Let the agent gather context
Pull research, files, emails, and records into the run so the output is grounded without manual tab work.
Sources stay in the same thread
Findings become workspace memory
Less jumping between apps
Tool actions
Let the agent act through tools after you approve the path.
Integrations are not a logo wall. They are action surfaces the workspace agent can use to gather context, send messages, update files, publish outputs, and automate repeatable work.
120+ searchable actions for connected apps
Email, files, scraping, storage, and publishing in one run
Permissioned tool calls before external actions happen
Less tab-hopping, more execution from workspace memory
Let the agent take approved actions
Send, save, publish, scrape, or update through connected tools after the user approves the path.
Permissioned external steps
Email, storage, publishing
Action logs beside the docs
Chain tools into workflows
Run multi-step tasks like lead research, outreach drafting, content production, and reporting from one workspace.
Lead research
Content ops
Reporting loops
Keep one source of truth
The docs remain the memory and output layer, so tool results do not disappear into disconnected chats.
Docs as memory
Tasks as execution history
Outputs stay linked
Why it matters
Tool actions make the agent useful outside the editor
The point is not a long list of names. It is compressing the path from "the agent understands the task" to "the work is done or ready for approval."
Workflow loop
1
Gather context from approved connected tools
2
Use docs and models to plan the right output
3
Execute, update, or send after the user approves
Keep actions where workspace memory lives
Tools support the task. Docs remain the source of truth, and outputs stay attached to the workspace.