CloudCruise: Product, team and how it was built
CloudCruise (Browser agents that build, run and repair healthcare portal automations, USA): Generally available with paid plans, November 2026. Evidence grade E: no revenue figure was checked. Team: 3 people in the Show HN launch post of April 2024 (2 listed founders); Y Combinator lists a team of 8 in November 2026, after a seed round. Launched 16 April 2024; still running.Added 6 November 2026 · updated 8 November 2026
- Outcome
- Generally available with paid plans, November 2026
- Evidence
- E no revenue figure was checked
- Launch
- 16 April 2024
- Status
- Still running
- Team
- 3 people in the Show HN launch post of April 2024 (2 listed founders); Y Combinator lists a team of 8 in November 2026, after a seed round
- Outside funding
- Y Combinator (Winter 2024); the company blog announced a US$5M seed round
- Main channel
- Not stated
What is not verified
No revenue figure for CloudCruise was found. We looked for a payment-processor profile, a founder interview with figures, and a filing. The home page's figures (5,000+ daily calls, 99.9% session availability) are one customer's numbers from the Trapeze case study. The US$5M seed round comes only from the company's own blog post. Every post header on the blog shows 19 March 2026, also on older posts, but the blog's post cards, which give older posts their own dates, list the seed post on 19 March 2026. A reader could look for a press report of the round or a funding filing. The home page shows a row of customer logos (named in the page code Charta, Medallion, Valerie Health, Substrate, Wisdom, Haven and Develop) under "Deployed at leading AI automation companies", and three quotes from customers' CTOs or founders; only Trapeze has a published case study, and we did not confirm the logos with the companies. This story is grade E: the business is real and its product, price and customers are described from public pages, but no revenue figure was checked. Aiolum shows it as an idea worth studying, not as proof that it makes money.
CloudCruise sells browser automation (software that operates a website through a real web browser, the way a person clicks and types) to other AI companies, mainly in health care. Its builder agent writes the workflow as a graph in its own DSL (domain-specific language, a small programming language made for one job), CloudCruise runs it on managed virtual machines, and a maintenance agent tries to fix a workflow when a portal changes its layout.
Adrian Ziegler and Felix Martin Eckert are the founders that Y Combinator lists; the company was in Y Combinator's Winter 2024 batch. Their Show HN post of 16 April 2024 was written by three people and launched a Chrome extension for general web workflows. By November 2026 the home page sold only healthcare automation (payer portals, EHRs, credentialing, revenue-cycle work), Y Combinator listed a team of 8, and the company blog announced a US$5M seed round.
No revenue figure was found, so this story has evidence grade E. The figures on the home page (5,000+ daily calls, 99.9% session availability) belong to one customer, Trapeze, and are not CloudCruise's revenue.

How much revenue does CloudCruise make?
Each row shows the evidence grade and the source numbers from the list at the end of this page. The grades are explained in How Aiolum verifies.
| Fact | Value | Grade | How we know | Sources |
|---|---|---|---|---|
| Status, November 2026 | generally available, with paid plans on the pricing page | C | Home page banner "CloudCruise is now generally available", read 6 November 2026. A Hacker News post by the founder's account, 23 June 2026, is titled "CloudCruise Is GA Now". | [1, 2, 6] |
| Revenue of CloudCruise | not published | C | No payment-processor profile, interview or filing with a revenue figure was found. See "What is not verified". | [1, 3, 4] |
| Prices | Starter free with 2 automation hours, then US$3 an hour; Scale US$700 a month | C | Pricing page, read 6 November 2026. | [2] |
| Customer figures on the home page | 5,000+ daily calls handled by Trapeze's AI voice agent for one customer; 99.9% session availability | C | These are one customer's figures, not CloudCruise's revenue or usage. The Trapeze case study is dated 30 September 2025. | [1, 7] |
| Team | 2 listed founders; team size 8 | C | Y Combinator company page data, read 8 November 2026. The Show HN post of 16 April 2024 was signed by three people. | [3, 4] |
| Funding | Y Combinator, Winter 2024; seed round of US$5M announced on the blog | C | The blog post names Floating Point, Meridian Street Capital, Twine and Refract Ventures as backers. Every post header on the blog shows 19 March 2026, also on older posts, but the blog's post cards, which give older posts their own dates (BADGER 26 June 2025), list the seed post on 19 March 2026. | [3, 5, 9] |
| Founded | 2024, San Francisco | C | Y Combinator company page. | [3] |
| AI models used | o3 and GPT-4.1 named; models from Anthropic, Google and OpenAI tested | C | Company blog post on prompt engineering: o3 for root-cause analysis in the maintenance agent, GPT-4.1 for narrow tasks. | [8] |
| Time from launch to the headline | 31 months | calc | Calculated: Show HN on 16 April 2024 to November 2026. | [4, 1] |
How to read this table: a grade C value is a statement by the company on its own pages, its posts or its Y Combinator page. A "calc" value was computed on this page. No value is grade A or B, and no row is CloudCruise's own revenue.
