EmpowerED AI Learning
Know Your Settings
A quick-reference guide to personalizing the AI tools you already use — Claude, ChatGPT, Copilot, and Gemini.
The future of work starts with you
Why this matters: every major AI assistant now lets you tell it who you are and how you want it to respond — but each one stores that information a little differently, and some of it is saved automatically unless you turn it off. Five minutes in your settings means you stop re-explaining yourself every conversation, and you stay in control of what the tool remembers.
Scope — this isn't per-conversation. Once you save a preference, instruction, or memory, it applies to every new chat on that account going forward, not just the conversation you were in when you set it up. On Copilot, it can follow you across Microsoft 365 apps too. That's what makes it useful — and it's exactly why what you put in there deserves a second thought.
Claude
AnthropicWhere to find it
- Click your initials, lower-left corner
- Select Settings
- Profile: "What preferences should Claude consider in responses?"
- Capabilities: memory and connected tools toggles
- Response Styles are set from the chat box itself
What to know
- Preferences are account-wide — they apply to every conversation, not just one chat.
- Styles (Formal, Concise, Explanatory, or a custom style built from your own writing) only change tone and formatting, not what Claude knows about you.
- Memory is a separate, opt-in toggle that lets Claude carry facts across sessions — you can review or turn it off.
- Paid plans add Projects, where instructions apply only within that project's chats.
ChatGPT
OpenAIWhere to find it
- Click your profile icon, bottom-left
- Select Settings → Personalization
- Open Custom Instructions
- Fill in "what should ChatGPT know about you" and "how should it respond"
- Manage saved Memory entries from the same tab
What to know
- Custom Instructions are two text fields with a roughly 1,500-character limit each; they apply to new conversations until you edit or disable them.
- Memory is separate — it builds automatically from things you mention in chats, unless you turn it off.
- You can browse, delete, or restore individual saved memories one at a time, not just wipe everything.
- A master Memory toggle sits beside "Reference chat history" — check both if you want full control.
Microsoft Copilot
Microsoft 365Where to find it
- In Copilot Chat, select the ··· menu, top right
- Select Settings → Personalization
- Open the Custom instructions tile to edit tone, format, or reading level
- Manage the separate Memory tile to view or clear what's been remembered
What to know
- Copilot is tied to your Microsoft 365 work or school account, so an organization's admin policy can limit or disable these settings.
- Custom instructions tell Copilot how to respond; Memory is what it remembers about you over time — the two toggle independently.
- You can turn Custom Instructions off entirely without affecting Memory, and vice versa.
- Because it runs inside Microsoft 365, this is one place professionals often already have organizational data protections in place — worth confirming with IT.
Gemini
GoogleWhere to find it
- Tap your profile picture
- Select Settings → Personal Intelligence
- Toggle Memory (learns from past chats) on or off
- Review Saved info for facts you've told it directly
- Set retention under Keep Activity (3, 18, 36 months, or never auto-delete)
What to know
- Personal context/Memory learns from past chats automatically and is on by default for eligible personal accounts.
- Saved info is different — explicit facts you've directly told Gemini to remember.
- Personal Intelligence can also connect Gmail, Photos, Search, and YouTube — this part is opt-in and off by default.
- Requires a personal Google Account (18+); not available on Workspace, school, or supervised accounts unless an admin enables it.
- Use Temporary Chat for a one-off conversation you don't want saved at all.
At a glance
| Platform | Settings location | "Tell it about you" | Automatic memory |
|---|---|---|---|
| Claude | Profile icon → Settings | Profile preferences (all chats) | Optional toggle under Capabilities |
| ChatGPT | Settings → Personalization | Custom Instructions (two fields) | Memory, on by default, editable |
| Copilot | ··· menu → Settings → Personalization | Custom instructions tile | Memory tile, admin policy may apply |
| Gemini | Profile → Settings → Personal Intelligence | Saved info | Personal context/Memory, on by default |
Three habits worth building
- Set it once, benefit every time. A few sentences about your role, your typical tasks, and how you like information formatted saves you from repeating yourself in every new conversation.
- Know what's automatic vs. what you control. Several of these tools now save details from your conversations by default — check the memory setting before you assume nothing is being remembered.
- Match the setting to the sensitivity of the work. For anything involving student data, personnel matters, or confidential business information, confirm your organization's AI policy and consider a private or temporary chat mode rather than relying on default settings.
What not to put in your settings
- Passwords, IDs, or account numbers. Settings fields aren't a password manager, and this information doesn't need to travel with every future conversation.
- Other people's private information. Student records, coworkers' personnel details, or a client's confidential business information belong in your organization's approved systems, not in a personal AI setting that's outside its control.
- Anything your organization's data policy restricts from AI tools. If your district or company has guidelines on what can go into an AI tool at all, those apply here too — settings don't create an exception.
- Health, financial, or other sensitive personal details you wouldn't want surfacing later. Because saved information can resurface in an unrelated conversation, it's worth asking, "would I be fine if this came up again in a completely different chat?"