Just ask · For IT and data protection
Just ask: security and privacy
How ToolSense connects to Claude, ChatGPT and Microsoft Copilot, what it can access, and how you stay in control.
Coming soonWrite access: act on the answer right from your assistant, for example to create a ticket.
How a question flows
- 1A user asks a question in their assistant.
- 2The assistant calls the ToolSense connector with that user's sign-in token.
- 3ToolSense returns only the records needed for the answer, within the user's permissions.
- 4The assistant writes the answer and links the records.
Access and control
Can it change data?
Not today. It only reads and cannot create, change, close or delete any record. Write access is coming soon.
What can a user see?
Only what their ToolSense role and site access allow. There is no shared service account or API key.
How do users sign in?
Through OAuth with their own ToolSense account, so the AI assistant never sees the password. The assistant stores the sign-in token for that user only, or the Claude Desktop app stores it on managed installs.
How is access removed?
A user disconnects ToolSense in their assistant, and access ends immediately. Admins can remove ToolSense for the whole workspace.
Do we need firewall changes?
No. Claude, ChatGPT and Copilot connect to ToolSense from their own cloud. Users only need to reach the ToolSense sign-in page in their browser.
Rollout
Claude
Team and Enterprise owners add it once for everyone. For Claude Desktop, the setup page has a macOS profile (.mobileconfig) and Windows files (.reg, .ps1) for your device management tools. The files contain no credentials, so each user still signs in.
ChatGPT
Admins publish it to the workspace. Enterprise and Edu admins choose which groups get it.
Microsoft 365 Copilot
Your Microsoft 365 admin adds it once and can start with a pilot group through Staged rollout. ToolSense data is fetched live and isn't indexed or stored in Microsoft 365.
Updates
Claude picks them up automatically. On ChatGPT Business, a published app can't be updated, so admins recreate and republish it. Enterprise and Edu admins click Refresh under Action control.
Auditing
Microsoft 365 Copilot activity can be audited in Microsoft Purview. ChatGPT Enterprise and Edu include app use in OpenAI's Compliance API.
Privacy and data
What is shared?
Only the records the assistant fetches to answer a question. Anthropic (Claude), OpenAI (ChatGPT) or Microsoft (Copilot) processes them under your organization's agreement with that provider.
Is it used to train AI models?
| Plan | Used for model training? |
|---|---|
| Claude Team, Enterprise | No, by default |
| ChatGPT Business, Enterprise, Edu | No, by default |
| Microsoft 365 Copilot | No. Microsoft says prompts and responses aren't used to train its foundation models |
| Claude Free, Pro, Max | Only if the user turns on Model Improvement in Claude's privacy settings |
| ChatGPT Pro | Yes, by default. Switch it off under Settings > Data Controls > Improve the model for everyone |
We recommend company plans for company data. Source: Anthropic, OpenAI and Microsoft policies as of November 2026.
Can answers include personal data?
Yes, if the user's role allows it: staff names and emails, site contacts and vehicle trips. Private trips appear without distance, duration or addresses. If you review AI tools with your data protection officer or works council, include Just ask in that review.
How long are conversations kept?
As long as your assistant's retention settings say, like any other chat.
Try it safely
Want to test without company data? Create an account at toolsense.io/sign-up. It comes with a sample company and connects the same way.
Create an account with sample dataQuestions from your IT or data protection team?
Customers: contact your Customer Success Manager. Everyone else: book a call with us.