In partnership with

When AI Stops Acting Like a Goldfish

One of the quiet revolutions happening in AI right now has nothing to do with model size or speed.

It’s memory.

The moment these tools stop starting from zero and begin remembering how you work, everything changes: your pace, your output quality, your cognitive load. But every platform is handling memory differently and most people have no idea what’s actually turned on.

ChatGPT: Opt-in and the most refined so far

ChatGPT’s memory stays off until you enable it.
Even once it’s on, it asks before saving anything. That means:

• it won’t guess what’s “important”

• it won’t silently store details

• you’re always in control

When memory is enabled, ChatGPT gets very good at:

• tone and writing style

• audience targeting

• ongoing projects

• formatting preferences

• your business context

It’s structured, stable, and surprisingly useful for longer-term work.

To access the options for memory in ChatGPT, go to Settings → Personalization.

Black Friday Deal: Genspark

In a sea of single-model chatbots and half-baked “agents” that still leave you copy-pasting between tabs, Genspark quietly stands out as the first true AI operating system for getting real work done: it orchestrates 30+ frontier models (GPT-5, Claude Opus 4.1, Gemini 2.5, etc.) behind a single prompt, autonomously browses the live web, extracts and structures data into editable AI Sheets, spins up presentations, websites, or even phone calls, and delivers polished, ready-to-ship outputs—all inside one ad-free workspace for $25/month (or try it free with daily credits).

Claude: Yes, it has memory now — and it’s more powerful than people think

Claude now has two separate memory systems that you can toggle on.

1. Search and reference chats

Claude can reach back into past conversations and pull relevant details forward.
This is more like retrieval than memory but it’s hugely helpful.

2. Generate memory from chat history

This is the big one.
Claude can now extract lasting context from your chats and store it automatically.

That means Claude can remember:

• your background

• your projects

• your preferences

• your workflows

…and use them across chats and Projects.

You can also delete specific memories, which gives you real control.

Claude doesn’t advertise this loudly, but it’s a major upgrade and it puts Claude much closer to ChatGPT in terms of long-term personalization.

Grok: Memory just arrived, and it’s in beta

Grok recently added a toggle called:
“Personalize Grok with your conversation history.”

Turn this on and Grok will:

• look at your past chats

• extract recurring details

• personalize responses going forward

Caveats:

• Not available in EU/UK without VPN

• Still beta—occasional drift on very long threads.

Gemini: Lagging behind

Gemini is the only major frontier model that still resets between chats.
It’s great at context inside a session, and it’s deeply integrated into Google Workspace…
but it doesn’t build a long-term profile of who you are or how you work.

If Google flips this switch it’ll be a massive shift. But as of today, Gemini is the “smart assistant with no memory,” while ChatGPT, Claude, and now Grok all cross that line.

Learn AI in 5 minutes a day

This is the easiest way for a busy person wanting to learn AI in as little time as possible:

1. Sign up for The Rundown AI newsletter

2. They send you 5-minute email updates on the latest AI news and how to use it

3. You learn how to become 2x more productive by leveraging AI

Sign up to start learning.

Why this matters for your work

Once an AI model can hold onto your context, you stop wasting time repeating yourself. And that’s when the leverage shows up.

Memory gives you:

• speed — less setup

• consistency — tone and structure stay tight

• continuity — projects don’t reset

• better thinking — the model makes connections across time

We’re entering the era where AI isn’t just answering your questions —
it’s learning how you think.

I’d like to hear from you!

If you would like provide me with any feedback, suggestions, thoughts, discoveries, or just say hello, please write to me (I promise that I will read every single email): [email protected]

Till next time,

Hock