Enable an ability
Owners switch on the plugins the workspace should have. Each has a try-it button.
Product · Build
Let the model do things.
AI chat plugins that work on every provider: search the web, read pages, draw charts, or run your own tool.
Someone asked about a price that changed this morning, and the answer came back with the page read.
- Web search, webpage reading, chart rendering
- Same plugins on every connected provider
- Custom plugins in client-side JavaScript
3 built-ins, 14 providersweb search, page reading, and charts on every tool-capable model you enable
How it works
Ask normally
Ask about something current, paste a link, or request a chart. The model calls the tool itself.
Read the result
Search results, page content, or a rendered chart arrive inside the reply.
01
One plugin set, eleven providers.
Tool calling normally means building against one vendor. Here the same three built-ins work on every model that supports tools, whichever key it runs on.
- Built-ins: web search, read webpage, render chart
- Tool-capable models labeled in the picker
- Runs limited per message to keep costs bounded
02
Bring your own tool.
Custom plugins are small JavaScript functions that run in your browser during the chat. Wire an internal lookup or a calculator without touching a server.
- Client-side execution, no deploy
- Per-plugin settings for keys and endpoints
- Owner-controlled, workspace-wide
Questions, answered
They work on every connected provider whose model supports tool calling. The picker labels which models those are.
A small JavaScript function that runs in your browser during the chat. Teams use it for internal lookups or calculators, with no server to deploy.
Tool runs are limited per message to keep costs bounded, and owners control which plugins the workspace has.
Size it with a free tool
Validate a tool definition
Catch schema errors before an API call rejects them.
Calculate API rate limits
Turn documented per-minute limits into safe concurrency.
Why startups need a model-agnostic AI workspace
The case for keeping model choice open as your team grows.