| Plan | Per month, billed yearly (US$) | Per month, billed monthly (US$) | Limit |
|---|---|---|---|
| Starter | 0 | 2 automation hours, then US$3 an hour; up to 3 seats; standard BAA | |
| Scale | 700 | unlimited seats, priority support, volume discounts, custom BAA | |
| Accelerate | custom | workflow build support, migration help, human maintenance fallback | |
| Enterprise | not listed | single sign-on, audit logs, reserved browser instances, custom SLAs |
One automation hour is 60 minutes of workflow runtime; the page says virtual machines, proxies and AI usage are included. No yearly price is published. Accelerate is listed as Custom; the Enterprise plan shows no price.
When did CloudCruise launch, and how did it grow?
From a free Chrome extension for any website to a paid platform for healthcare portals, in about two and a half years. Each entry carries its grade.
- Winter 2024CY Combinator batch. YC lists the company as founded in 2024 in San Francisco. [3]
- 16 April 2024CShow HN: a graph-based workflow builder for web agents, with a Chrome extension free for up to 200 browser actions a month. [4]
- 23 May 2025CThe company publishes BADGER, the type definitions of its workflow DSL, on GitHub (Hacker News post of that date). [6, 9]
- 30 September 2025CCase study: CloudCruise gives Trapeze's AI voice agent logged-in EHR sessions. [7]
- 23 June 2026CA Hacker News post from the founder's account: "CloudCruise Is GA Now". [6]
- November 2026CThe home page sells healthcare automation only; pricing from free to US$700 a month and custom plans; YC lists a team of 8. [1, 2, 3]
Why did CloudCruise work?
Four causes, in the order they mattered. Each one is a reusable lesson.
Generate code, not prompts at run time
The company's blog says most browser agents ask a language model what to do at every step, which is slow and hard to debug. CloudCruise uses the model to write a workflow graph once and then runs that graph as normal code.
Repair as a product
Portals change their pages. A maintenance agent classifies each failure (changed element, expired login, site down) and either fixes the step, asks the user to log in again or retries, so customers pay for upkeep they would otherwise do by hand.
One vertical
The 2024 launch was a general web tool. By 2026 the site speaks only to healthcare automation companies, with HIPAA terms (a BAA, Business Associate Agreement) in every plan.
How it was built
Only what the company's own pages and posts say. [10, 8, 9, 2, 7]
- A workflow DSL called BADGER: each node is a browser action (click, input text), a control step (if, loop) or a data step (extract, upload); edges give the order.
- A builder agent that fills in the graph from a plain-English task, and a maintenance agent that handles failures by an error taxonomy, with deterministic rules where possible.
- OpenAI's o3 for root-cause analysis and GPT-4.1 for narrow tasks; the team says it also tested models from Anthropic and Google.
- Runs on virtual machines with proxies included, credentials stored in a vault, and per-action logs, screenshots and run videos for audits.
- BADGER was written as an answer to the limits of plain Playwright scripts, according to the company's BADGER post.
Tools named in the sources
Each one opens its page in the Tools tab, with every story that uses it.
Who tried this before CloudCruise, and why did they fail?
Three companies in our database of failed startups promised to automate work inside other people's software, or to turn websites into data, and failed. The values below are from that database and are not independently verified.
Olive AI
The promiseAn AI workforce of software bots for hospital administration: prior authorizations, claims and billing.
Why it diedIt sold autonomous AI, but delivery was custom RPA (robotic process automation) that needed many engineers for each customer.
Dashblock
The promiseTurn any web page into an API without writing code.
Why it diedKeeping up with changing page structures drained its resources, and competitors and AI tools took the market.
Solid
The promiseOne API layer to connect a company's business systems.
Why it diedEach system needed custom work and support, and long sales cycles made the costs too high.
The failed startups come from our database of 1,749 dead startups, retold here in our own words. Its entries are AI-assisted summaries of public sources, so these cards carry no evidence grade.
What changed
Dashblock and Olive AI both broke on the same problem: websites and hospital systems change, and every change cost engineering time. CloudCruise sells the repair step itself, done by a language model, and limits the product to one industry. It sells to the AI companies that build agents, not to hospitals.
What did not change: the systems it automates belong to other companies. A portal that changes its terms or blocks automated logins can break the workflows that CloudCruise runs.
Can one person copy CloudCruise now?
The pattern, yes. Not alone.
Infrastructure for one hard kind of website in one industry, sold to the startups that build agents there. It needed a team, funding and healthcare compliance work, so it is hard for one person to copy as a whole.
