GenAI UX Principles

Always Signal When AI Is Involved

The user must always be aware when they’re about to interact with AI and when content on the screen was generated by AI. Use the GenAI buttons and GenAI embedded icons for every AI entry point. Keep the interface simple and uncluttered and use consistent, visible entry points to help the user:

• Understand what the AI capability can do for them.
• Decide whether they want to use it.
• Avoid triggering AI capabilities accidentally; AI actions run only on explicit user action and never on hover, focus, or page load.

Include a Disclaimer with Every Response

Every AI-generated response in an embedded experience, whether from Infor GenAI or third-party generative AI, must display this disclaimer: “Responses may be based on Infor data but are AI generated. Verify the information.”

• Place it next to the response, keep it visible without requiring interaction, and don’t allow the user to dismiss it permanently.
• When AI appears multiple times on a page, show the disclaimer only once—near the bottom or next to the relevant output.

To learn more, refer to the GenAI Transparency Note and Disclaimer at the Infor GenAI Wiki, maintained by the Infor GenAI team. (An Infor SAAM request is required for Wiki access.)

Explain What AI Output Is Based On

The user should be able to see where an AI response came from and how reliable it is so they can decide whether to use it.

• Where available, show the source(s) the response is based on so the user can verify it.
• So a weak answer isn’t mistaken for a solid one, make it clear if an answer is uncertain, partial, or incomplete.
• Reveal detail progressively. Show the essential information first and let the user expand to see the full sources, the data that was used, or the rationale for why AI gave this result. Don’t interrupt the task with explanations.

Keep the User in Control

The user must remain the decision-maker. Treat AI output as a draft or a suggestion until the user explicitly accepts it.

• The user’s judgment has the final say on whether AI output is used.
• AI output is never saved, sent, or applied to downstream work without explicit user confirmation, so errors can’t pass silently into stored data or outgoing messages.
• The user should be able to review, edit, or undo AI-generated content by default. If an action can’t offer one of these, document the exception and explain why in the UI. Provide a way for the user to give feedback about AI responses (helpful, not helpful, or report an issue).

Communicate Status

The user must always know what the system is doing at every stage of an AI workflow (generating, complete, partial, failed, or no result).

• Never leave the user at a blank screen or an unresponsive screen. Show a GenAI loading indicator while AI generates a response.
• For longer operations show progress and let the user cancel.
• Give every state a clear next step. If generation fails or returns no useful result, explain what happened in plain language and offer a retry or a manual alternative so the user can still complete their task.

Use a Plain, Honest Voice

The user must always know that they’re working with a tool—not a person. AI should sound neutral and factual and follow Infor’s voice and tone principles.

• Don’t give AI a name, an avatar, or any other human persona.
• Don’t use expressions of feeling or opinion, like “I think” or “I’m excited to help”.