How do I build an AI agent in Agentic OS?
Overview
Agentic OS, Syracuse's own AI platform formerly called mentorAI, lets students, faculty and staff build AI agents that answer from material they choose. Sign in at mentor.ai.syr.edu, select New Agent, add your documents and set how it responds. You decide who can use it.
Steps
Go to mentor.ai.syr.edu (opens in a new tab) and sign in with your Syracuse account.
Select New Agent from your dashboard. Give it a name, a short description of what it helps with and a category so people can find it.
Add your material: in your agent's Edit Agent screen, go to Integrations > Datasets and select Add Resource to upload documents or connect the sources it should answer from.
Shape how it responds: go to Configurations > Prompts and edit the System Prompt (the instructions that set its purpose and voice). Claude can help you write it: describe what the agent is for and ask Claude to draft the prompt.
Choose who can view and chat with it: go to Configurations > Access and select Create role access. Then share the link, or put it on a webpage from Integrations > Embed.
What you can add
Agentic OS reads text, so text-based sources work best. A scanned PDF or an image-heavy slide gives it little to work with.
- Files: PDFs, Word documents, PowerPoint decks, spreadsheets and plain text.
- Connected sources: Google Drive, OneDrive, a website, a GitHub repository or a video with captions.
Additional information
- Pick the model behind your agent (Claude, ChatGPT or Gemini), and switch any time.
- Optional: turn on Guided Prompt under Configurations > Prompts to suggest questions to people chatting with your agent.
- Optional: review Configurations > Safety. Safety settings can filter inappropriate messages and keep your agent away from medical, legal or financial advice, pointing people to the right campus resource instead.
- If web search is turned on under Integrations > Tools, your agent can also answer from the web, not only from your material.
- You can require a Syracuse sign-in to chat with your agent.
- The person or team who builds an agent can read people's chats with it, to improve it. See what happens to your data.
- The Responsible AI Guidelines (opens in a new tab) apply to whatever you build. Check them before you add student records or other confidential University data.
Official links
Still need help?
If this did not answer your question, submit a request and the ITS Data and AI team will follow up.
Contact ITS AI Support