Asteroid

What do they actually do

Asteroid runs AI-controlled browser sessions to automate manual, browser-based back‑office tasks on sites that don’t expose APIs (logging in, navigating portals, filling forms, moving or scraping data). It’s used today in regulated and legacy workflows like insurance quoting and clinical data entry; public case studies include Meshed (automated carrier quoting) and Delfa (putting voice transcripts into medical forms) Meshed, Delfa, site.

Customers build agents with a graph-based visual builder, configure an Agent Profile with proxies, cookies, credentials and TOTP/2FA, then run agents from the dashboard or via SDK/API. Executions run in real browsers hosted by Asteroid, with live observability, logs/recordings, and results returned via SDKs or webhooks docs, graph agents, profiles/TOTP.

The platform emphasizes reliability (persistent sessions, 2FA handling), auditability (execution records and failure detection), and enterprise controls (concurrency, SLAs, BAA/HIPAA claims) docs, site. They’ve begun shifting from a white‑glove service to self‑serve through the new builder and templates so non‑developers can assemble automations docs, founder note.

Who are their target customer(s)

• Insurance quoting operations at carriers or insurtechs: Teams manually log into multiple carrier portals to pull/submit quotes; work is slow, inconsistent, and error‑prone. Asteroid has a public case showing automated quoting across legacy carrier sites Meshed.
• Healthcare teams and digital health startups needing to enter clinical data into EHRs: Clinical notes or transcripts often must be typed into legacy EHR UIs without APIs; it’s slow, requires oversight, and must be handled in a HIPAA‑safe way Delfa, site.
• Enterprise back‑office teams (finance, procurement, benefits) dealing with legacy vendor portals: One‑off browser workflows resist integration; teams rely on brittle scripts or manual rekeying without observability or audit trails docs.
• Product/automation teams building voice or conversational agents that must act on legacy systems: The conversational layer can decide what to do, but there’s no reliable way to push actions into portals/EHRs without APIs; they need a robust “last mile” to execute in web UIs blog.
• Security/compliance/IT reviewers at regulated organizations: They need to understand credential storage, 2FA handling, run auditability, and require SLAs and compliance attestations (BAA/HIPAA, SOC2 claims) profiles/TOTP, site.

How would they acquire their first 10, 50, and 100 customers

• First 10: Founder‑led, white‑glove pilots with a few high‑value healthcare and insurance teams; Asteroid builds the initial agents, demonstrates reliability and measured time savings, and supplies BAA/HIPAA and audit logs to speed security review Meshed, Delfa, docs.
• First 50: Package repeatable templates (e.g., insurance quoting, EHR form filling) in the graph builder and cookbook, pair with a short onboarding service and a solutions engineer to keep time‑to‑value under a few weeks cookbook, graph agents.
• First 100: Scale through partners (RPA integrators, voice/conversational platforms, SIs) and procurement‑friendly motions (SLAs, BAA/SOC2 evidence, credential/2FA documentation) to resell/embed Asteroid runs voice agents blog, site.

What is the rough total addressable market

Top-down context:

Asteroid sits within enterprise automation where RPA is ~$18.2B in 2024, with adjacent spend in back‑office automation (~$5.5B), EHR (~$28–33B), and insurtech (~$25–26B). These figures show overall budget pools but overlap significantly, so they’re not additive RPA, back‑office, EHR, insurtech.

Bottom-up calculation:

Using published pricing, a mid‑tier enterprise package around $3k/month implies $3.6M ARR at 100 customers, $18M at 500, and ~$36M at 1,000; this frames a realistic near‑term revenue path while templates and capacity scale site. Scenario slicing of the broader pool (5–20%) yields roughly $3.9B–$15.6B for browser‑only, compliance‑sensitive automations, but actual capture depends on legacy persistence and execution.

Assumptions:

• Only a subset of automation spend targets browser‑only legacy UIs in regulated settings (5–20% share used for scenarios).
• Legacy systems without APIs persist long enough for agents to displace manual work at scale.
• Mid/upper‑tier pricing (~$3k/month) remains viable for enterprise workflows requiring auditability and compliance.

Who are some of their notable competitors

• UiPath: Large enterprise automation platform spanning RPA, AI, and orchestration with governance; widely adopted across regulated industries UiPath, product.
• Automation Anywhere: Cloud‑first intelligent automation/RPA suite with agentic process automation, document automation, and orchestration features products.
• Microsoft Power Automate: Low‑code automation across cloud and desktop with built‑in RPA (desktop flows), hosted RPA, and 1,000+ connectors; strong enterprise governance Microsoft.
• Robocorp: Python‑centric, open‑source RPA stack with a cloud Control Room and libraries for browser/desktop automation; popular with developer teams portal, rpaframework.
• Browserbase: Serverless headless browsers and tooling for AI agents and web automation, with session replay, proxies, and compliance options (SOC2/HIPAA) site, docs.