Pulse AI's behavior is shaped by system prompts — the instructions that tell the AI how to write briefings, deep-dives, chats, advisory drafts, and assist text. You can view and edit these prompts from the admin console. This guide covers the AI prompt settings.
💡 Tip — Who can use this For platform admins, under Settings (AI Configuration).
Quick reference
| Step | Action |
|---|---|
| 1 | Edit a prompt and save (or reset to default) |
Step 1 — Edit an AI prompt
Go to Settings (the AI Configuration page). You'll see tabs for each AI surface — the executive briefing, pillar deep-dive, copilot chat, AI assist, and the auto-advisor. Each tab shows a description of what that prompt controls and a textarea with the current prompt text.
To change one:
- Edit the prompt text in the textarea.
- Click Save Prompt. The change takes effect immediately and invalidates any cached AI output for that surface, so new generations use the new prompt.
Use Reset to default to restore the textarea to the built-in default (this doesn't auto-save — you still need to click Save to apply it).
📝 Note — Saved prompts override the defaults The built-in defaults live in the code, but any prompt you save is stored in the database and takes precedence. Resetting to default and saving restores the code's version.
⚠️ Warning — Changes affect all AI output Editing a system prompt changes how Pulse AI writes for every client on that surface. Test thoughtfully — a vague or conflicting prompt can degrade output quality platform-wide. It's wise to keep a copy of the previous prompt before experimenting.