How to find
- In the left-hand menu click Scrapers
- Select Scraper Studio
How it works
Every scraper in Scraper Studio performs two core operations regardless of how it’s created:- Interaction: Navigating to a target URL, handling pagination, clicking elements, or sending HTTP requests.
- Parsing: Reading the page content (HTML or JSON) and extracting structured fields into a defined output schema.
Development Modes
Scraper Studio offers two ways to build that logic:| Mode | How it works | Best for |
|---|---|---|
| AI Agent | Describe the data you want in plain language. The AI generates a schema, then writes the scraper code for you. | No-code users, fast prototyping |
| IDE | Write and test JavaScript directly in a browser-based code editor with debugging tools built in. | Developers who need full control |
When to use Scraper Studio vs. other Bright Data products
| Scenario | Recommended product |
|---|---|
| Need data from a popular site with zero setup | Datasets Marketplace |
| Need a scraper built and fully maintained by Bright Data | Managed Services |
| Need to build a custom scraper yourself using AI or code, and run it on Bright Data’s infrastructure | Scraper Studio |
Use Scraper Studio when the data you need isn’t in the Marketplace, you want ownership of the scraper logic, and you don’t want to manage proxies or infrastructure yourself.
Key trade-offs: AI Agent vs. IDE
| AI Agent | IDE | |
|---|---|---|
| Setup time | Minutes - describe and run | Longer - write, test, debug |
| Code control | Limited - AI writes the code | Full - you own every line |
| Customization | Via Self-Healing Tool prompts | Direct JavaScript editing |
| Best for | Fast scraper creation, non-technical users | Complex logic, multi-stage scrapers, performance tuning |
Cloud infrastructure trade-offs
Scraper Studio runs entirely on Bright Data’s infrastructure. This means:- No server setup or proxy management - all included.
- Billing is based on page loads (CPM) - 1 CPM = 1,000 page loads. Browser Workers typically consume more page loads than Code Workers for equivalent tasks.
- Parallel job limit - up to 1,000 batch jobs can run simultaneously. Additional jobs queue automatically.
- Snapshot retention - batch collection results are stored for 16 days; real-time results for 7 days. Export your data before expiry.