AI can remember more now.

A cozy corner of the internet where AI makes sense — no degree required.

This week: AI assistants are getting better at remembering your preferences, projects, and habits. That can make them more helpful — if you review the memory once in a while.

For a long time, using AI felt like starting over every time.

You would explain who you are.
You would explain what you are working on.
You would explain your preferences.
Then, a few days later, you would do it again.

That is changing.

OpenAI has started rolling out a more capable memory system for ChatGPT, designed to keep context fresher and more useful over time. In plain English, ChatGPT is getting better at remembering useful things from past conversations, not just the exact facts you deliberately told it to save.

Google is moving in the same general direction with Gemini becoming more proactive and more connected to your everyday tools. Anthropic keeps improving Claude for longer work, collaboration, and agent-style tasks.

The bigger pattern is simple:

AI tools are trying to understand your life with more continuity.

That can be genuinely useful.

If an AI remembers that you prefer plain English, it can stop giving you jargon.
If it remembers your project goals, it can give better next steps.
If it remembers your constraints, it can stop suggesting things that do not fit your real life.

But memory also needs care.

Because useful context can become stale.

Maybe you changed jobs.
Maybe a project ended.
Maybe your budget changed.
Maybe you told the AI something during a stressful week that is no longer true.

A human assistant would need updates too.

So the new habit is not “let AI remember everything.”

The habit is:

Treat AI memory like a little notebook you review, correct, and clean up.

That keeps personalization helpful instead of weird.

If your AI tool has memory, custom instructions, personalization, or saved preferences, do a quick check this week.

Open the settings and look for anything like:

Then ask your AI:

You do not have to make it perfect.

Just clean up the obvious stuff.

Remove anything too private.
Correct anything outdated.
Add one preference that would make future answers easier to use.

For example:

Simple rule:

The big AI trend this week is that assistants are moving from one-off chatboxes toward ongoing helpers.

OpenAI is improving ChatGPT memory so conversations can carry forward more useful context. Google is pushing Gemini toward richer, more app-connected assistance. Microsoft says computer-using agents in Copilot Studio are now generally available for business workflows. Anthropic’s Claude Opus 4.8 update focuses on better collaboration, stronger agentic work, and more control over how much effort Claude spends on a job.

Different companies. Different products. Same direction.

AI is not just answering questions anymore. It is starting to remember, plan, use tools, work across apps, and stay involved for longer stretches of time.

For regular people, the takeaway is simple:

The more AI can do for you, the more important it becomes to know what it knows, what it can touch, and when you want a clean slate.

That does not mean you need to become technical.

It means the basic habits matter more:

The best AI tools will not just be smart.

They will make control feel obvious.

This week’s pick is Temporary Chat.

If memory is one side of useful AI, Temporary Chat is the other side.

It is for the moments when you want help, but you do not want that conversation to shape what the AI remembers about you later.

Use it for things like:

Think of it like opening a fresh notepad.

You can still get useful help, but you are not adding that conversation to the ongoing story.

Different AI tools use different names for this idea. ChatGPT has Temporary Chat. Other tools may have private chats, incognito modes, history-off settings, or project-specific spaces.

The button matters less than the habit:

That little choice gives you more control without making AI harder to use.

Personal AI gets better when it knows a little about you.

But “a little” should still be chosen on purpose.

You do not need to share everything to get better help. Start with preferences, goals, working style, and recurring tasks. Keep sensitive details out unless you understand the controls and truly need the help.

AI memory should feel like a helpful notebook, not a mystery file.

And when something does not belong in the notebook, use a fresh page.

Hit reply: what is one thing you would actually want your AI helper to remember about how you like to work? 🧡

The Cwtch is published weekly. Unsubscribe anytime — no hard feelings.

This newsletter may contain affiliate links. They cost you nothing.