Stanford Financial Management Services (FMS) continually assesses the financial ecosystem for opportunities to improve and simplify. This includes helping faculty and staff access the information they need to understand and take action on financial activities.
In December, FMS is piloting a new agent in Stanford’s AI Playground called the Financial Info Navigator (FIN), which aims to make financial information easier to find and apply appropriately. The pilot will provide the opportunity for users across the university to test the new tool and provide feedback, which will inform planning and readiness for its next phases.
What is FIN and what can it do?
FIN is a chat assistant designed to help Stanford faculty and staff explore financial questions and transactions. It is plugged into specific sources of information:
- Fingate (fingate.stanford.edu) is the website managed by FMS, including financial activity overviews, guidelines, and steps involved in successfully completing financial activities.
- Stanford's Administrative Guide (adminguide.stanford.edu), also referred to as the AGM, contains university-wide policies.
- Specific Oracle Financials modules: FIN retrieves real-time information from Expenditure Type Query, Requisition and Purchase Orders Query, Supplier Query and Request, and Financial Approver Authority Query.
Please note that to include certain financial details in FIN responses, such as purchase order or invoice status and transaction history, you must have current access to iProcurement systems. In instances where you don’t have access to the information, FIN will reply: "Permissions prevent the display of that information for you. FIN restricts access to certain information based on your Oracle authority. Refer to the FIN News page for more information.”
With all of this information pulled into one agent, FIN can produce summarized information that can guide users through policy and procedure questions such as:
- “What are meal reimbursement limits?”
- “How do I add a vendor or request a purchase order?”
- “What’s the status of PO 12345678?”
- “What’s the setup status of supplier Roche Diagnostics?”
The goal of the tool is to surface answers quickly and effectively, avoiding errors and the need to seek support.
Get the most out of FIN
Unlike other agents, FIN is specifically designed to answer questions in a standardized manner consistent with its specified information sources; it is not designed to generate content or complete a task. To get the most out of FIN, use well-crafted prompts (refer to UIT’s GenAI Prompt Guide) that will help the tool help you.
For example:
- Use a clear instruction or question with as much information as possible, for example: “What are travel meal options when the source of funding is a sponsored award and the traveler is an employee?”
- Ensure you are including the key term in your question, for example: “Which expenditure type should I use for office supplies?”
Finally, remember that FIN is not perfect; responses may be incomplete or change. Always verify critical details in the linked resources that FIN provides, or contact the Financial Support Center for assistance.
Try FIN and provide feedback
To get started with FIN:
- Watch the short video introduction.
- Navigate to the Stanford AI Playground and log in with Single Sign On (SSO) using your SUnet ID.
- From the Models/Agents drop-down list, select Financial Info Navigator.
- Start a conversation by typing in the message field (bottom center) or use the microphone to speak your request.
- Prompts are powerful. Give clear, specific prompts with context, goals, and constraints to get better FIN responses. Check the GenAI Prompt Guide for more tips and examples.
Provide feedback
Users of FIN can submit feedback in two ways to help FMS enhance the effectiveness of FIN:
- Thumbs up and Thumbs down rating
- Detailed feedback to help FMS enhance the effectiveness of FIN. Share feedback about your experience by submitting a Financial Info Navigator feedback or assistance ticket.
What happens next?
After the pilot period ends in early 2026, FMS will utilize feedback to inform the next phases of the project. Stay tuned!