A visual way to build Playwright — and plain Playwright on the way out
Orchestra is not a replacement for Playwright. It is a studio that builds flows visually and exports them as ordinary Playwright TypeScript, so the fast path and the portable path are the same path.
Playwright is excellent. Writing it is still slow.
Nobody sensible argues with Playwright as a runtime — it is fast, it is reliable, and its auto-waiting removed a whole genre of flaky test. The cost is upstream of that: finding a selector that will still match next month, restructuring around a loop, re-running the file to discover the page moved on, and doing all of it in an editor while the actual page lives in another window.
`playwright codegen` helps and then stops helping. It emits a linear transcript of what you did, with no loops, no conditions and no structure — useful as a starting point, awkward as a thing to maintain.
What a visual builder changes
Orchestra keeps the page and the flow side by side. You click an element and see immediately what the proposed selector matches and how many things it matches, which is the moment selector bugs are cheapest to fix.
- Structure while you record
- Loops, conditions, try/catch and assertions are steps you add as you go, not a refactor you perform afterwards on generated code.
- Run any step in isolation
- Re-run one step against the current page state instead of re-executing the file from the top each time.
- Selectors you can interrogate
- See the matches highlighted, then loosen or tighten before committing — rather than finding out at run time.
- Readable export
- The output is TypeScript a person would write: named steps, real Playwright locators, no Orchestra imports and no runtime to install.
No lock-in, by construction
The exported file has no dependency on Orchestra. No SDK, no licence check, no proprietary format, nothing that phones home. It is a Playwright script — `npx playwright test` runs it, CI runs it, a developer who has never heard of this app can read and extend it.
So the question is never "can we get our automations out". They were always out.
When to use which
If your team writes Playwright comfortably and lives in the editor, keep doing that — Playwright is the better tool for people who are already fluent in it, and this page is not going to pretend otherwise.
Orchestra earns its place when the person who understands the process is not the person who writes TypeScript, when you want a working automation this afternoon rather than this week, or when you want the two to collaborate on one artifact: built visually, handed over as code.
Common questions
Is Orchestra a replacement for Playwright?
No. It is a visual builder that generates Playwright. The code it exports uses Playwright as its runtime, so you are adopting Playwright, not avoiding it.
How is this different from Playwright codegen?
Codegen records a linear transcript of your actions with no loops, conditions or error handling, and stops there. Orchestra lets you build structure as you record, re-run individual steps, inspect what a selector matches before committing to it, and edit the flow afterwards.
Is the exported code actually readable?
Yes — plain TypeScript with real Playwright locators and named steps, with no Orchestra imports, no SDK and no licence check. It is meant to be handed to a developer and extended by hand.
Can I run the exported code in CI?
Yes. It is an ordinary Playwright script with no dependency on Orchestra, so it runs anywhere Playwright runs — GitHub Actions, GitLab CI, a container, a cron job on a server.
Do I need Playwright installed to use Orchestra?
Not to build and run flows inside the app — Orchestra ships its own browser runtime. You need Playwright only when you take the exported code elsewhere and run it yourself.