Any site becomes callable JSON.
Point Writ at a page you are authorized to use (even behind a login or heavy JavaScript) and get back clean, structured JSON you can call by API or MCP.
The pages ordinary tools can't reach
A recorded browser path plus AI navigation drives the real site the way a person would, so the cases that stop brittle integrations just work. Writ handles these as authorized actions on accounts you control.
- ✓Login walls
- ✓TOTP + mailbox-OTP 2FA on your own accounts
- ✓Verification on your own authenticated accounts
- ✓Dynamic / JS-heavy pages
- ✓Pagination & infinite scroll
- ✓Multi-tab flows
A list-and-detail page becomes a structured table
Record the path once. An extract step reads the list, a navigate-and-extract pair pulls each detail page, and a click-next loop follows pagination, then the records land in a table you can call by API or MCP.
Recorded workflow
Structured records
Build the extractor however you like
Record it by clicking through the site, describe the goal and let Scribe, our AI concierge, build it for you, or discover it from the page's own network calls.
Extract a list, page by page.
# Workflow steps (recorded or AI-built):
# navigate → fill (login) → twofa → evaluate (extract list)
# → click (next page) → wait_for_change → extract → repeat
POST https://api.usewrit.app/v1/catalog-extract/run
Authorization: Bearer wt_…
→ { "records": [ { "title": "…", "price": 12.5, "url": "…" }, … ], "pages": 8 } 30+ step types compose into a durable extractor: navigate, click, fill, type, select, wait_for_change, extract, evaluate (JS), api_call.
Free on your machine, or managed in the cloud
The app is free and runs on your own machine; data never leaves. Reach for the managed cloud when you want it always-on.
Local · BYO agent: free
- ✓Runs on your own machine; extracted data never leaves your tenant
- ✓Reaches intranet and VPN-only sites the cloud can't
- ✓Bring your own AI keys: AI steps run unbilled
Managed cloud: metered
- ◍Always-on cloud fleet, no machine of yours to keep running
- ◍Running time ~$0.12 per browser-hour, metered by time — included usage first, then your balance
- ◍Managed AI is metered by tokens used, with the price shown before you run, drawn from your prepaid balance
Local and BYO runs carry no compute charge. You only pay when you choose the managed cloud.
When the site moves, the extractor keeps up
Selectors drift; sites redesign. Writ heals the target automatically wherever it runs, local or cloud. On a hard failure, the managed cloud goes a step further and rewrites the recipe with AI. Cloud AI auto-repair is a cloud-only feature; self-healing targets run everywhere.
Web data extraction questions
Is web data extraction legal?
It depends on the site and the data; Writ is built for authorized access (public data, your own data, or your own accounts), and you're responsible for respecting each site's terms.
Does it handle logins and 2FA?
Yes, on your own accounts, with TOTP and mailbox-OTP handled by Personas.
What about sites that change?
Recorded recipes plus AI navigation adapt when the page moves; self-healing targets run everywhere, and on a hard failure the managed cloud rewrites the recipe with AI (a cloud-only feature).
How do I get list, detail, and paginated records?
Compose a workflow: an extract step reads the list, a navigate + extract pair pulls each detail page, and a click-next loop follows pagination. Record it once, then run it as a callable endpoint.
Turn any site into structured JSON
Record an extraction once and run it as a callable endpoint, on your own machine, free.