Where it is still open: The same layer for government portals, insurance carrier sites in other countries, or freight and customs systems.
- 1Pick an industry where agents must use websites without APIs, and learn its portals.
- 2Write workflows as structured code that a model generates once, not as prompts run every time.
- 3Sell maintenance: detect and fix broken steps faster than the customer could.
- 4Settle the compliance contract (here, a BAA under HIPAA) before the first paying customer.
What could go wrong for CloudCruise?
The risks are the mirror image of the causes above.
What is not verified
No revenue figure for CloudCruise was found. We looked for a payment-processor profile, a founder interview with figures, and a filing. The home page's figures (5,000+ daily calls, 99.9% session availability) are one customer's numbers from the Trapeze case study. The US$5M seed round comes only from the company's own blog post. Every post header on the blog shows 19 March 2026, also on older posts, but the blog's post cards, which give older posts their own dates, list the seed post on 19 March 2026. A reader could look for a press report of the round or a funding filing. The home page shows a row of customer logos (named in the page code Charta, Medallion, Valerie Health, Substrate, Wisdom, Haven and Develop) under "Deployed at leading AI automation companies", and three quotes from customers' CTOs or founders; only Trapeze has a published case study, and we did not confirm the logos with the companies.
Built-in stealth and CAPTCHA solving
The home page sells "Built-in stealth": "Every run executes in a VM with laptop-like fingerprints and human-like input patterns. No bot flags, no CAPTCHAs, no blocked runs." The documentation describes a Captcha step that "solves Cloudflare Turnstile, reCAPTCHA v2, hCaptcha, AWS WAF CAPTCHA, and a site's own text or image challenge", and a managed proxy pool (a random proxy, one in a chosen country, or the one closest to a target IP); Enterprise customers may also bring their own proxies. The sites automated are payer portals, EHRs and credentialing boards, logged in with credentials that the customer stores in CloudCruise's vault; in the Trapeze case study these are logins for the EHRs of Trapeze's clinic customers. CloudCruise's terms forbid customers to "Violate any law, regulation, or third-party terms of service (e.g., scrape sites in breach of their rules)". Some portals' terms may not allow automated access or CAPTCHA solving; we did not check any portal's terms.
Earlier use cases involved people's data
The Show HN post of April 2024 listed early users' tasks such as finding "the contact information of fire department chiefs in the US" and "background research on people they have meetings with". The product sold in November 2026 is healthcare portal automation, and we found no such use case on the current site.
Competition and model providers
Plena Health and Develop Health, with stories on this site, sell their own AI for payer portals and prior authorization; CloudCruise sells the browser layer to companies like them. The company's prompt-engineering post names OpenAI's o3 and GPT-4.1 and says it tested Anthropic and Google models, so it depends on these providers, while running workflows as code limits model calls at run time.
Sources
The grade next to each source is the best grade that the source can support. "Read in full" means that the text of the page was downloaded and the numbers were checked in it.
- C CloudCruise home page, read 6 November 2026Read in full. Source for the product, the GA banner, the customer figures, the testimonials and the stealth text.
- C CloudCruise pricing page, read 6 November 2026Read in full.
- C Y Combinator company page: CloudCruiseRead the company data in the page: founders, team size, batch, founding year and description.
- C Show HN, 16 April 2024: CloudCruise (YC W24) - Graph-Based Workflow Builder for Web AgentsRead the post text through the Hacker News Algolia API (item 40052809). The founders' launch post.
- C CloudCruise blog: CloudCruise Raises $5M Seed Round for Healthcare AutomationRead in full. The page date is not reliable (see watch).
- C Hacker News search results for CloudCruise (Algolia API)Read the list of posts with dates: the GA post of 23 June 2026 and the BADGER posts of 2025. The GA post links to an X post, which was not opened.
- C CloudCruise case study: Trapeze, 30 September 2025Read in full. A customer's figures, published by the company.
- C CloudCruise blog: Prompt Engineering at CloudCruiseRead the sections on model choice and the maintenance agent.
- C CloudCruise blog: BADGER: Graph-Based Browser AutomationRead the sections on the DSL and Playwright.
- C CloudCruise blog: A Better Way to Build Browser AgentsRead in full. Source for the builder and maintenance agents.
- C Database of failed startups, local exportLocal export. AI-written summaries, not independently verified. Source for the failed predecessors.
- C CloudCruise docs: Captcha nodeRead in full, 8 November 2026. Source for the CAPTCHA solver in watch.
- C CloudCruise docs: Workflow SettingsRead the sections on captcha and proxy settings, 8 November 2026.
- C CloudCruise Terms of Service, effective 29 June 2025Read the section on acceptable use, 8 November 2026.
Run this company? Send corrections or a verified figure to aiolum@livyden.com