Built in, not an integration
A full AI workspace, right at your desk
One chat for everything. Each message gets the kind of answer it needs, whether that is a reply, a side-by-side comparison of models, a multi-step pipeline or an image, so nobody has to pick a mode first. It lives in the same office your team works in.
Model APicked
- 1. Feature flag depends on Thursday's migration.
- 2. Announcement copy not yet reviewed.
Model B
- 1. Feature flag depends on Thursday's migration.
- 2. Announcement copy not yet reviewed.
Routing picks the model for each task, so nobody has to.
- Meet
- Capture
- Act
One chat, no mode picker
Every message gets the answer it needs
Ask a quick question, ask to see two models answer the same prompt, hand over a multi-step job, or describe a picture. The workspace works out which of those you meant for each message, and routes it to a suitable model. If you would rather choose, pin a model and the whole chat uses it.
- A reply, a side-by-side comparison, a multi-step pipeline or an image
- Image requests are handed to Studio
- Pin a model whenever you want to choose
- Chats are titled for you and searchable later
- What changed in the onboarding flow this sprint?A reply · Answered by the model routed for the task.
- Show me how two models would word this announcement.A comparison · Two models answer side by side.
- Summarise the thread, then draft the follow-up email.A pipeline · Several steps run in order, each shown.
- A calm cover image for the launch post.An image · Handed to Studio to generate.
Prefer a particular model? Pin it, and every reply in the chat uses it.
Bring your material
Attach it, drag it in, or say it out loud
Attach documents including Word files, images including iPhone HEIC photos, and PDFs that are read page by page. Pick files from Drive, drag and drop them, or dictate your prompt. Save your custom instructions once, and keep the prompts you reuse in a library.
- Word documents, images (HEIC too) and PDFs read page by page
- From your computer, from Drive, or dragged in
- Voice dictation, custom instructions and a prompt library
- launch-brief.docxWord document
- IMG_2041.HEICiPhone photo
- vendor-contract.pdfPDF · read page by page
Compare the contract against our launch brief and list anything that clashes.
Research, files and charts
From a question to a report, a chart or a file
Switch on research for a cited report built from the web and your own work. Ask for a spreadsheet, a PDF or a Word document and it appears under the reply, with charts in the chat and inside the files, and totals calculated exactly from the data you attach.
- Deep research with citations, exported to PDF or Word or saved to Docs
- Excel, PDF, Word and more as files you can open and revise
- Interactive charts with a data table and PNG download
Summarise the attached Q4 numbers as a workbook, a one-page PDF and meeting notes.
Done. Totals are calculated from your data file, not estimated.
- q4-summary.xlsxExcel · 2 sheetsSave…
- q4-summary.pdfPDF · 1 pageSave…
- q4-notes.docxWord · 2 pagesSave…
Projects
Projects your team shares, next to your work
Group chats into a project with its own instructions, knowledge files and memory of what has been decided. Share it with your team, and link a Docs space or a Tracker project so answers draw on live work, each read with the viewer's own access.
- Instructions, knowledge files and automatic project memory
- View or edit members, and chats you can share
- A shared project drive that follows project membership
Instructions
Write in a warm, plain tone. Always name the owner of an action.
Knowledge
- brand-voice.pdf
- launch-plan.docx
Project memory
- Launch moved to the second week of March.
- Pricing page copy is owned by Lucía.
Live context
- Docs · Marketing
- Tracker · Website
- Amara· Edit
- Arjun· Edit
- Lucía· View
Keys and budgets you control
Your providers, your keys, your spending limits
Admins add your AI providers' API keys once; they are encrypted at rest and used only on the server, never in the browser. Admins can set a budget for each member and limit which models a person can use, and members can see what they have used.
- Provider keys entered by admins, encrypted at rest
- Per-member budgets with an alert before the limit
- Model allow-lists, and usage charts for admins
Your allowance
Eligible
$20.00
this month
Used
$7.40
37% of budget
Remaining
$12.60
resets 1 Oct
Budget meter
Related
Works with the rest of the office
Do I have to pick a model?
Can it search the web?
Whose API keys does it use?
Is my data used to train models?
Which files can it create?
Can it change things in my workspace?
How is it different from a chatbot in a browser tab?
Ask, and get on with the work
The AI workspace lives in the same office as your team, with keys your admins control.