Your assistant acts on what your company knows.
Ask a question and get an answer drawn from your own documents, with inline citations and the asker's own permissions applied every time. It also does the work for you: drafting documents, running deep research, and turning data into charts. Model 'Auto' keeps sensitive work on local models, so nothing leaves your perimeter.
Chat answers questions and gets the work done.
Every answer traces back to its source. Quick actions create and research on your behalf, and model routing keeps sensitive work on local models inside your walls.
Grounded and cited
Every answer links back to the documents it came from, and each question runs under the asker's own permissions, so people see only what they are cleared to see.
Draft documents and run research
Draft a document as a Doc, Sheet, or Slides, run deep research across your company knowledge, or turn data into charts rendered on your own machines. No data leaves to build them. Image generation activates only when a local image model is registered.
Local models by default
'Auto' picks a local model by default and reaches for the cloud only where your workspace policy allows it, so sensitive work stays inside.
Answers from the sources you pick
Point the assistant at the right sources or attach files to any message, then check the citation behind every claim it makes.
Start in a workflow
When a conversation becomes a task you repeat, promote it to a governed workflow in one step.
Nothing leaves
The assistant runs on your own infrastructure, so your data stays inside the perimeter and never travels to a vendor.
See the assistant run on your own knowledge.
You get cited answers you can trace and quick actions that draft and research for you. Model 'Auto' keeps sensitive work on local models inside your perimeter.