I stopped writing complex AI prompts — this 60-second 'memory' trick works 10x better

There’s a lot of talk lately about switching chatbots. Companies like Google and Anthropic have made it easy to switch from ChatGPT to Gemini or Claude with a simple prompt, thanks to each model’s advanced memory capabilities.

Despite this easy, 60-second switch, the move is not exactly seamless. Once you start chatting, you’ll notice that what you told ChatGPT may still appear in Claude or Gemini, but the context is gone and the nuance has disappeared. It’s like chatting with a stranger and referencing an experience you had with a good friend. The stranger kind of gets it, but the vibe is off. It’s a “you had to be there” moment.

Essentially, the way you manage memory is just as important as the types of prompts you use. And while this may sound like one more complicated piece of the AI puzzle, it’s actually simpler than it may sound.

Once you understand how to invest some time into improving a chatbot’s memory, you’ll realize you’ve been using AI the hard way this entire time.

What a 'memory manager' actually is

To be clear, a memory manager isn’t a feature you toggle on. It’s a shift in how you use AI. Instead of starting from scratch every time, you:

• Store key context (who you are, what you care about, how you work)
• Reuse it across conversations
• Refine it as you go

You are essentially treating ChatGPT or another AI chatbot like a system that actually understands you. This is a key shift that makes everything faster and boosts productivity exponentially.

The problem with prompt engineering

As a prompt engineer, I can honestly say there's a problem that nobody talks about. Although I've seen success from numerous prompts I've created, without memory, even my best prompts wouldn't perform as well.

If you're using prompts like the 'potato' prompt, 'reverse brief' prompt or 'glitch prompt' and not enabling memory, you're not getting the full benefit of AI. You're just giving yourself busywork rather than boosting your productivity — and this is a trap a lot of users fall into.

To avoid this issue, try something I swear by: the reusable 'memory layer.'

Here’s what that looks like for me:

• Focus on facts
• Avoid jargon
• Cite sources and provide real examples
• Keep things clear, practical and easy to understand
• Avoid being overly chatty or pleasing
• When responding prioritize clarity over cleverness
• Do not add a follow up question unless prompted to

This is just a core memory block, but you can add in anything else that matters to you to help you stay focused and productive. Now, instead of starting from zero, I drop this in — or build on it. And the difference is immediate.

Building memory over time

This is the real upgrade and where you'll notice how the AI goes from good to great. A memory manager isn’t just one block, it’s something you evolve and continue to add to overtime. You can even tell the AI to "remember this" or give it feedback such as "I like that response, keep doing that/responding like that."

Overtime, the AI will become more naturally immersed in your work so you don't have to continue to tell it what to do and how to act, it will organically pivot with you. This is true for ChatGPT, Gemini or Claude. The AI will start feeling like something that actually understand how you work.

To try a memory manager for yourself, you don't need a new tool, special app or new subscription. Simply write your core context, like I shared above. That should include who you are and why you're using AI, what you want from AI and how you prefer it to respond.

Be sure to save this memory manager in a notes app, pinned doc or reuse it directly in ChatGPT with Projects (this also works for Claude Skills and Gemini Gems). Then continue to build it — keep what works and remove what doesn't or is simply slowing you down.

The takeaway

The more you use AI, you'll discover that prompt engineers no longer have the edge. It's those who have learned to build context and reuse it intelligently. By leaning into what a memory manager does and exploring how to implement it into your workflow, you'll notice yourself working smarter with less effort.

Give it a try and let me know in the comments what you think. Once you start using a memory manager, it’s very hard to go back (and you won't want to).

Follow Tom's Guide on Google News and add us as a preferred source to get our up-to-date news, analysis, and reviews in your feeds.

More from Tom's Guide

• I asked ChatGPT to help me fix 7 everyday problems — I didn’t expect this
• I used ChatGPT wrong for months — these 3 simple 'system' prompts changed everything
• I gave my parents 7 simple prompts — now they use AI every day