updates
Asteroid merged its Vault and Agent Profiles into one Logins page that holds every account its browser agents sign in as, including groups of interchangeable logins for the same portal. The workflow builder adds a Runs as control that limits which logins each workflow may use, and the API gains matching routes while the old vault and profile paths are deprecated.
Bears on: Security / enterprise
View sourcePricing
Individual is free to start, with pay as you go credit top ups. Startup is $300 a month ($330 of credits) and Scale $1,000 a month ($1,150 of credits). Enterprise is custom.
Credits that cover runtime, LLM usage and proxies, bought through pay as you go top ups on the free Individual plan or through a monthly plan that includes more credits than its fee.
What is public
All four plans are public with their prices. Individual is free with pay as you go top ups, Startup is $300 a month with $330 of credits, Scale is $1,000 a month with $1,150 of credits and managed delivery, and Enterprise is a custom credit commit. Credits cover runtime, LLM usage and proxies.
Variable cost rationale
Credits are the meter, and they pay for runtime, LLM usage and proxies, so the bill grows with the number and length of agent runs. Asteroid compiles a mature workflow into a deterministic script that replays runs without the model at a fraction of the cost, which reduces model spend per run over time. The Individual plan has no monthly commitment and runs on credits topped up as needed. Startup and Scale are fixed monthly plans that include credits worth more than the fee, and Enterprise is a custom credit commit.
Sales call required
No, self serve available
Free / trial
Individual plan starts free with no monthly commitment, and credits are topped up as you go
Lowest paid plan
Startup, $300 a month with $330 of credits (10 percent bonus)
Cancellation / refund
Individual has no monthly commitment. Scale carries a money back guarantee on scoped workflows, and Enterprise has custom terms.
Support SLA / resale
Priority Slack support from Startup up, FDE onboarding on Scale, and a dedicated AI engineer with SLAs on Enterprise.
Related vendors
- AgentQL — A self healing query language for the web, with natural language…
- AGI, Inc. — AGI, Inc
- Airtop — Cloud browsers for AI agents that compile plain English workflows…
- Automat AI — AI agents that operate computers visually, sold as Automat Workforce…
- Autotab — General AI agent that learns a workflow from a demonstration and…
- Axiom.ai — No code browser automation tool and Chrome extension for building…
Alternatives to Asteroid
The closest documented capability profiles to Asteroid among browser and computer-use agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Grok Bot10.5 / 14A lighter documented profile than Asteroid
- Manus10.0 / 14A lighter documented profile than Asteroid
- Automat AI11.5 / 14Adds documented Deployment & Data Residency
- Airtop9.0 / 14A lighter documented profile than Asteroid
- Browser Use11.5 / 14Adds documented Deployment & Data Residency
- Meta Muse8.5 / 14A lighter documented profile than Asteroid
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded