Free
€0no card needed
See what Holmes finds on your product.
Start free- 1 product
- 5 runs a month
- Unlimited users
- Self-serve setup

Your all-in-one QA platform for web apps, mobile apps, and APIs.
Runs autonomously from day one.
52%
of customers quit a brand after a bad experience
56%
of software testing is still done by hand
1 day
a week on testing for 4 in 10 developers
€250K+
per hour of downtime for 9 in 10 larger firms
A case can start with a pull request, a ticket, a bug hunt, or the nightly regression run. Pick one to see what starts it, what Holmes reads, and where the verdict lands. Or click the tools to build your own.
STARTS A RUN
READS
REPORTS TO
HOLMES
When a pull request opens, Holmes reads the code change and the ticket, tests what the pull request changed on its preview, and reports back with a check on the pull request and a message in Slack.
“We wanted our smoke tests automated: big groups of tests that run once a week, so our manual QA people can focus on new features.”
Manu FourneauSagetapA run per ticket
Holmes picks up the Linear ticket and tests it on that ticket’s own QA environment.
One test, several portals
A single test follows a flow from the vendor side to the executive side.
Weekly regression
The smoke suite runs on its own, so the QA team spends its time on what is new.
Design against build
Holmes compares the Figma design with what actually shipped.
Unlimited users on every plan. Priced on how much Holmes tests, never on seats.
€250 in free credits on every plan.Sign up and spend it on any run, before you pay a cent.
Claim your €250€0no card needed
See what Holmes finds on your product.
Start free€199per month
For a small team with one product to keep green.
Get startedfrom€499per month
For teams that ship every week.
Book a demoCustomusage-based
For organisations with security and compliance needs.
Get a quoteHolmes is an AI QA platform. AI agents test your web app, mobile app, and APIs the way your customers use them, and report every issue with a recording, screenshots, and the steps to reproduce it. It covers the moments that matter: regression testing, feature and exploratory testing, plain-language summaries of what changed, and persona and accessibility testing. Setup takes under ten minutes: add your URL and a test account, and most teams see their first results within the hour.
Web applications in a real browser (Chromium, Firefox or WebKit), native Android and iOS apps on emulators and simulators, and the API requests behind every journey, so a failing endpoint never hides behind a screen that looks fine. Holmes also walks your product as the personas you build for, and runs accessibility audits against WCAG success criteria with keyboard navigation, which helps you prepare for the European Accessibility Act. An automated audit supports, but does not replace, a formal accessibility certification.
No. Holmes explores your product and builds test flows from what it finds, from your tickets, or from a plain-language description. There is no Selenium, Cypress, or Playwright code to write or maintain, and you can import existing flows if you have them. That is the main difference with traditional test automation: scripted end-to-end tests break when a selector or a label changes, while Holmes agents understand the page, adapt when the UI changes, and repair their flows. They also look beyond the script and investigate paths nobody wrote down.
Holmes saves your critical customer journeys, like sign-up and checkout, and replays them on every release, on a schedule, or on a preview deployment. When a journey breaks, you see the exact step that failed, with a recording and the console and network logs. For new work, Holmes reads the ticket or pull request, tests the feature against its requirements, and explores the edge cases a happy-path test would miss, so you get a verdict with evidence before you merge.
Connect GitHub, GitLab, or Bitbucket and Holmes tests pull requests, deploys, and preview deployments; on GitHub and Bitbucket the result comes back as a check and a comment on the pull request. Runs can also start on a schedule, from a webhook in your pipeline, from a ticket in your ticketing system, from Slack, or from coding agents like Claude and Cursor through MCP. Holmes reads context from your code, tickets, specs, designs, and analytics, and sends findings to pull requests, tickets, Slack, Microsoft Teams, and email.
Every issue goes through multi-stage validation before it reaches you: deterministic confidence scoring, screenshot evidence, and optional LLM-based verification. Borderline issues are re-tested by an independent evaluator, so what lands in your inbox is a bug worth looking at, with the evidence to prove it.
You can start for free with one product, five saved flows, and five runs a month, no credit card needed. Paid plans start from €199 ($219) per month. Every plan includes unlimited users, and pricing is based on how much Holmes tests, never on seats. A small group of founding customers gets design-partner pricing in exchange for feedback.
All data is encrypted at rest and in transit, and test credentials are stored with AES-256 encryption. Holmes runs in isolated cloud containers that are destroyed after each session. We are working towards SOC 2 Type II and ISO 27001 certification; the Trust Center describes how we handle your data, code, and credentials.
Open your first case today. Holmes reports back within minutes.