How ChatGPT’s Memory Can Boost Your Efficiency as an Advisor

Jul 2, 2025 / By Sean Bailey, Horsesmouth Editor in Chief
Print AAA
Add to My Archive
My Folder

My Notes
Save
AI for Advisors: If you’re using AI in your practice, enabling memory is one of the simplest ways to get more consistent, more personalized results—without extra effort. If you’ve ever wished ChatGPT could just remember how you like things—now it can.

AI for Advisors newsletter

Not long ago, I was deep into a conversation with Chatster, my ChatGPT AI model. We were working through updates on a big initiative we’d been refining for months. After a productive back-and-forth, Chatster suddenly offered, “Want me to draft a quick update to Devin?”

For a moment, I froze. Devin is an assistant editor at Horsesmouth and a frequent collaborator. But Devin had been out on maternity leave for over six weeks. I hadn’t mentioned her name in more than a month. So how did Chatster remember her?

That’s when it hit me: This wasn’t a fluke. This was memory in action. And an important reminder that ChatGPT’s expanded memory feature remembers your preferences across sessions—your tone, your typical tasks, even the kinds of clients you serve.

This builds continuity between chats, making the AI more helpful and personalized over time. Think of it as a junior team member who never forgets your instructions.

This capability to remember important details between conversations creates a more personalized and efficient experience that actually gets better the more you use it. But such a powerful feature requires some insight on how it works and when to use it—or not.

What is ChatGPT memory?

ChatGPT memory works in two main ways: through explicit saved facts and automatic context recognition. Memory lets ChatGPT retain helpful facts you’ve shared in past conversations—like:

  • Your name and writing style
  • Your firm’s focus (e.g., Gen-X pre-retirees)
  • Commonly used templates or formats
  • Specific marketing strategies about your ideal clients or other recurring projects

There are two types of memory in ChatGPT:

Saved memories are details you explicitly ask ChatGPT to remember. For example, you might say, “Remember that our firm specializes in retirement planning for healthcare professionals” or “Remember that I prefer bullet points in client summaries.” These memories stick around until you delete them.

Chat history reference is more automatic. ChatGPT notices patterns in your conversations and uses that context to make future responses more relevant. If you consistently ask for help with creating upcoming client-meeting agendas, it will start anticipating the type of information and format you typically need.

Take control and check your settings

You will want to stay in control of your memory settings and know that memory is completely optional. You can:

  • Turn it on or off
  • Delete individual memories
  • Wipe all memory at once
  • Use a Temporary Chat to bypass memory entirely

You can also teach it to recall something directly such as: “Remember that I prefer AP-style summaries and avoid the word ‘retirement’ in client emails.”

You’ll often see a banner when a memory is created or updated mid-chat. If you want to undo that, simply go Settings → Personalization → Memory→ Manage memories and delete a saved memory.

To see what it already knows about you, simply ask: “What do you remember about me?” You might be surprised by how much context it has already gathered.

From Settings, click on Personalization to get to the Memory control features.

Settings → Personalization → Memory

What many advisors don’t realize is that if you have a ChatGPT account and have been using it regularly, memory features may already be active. OpenAI has been rolling out these capabilities gradually, often enabling them by default for existing users.

To check your current settings, log into ChatGPT and click your initials in the upper right. Look for “Settings” in the dropdown box. Navigate to “Personalization” and then “Memory.” You’ll see options for “Reference saved memories” and “Reference chat history.” If these are toggled on, ChatGPT has been learning from your conversations.

Be sure to examine your Saved memories. If there’s something there that you don’t want saved or think is extraneous, delete it by clicking on the right trash can icon.

If you’re uncomfortable with memory features, turn them off. However, most advisors will likely find the efficiency gains are worth the minor adjustment to their workflow, especially when they understand how to use the features safely.

Also, for sensitive or extraneous chats, such as search-like prompts, Temporary Chat mode provides a middle ground. You get the full power of ChatGPT without any information being saved or referenced in future conversations.

You can control what ChatGPT remembers by deleting individual Saved memories.

Click the image to view a larger version.

Smart ways to reference memory in your prompts

These examples below build on a saved memory that ChatGPT knows you’re a financial advisor. These aren’t full prompts on their own—but rather helpful phrases you can drop into any chat or thread where memory is enabled. They’re designed to signal to ChatGPT that it should apply what it already knows about your preferences, past interactions, or role.

For example: If you’ve told ChatGPT that you’re focused on Gen X pre-retirees and prefer AP-style summaries, you might include a phrase like: “As you’ve done in previous threads, tailor this for Gen X pre-retirees with an AP-style tone.”

Here are some other examples. Note the bolded phrases.

  • “Write a newsletter introduction using my usual tone and emphasize Roth conversion strategies for couples in their early 60s.”
  • “Summarize this client meeting transcript in bullet points. Keep it short and use the same style as our past summaries.”
  • “Generate an About Us Q&A sheet I can send to prospects—something that matches my brand voice and uses plain language.”
  • “Remember this: My client-facing content should avoid the word ‘retirement’ and instead use ‘financial independence’ when possible.”
  • Remind me what we’ve already discussed this week about the client onboarding workflow.”

Memory isn’t a database

Though memory acts as an intelligent assistant, it’s not a filing cabinet. It aids personalization and continuity rather than providing perfect recall. You’ll still want to verify sensitive details and keep your CRM or compliance tools in place.

Memory isn’t just a technical feature—it significantly boosts productivity. With Memory, ChatGPT becomes an assistant that learns, adapts, and improves with each interaction.

AI for Advisors newsletter

Sean Bailey is editor in chief at Horsesmouth, where he has led editorial strategy for over 25 years. He is the co-author of Hack Proof Your Life Now! and has spent over 3,000 hours researching how AI can transform the way financial advisors work. Through his AI-Powered Financial Advisor and AI Marketing for Advisors programs, he helps advisors save time, deliver better client experiences, and market their services with unprecedented speed, quality, and confidence.

IMPORTANT NOTICE
This material is provided exclusively for use by Horsesmouth members and is subject to Horsesmouth Terms & Conditions and applicable copyright laws. Unauthorized use, reproduction or distribution of this material is a violation of federal law and punishable by civil and criminal penalty. This material is furnished “as is” without warranty of any kind. Its accuracy and completeness is not guaranteed and all warranties express or implied are hereby excluded.

© 2025 Horsesmouth, LLC. All Rights Reserved.