> ## Documentation Index
> Fetch the complete documentation index at: https://docs.stably.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Why Migrate from Classic
> How the Stably Web Editor + CLI compares to Classic — and why teams that migrate see dramatically lower costs.
The Stably platform has evolved. The [Web Editor](/stably/stably-web-editor), [CLI](/stably-cli/quickstart), and [Cloud Runner](/run-tests/run-tests-on-cloud) are built on a fundamentally different architecture: your tests are real Playwright code stored in Git, maintained by AI agents that fix root causes instead of patching symptoms at runtime.
Teams that have migrated consistently report **at least 2x lower AI costs** for the same coverage — with many seeing 50–80% savings.
Your Classic tests won't appear in the Web Editor automatically — they need to be [migrated](/no-code-tests/migrate-to-agent-2) first. Migration is AI-assisted and typically takes minutes, not days.
## At a Glance
| | Classic | Web Editor + CLI |
| ------------------- | ------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------- |
| **Test format** | No-code steps in Stably UI | Playwright TypeScript in Git |
| **Editing** | Web UI only | Web Editor, CLI, or any IDE |
| **Auto-fix** | Runtime auto-heal on every run | Post-run [Autofix](/getting-started/autofix-for-tests) — analyzes root cause, applies a permanent code fix, validates in a real browser |
| **AI cost** | Auto-heal burns tokens on every execution | AI only invoked when something actually breaks |
| **Cost visibility** | Limited | Per-run cost breakdown with [spending limits](/pricing#usage-limits) |
| **Flakiness** | Auto-heal can mask underlying timing issues | Deterministic tests with proper wait strategies |
| **Debugging** | Limited visibility | Full traces, screenshots, DOM snapshots, commit-aware triage |
| **Execution** | Stably Cloud only | Anywhere — Cloud, CI, or locally with `npx playwright test` |
| **Version control** | None | Git-backed with full history |
## Key Benefits
### Dramatically Lower Costs
Classic auto-heal triggers AI calls on every test execution. These costs compound quickly — especially on scheduled suites — and they're hard to predict because the amount of healing varies run to run.
With the Web Editor + CLI, tests are resilient from the start. AI is only invoked when something genuinely breaks, and [Autofix](/getting-started/autofix-for-tests) fixes the underlying code so the same issue doesn't recur.
One team eliminated **80–90% of their AI calls** after migrating. The auto-heal overhead that had been silently inflating their bill disappeared because their tests no longer needed constant runtime patching.
Every run now includes a detailed cost breakdown — AI tokens and browser minutes — so you can see exactly where spend goes. You can also set [spending limits](/pricing#usage-limits) to prevent surprise bills.
### Smarter Auto-Fix
Classic auto-heal patches broken selectors live during each run. If you don't accept the fix, it triggers again next run, and again the run after that.
[Autofix](/getting-started/autofix-for-tests) takes a different approach:
* **Root cause analysis** — compares recent commits with past runs to pinpoint why tests broke, not just what broke
* **Permanent fixes** — edits your test code (or application code) rather than patching at runtime
* **Validated with proof** — re-runs the fix in a real browser and provides screenshots, traces, and DOM snapshots before anything reaches your codebase
* **Delivered as a PR** — when your [repo is connected](/stably2/bring-your-own-repo), fixes arrive as reviewable pull requests
* **Grouped by root cause** — catches shared issues across multiple failing tests instead of fixing them one by one
### Tests Live in Your Codebase
After migrating, your tests are standard Playwright TypeScript in a Git repository — either your own or a Stably-managed one.
* **Edit anywhere** — Web Editor, VS Code, Cursor, or any IDE. Changes sync automatically.
* **Review changes** — standard code review workflows with diffs, PRs, and branch history
* **Run anywhere** — `npx playwright test` works out of the box, no Stably dependency required
* **No lock-in** — your tests are Playwright code you own. If you ever leave Stably, the tests still work.
### Less Flakiness, Faster Runs
Classic auto-heal can mask flakiness — a test with a timing issue might "pass" after a few retries, but the underlying problem persists and the retries burn credits every run.
The Web Editor + CLI produce deterministic tests with proper wait strategies, stable selectors, and the SDK's [AI locator](/stably-sdk/ai-locator) when needed. When a test fails, it fails for a real reason — and you get full [traces with screenshots](/run-tests/test-results) to debug it.
Without retry loops, tests also run faster. The Cloud Runner uses auto-scaling infrastructure for parallel execution, and the CLI can connect to [dozens of cloud browsers](/stably-cli/cloud-browser) simultaneously.
### Flexible Test Creation
Classic offers one way to build tests: the no-code editor in the browser. The Web Editor + CLI give you multiple paths depending on your workflow:
| Approach | Best for |
| ----------------------------------------------- | ----------------------------------------------------------------------------------------------- |
| [Web Editor AI chat](/stably/stably-web-editor) | Quick creation — describe what you want in natural language, iterate conversationally |
| [Browser recorder](/stably2/web-quickstart) | Capturing real user flows by clicking through your app |
| [CLI (`stably create`)](/stably-cli/create) | Terminal-first workflows, CI automation, creating tests alongside code changes |
| [Your IDE + Stably SDK](/stably-sdk/overview) | Full control — write Playwright tests directly with AI coding assistants like Cursor or Copilot |
You can create multiple tests in a single session, and the AI references patterns from your existing tests for consistency.
## FAQ
They remain available but won't appear in the Web Editor until you [migrate them](/no-code-tests/migrate-to-agent-2). Migration is AI-assisted — an agent converts your Classic steps to Playwright code, applies compatibility fixes, and commits the result for your review. Most teams complete migration in a single session.
No. The Web Editor provides the same visual, no-code editing — drag-and-drop steps, click-to-edit selectors, AI assistance — but backed by real Playwright code and Git version control. You can switch between the visual view and code view at any time.
No. Your Classic scheduled runs continue as-is until you're ready to switch. You can set up new schedules for migrated tests in parallel, verify they work, and then retire the Classic ones on your own timeline.
Yes. Migrate one suite at a time. Classic and Web Editor tests coexist in the same project during the transition — there's no all-or-nothing cutover.
The AI-assisted migration uses a small amount of AI credits — negligible for most projects. Contact support if you'd like migration credits applied to your account.
## Ready to Migrate?
Step-by-step instructions — AI-assisted or manual export
See the Web Editor in action
Create and run tests from your terminal
See how the new Autofix works