Shared Agents Share Access

What Is Shared Agents Share Access?

An AI agent you build on your own files and share with your team answers with your access, not theirs. Left on the default of all my files, it can hand any document you can open to anyone on the audience list, one question at a time. In this exercise you build a helper for your team, and two days later a colleague reads a confidential restructure document through it. You practise pausing the agent, reading its usage log, narrowing it to one folder, and telling the people affected.

What You'll Learn in Shared Agents Share Access

• Recognise that a shared no-code AI agent reads its files with the builder's permissions, so its audience inherits the builder's access to everything in scope
• Choose an agent's knowledge scope deliberately, selecting the folder it needs instead of accepting an all-my-files default
• Contain an exposure by pausing the agent before investigating, then use its usage log to establish who asked what and which files answered
• Apply scope as the fix rather than deleting individual documents, since every file in reach was equally exposed
• Disclose the exposure to the document's owner and make sure the people affected hear it from the right person

Shared Agents Share Access — Training Steps

1. The Same Questions Every Week

   It is Tuesday, 10:15, and you have just answered the renewal-process question for the third time this month. Orlenby Assist rolled out Agent Studio last week: anyone can build an assistant on their own files and share it with the people they choose. Your drive holds the team's playbooks. It also holds a People folder with your 1:1 notes, and a document Hanne Vestergaard shared with you in confidence last week: the Q4 restructure plan, with three roles at risk on your own team.

2. Open Agent Studio

   Agent Studio lives in the browser. Colleagues in IT, Sales and People have already published assistants there.

3. Start A New Agent

   The list shows what your colleagues have built and who each one is shared with. Yours will sit beside them.

4. Name It

   Give the agent a name the team will recognise and tell it what it is for.

5. What It Can Read

   The knowledge step asks which files the agent may read. All my files is preselected, and the playbooks are in there, so it will work.

6. Who Can Use It

   The whole point is that the team can ask it instead of asking you. Share it with Customer Success and publish.

7. Try It

   The agent is live. Test it with the question you answer most often.

8. It Works

   The agent reads the playbook and answers. Watch it write.

9. A Good Answer

   This is exactly what you built it for: the right steps, in order, with the playbook cited underneath.

10. Sarah Wants To Show You Something

    Thursday morning. Sarah Chen, a specialist on your team, messages you. She sits two desks away.