Chat page
A Fluid page that users open from a tile on their homepage or the NavBar, for open-ended questions, help finding pages, and guided tasks.
PeopleSoft Assistant is a conversational chat UI built into PeopleSoft Fluid. Instead of hunting through navigation and search pages, users ask in plain language. The Assistant answers with live PeopleSoft data, takes them straight to the right page, or walks them through the task.
Every request runs with the signed-in user’s own PeopleSoft security, so the Assistant can never show more than that user could see on their own.
Chat page
A Fluid page that users open from a tile on their homepage or the NavBar, for open-ended questions, help finding pages, and guided tasks.
Magic Button
An Ask the Assistant action on the fields you choose. It opens the Assistant already knowing the page, the field, and its value, so users can ask “What does this mean?” or “What should I enter here?”
Magic Buttons are field-level related actions, set up under PeopleTools > Portal > Related Content Service, so delivered page definitions stay unchanged. Magic Buttons shows how.
The user asks on the chat page or through a Magic Button. The question goes to the Assistant service, with the user’s access token and any page and field context.
The LLM plans the answer. It works out what information it needs, and which of the Assistant’s tools can provide it.
PeopleSoft APIs do the work. Each tool calls a REST API built with the Application Services Framework (ASF), with a token for the signed-in user. PeopleSoft checks the user’s permission lists, and the API applies row-level security.
The user gets an answer grounded in live data, with links to the relevant PeopleSoft pages. When they want to change something, the Assistant links to the page and says what to enter; the user makes the change there.
Set up the Assistant takes you through each step.
Oracle also offers chatbots for PeopleSoft, built on Oracle Digital Assistant with the PeopleSoft Chatbot Integration Framework. The Assistant described here takes a different path: it works with any LLM that supports tool calling, and its tools are your own application services.