Ghostship

What do they actually do

Ghostship runs AI browser agents that explore your web app like a user, try common and adjacent clicks, and report bugs with session replays and reproduction steps. Teams connect it to GitHub so agents run on pull requests (or on a provided URL), and the UI shows what the agent did alongside a list of issues it found tryghostship.dev YC launch HN launch thread.

In practice, you set up a GitHub‑native workflow, optionally define critical user flows that should always be tested, and get a per‑PR report with session replay and steps to reproduce failures. The product focuses on running as an automated check tied to PRs and CI, aimed at catching regressions before merge tryghostship.dev LinkedIn summary.

It’s early stage (two‑person founding team) with reported initial traction of about $13.2k MRR shortly after launch, so expect active iteration on reliability, coverage and integrations as they mature YC company profile.

Who are their target customer(s)

• Early-stage engineering teams at fast-shipping startups: They lack bandwidth for manual QA and need lightweight checks that run on every PR to catch regressions without slowing releases. A GitHub‑native agent with session replays fits their workflow tryghostship.dev.
• Mid‑sized SaaS product teams with frequent releases: Complex user journeys and rapid shipping create regressions that slip to production. Autonomous agents that explore flows and surface issues with replays reduce misses vs. ad‑hoc manual testing HN launch.
• QA engineers/test-automation owners maintaining brittle suites: They spend time fixing flaky tests and pruning cases instead of finding new bugs. Exploratory agents with concrete repro steps and replays can cut triage time Fondo writeup.
• CI/CD or release engineers responsible for PR gates: They need reliable, fast automated checks that integrate with PRs without adding heavy maintenance. A GitHub/CI‑oriented agent helps block regressions earlier LinkedIn summary.
• Product managers and customer‑facing teams focused on UX quality: Intermittent or hard‑to‑reproduce UX bugs are difficult to prioritize. Session replays and step‑by‑step failure reports make validation and prioritization faster YC launch.

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

• First 10: Run hands‑on pilots with YC founders and fast‑shipping startups; sit with their engineers during a few active PRs, tune flows, and deliver short case studies that highlight real bugs found and time saved YC profile.
• First 50: Publish a one‑click GitHub integration and a simple free tier with a few agent runs, plus guides and short demos; amplify via developer communities and early customer case studies to reduce trial friction tryghostship.dev.
• First 100: Layer developer‑focused ads/newsletters and light outbound to mid‑market SaaS, offer time‑boxed pilots with clear success criteria, and pursue co‑marketing/integrations with CI/ops platforms and accelerators to drive bundled intros.

What is the rough total addressable market

Top-down context:

Broad software testing/QA is estimated around $50–60B in 2024 depending on scope, while “automation testing” reports show roughly $17.7B in 2024 with strong growth GMI Fortune Business Insights.

Bottom-up calculation:

About 27M developers worldwide implies ~2.7M teams at 10 devs per team; if 30–50% are web/CI‑oriented and a fit for GitHub‑native exploratory QA, that’s roughly 0.8–1.35M target teams Evans Data.

Assumptions:

• Average shipping product team ≈ 10 engineers.
• 30–50% of teams build web apps and use GitHub/CI in a way that fits this product.
• A meaningful subset will pay for PR‑driven exploratory browser agents at SMB/mid‑market tooling price points.

Who are some of their notable competitors

• mabl: AI‑first end‑to‑end testing that creates, runs, and self‑heals tests with CI/CD integrations; overlaps on agentic testing and PR checks but is a broader enterprise platform focused on test creation/maintenance mabl CI docs.
• Testim: AI‑driven UI test automation with AI locators and agentic workflows to generate/stabilize tests and run in CI; competes on authoring and CI runs rather than autonomous exploratory crawling docs.
• Autify: No‑code/low‑code AI test platform (incl. Autify Genesis) for cross‑browser/mobile with CI integrations; emphasizes test scenario creation over autonomous exploratory agents product docs.
• Ghost Inspector: Record‑and‑replay browser testing with schedulable runs, CI hooks, and screenshots/videos; similar session replays but centered on recorded tests, not autonomous exploration docs.
• Applitools: Visual AI for detecting UI regressions with tight CI integration; typically complements functional tests rather than exploring user journeys autonomously platform CI guide.