How to Stop Re-Explaining Yourself to AI

AI is smart, but it doesn't know you, so every new chat starts like a call to a support line. Write down what you keep repeating, once, and put it where your AI reads it at the start of every chat. If you use more than one AI tool, connect them all to one shared memory so there's only one copy to keep.

Why you keep repeating yourself

Think about calling a support line. The person who picks up is smart and polite, your account is on their screen, and you still explain who you are, what's wrong and what you've already tried. Get cut off, and you explain it all again to someone new. Now think of the person you'd call first for advice. You don't explain yourself to them. They already know your circumstances.

Today's AI is the support line. A new chat starts with almost nothing. It doesn't carry your last conversation into this one unless a memory feature brings it along, so it fills the gaps with the most average version of you there is.

ChatGPT and Claude both have built-in memory now, and it helps. But each one keeps its memory inside its own app, and neither can see the other's. They also decide much of what gets saved, so the detail you need today may not be there, and an old detail may still be.

The fix is to stop relying on the AI to notice things. Decide what it should always know, and put it in a place it always reads. The five steps below need no extra tool, and most people can do them in one sitting.

Step 1: Write a one-page brief

Open a blank note and answer these in plain sentences:

Who you are and what you do.
What you're working on right now.
How you like answers: length, tone, format.
What the AI should never do, and what it should always check.
The people, products and terms that come up most.

Keep it short. ChatGPT limits custom instructions to 1,500 characters on Free and Go plans and 5,000 on Plus and higher plans, so one tight page is the right size.

A quick way to start: look back at your last five chats and copy out every sentence where you explained something about yourself. That's your first draft.

Step 2: Put the brief where the AI reads it every time

Both apps have a setting for standing instructions that apply to every chat.

In ChatGPT, on the web or desktop:

Open Settings and select Personalization.
Check that "Enable customization" is on.
Paste your brief into the Custom Instructions field.

OpenAI says custom instructions are available on all plans.

In Claude:

Click your initials in the lower left corner.
Select Settings.
Paste your brief under "Instructions for Claude."

Anthropic says anything you add there is applied to all of your conversations with Claude.

Step 3: Give each ongoing job its own project

Your brief covers you. It shouldn't also hold the details of every client, class or codebase. Put those in a project.

In ChatGPT, select New project in the sidebar and name it. Add instructions in the project's settings, and upload the files the AI should refer to. OpenAI says project instructions apply only inside that project and override your custom instructions there.

In Claude, open Projects in the left sidebar and click "+ New Project." Click "Set project instructions," then add files to the project's knowledge. Anthropic says projects are available on all plans, and free accounts can create up to five.

Now a new chat inside the project starts with the background already loaded.

Step 4: Check what built-in memory has saved

Built-in memory works in the background, so look at it once in a while.

In ChatGPT, go to Settings, then Personalization, then Memory. OpenAI describes two parts: saved memories, and a setting that lets ChatGPT draw on your past chats.
In Claude, go to Settings, then Memory. Anthropic says everything Claude remembers is listed under Topics, and you can open a topic to read, edit or delete it.

Delete what's wrong or out of date. An old fact is worse than no fact, because the AI will use it with confidence. Say you raised your rate in March: if memory still holds the old one, that's the number that lands in the next quote. If something important is missing, tell the AI to remember it.

Step 5: End long chats with a handoff note

When a chat gets long, or you're stopping for the day, ask:

"Summarize this conversation for a new chat. List the decisions we made, what is still open, and anything about my preferences you learned."

Paste the answer into your next chat. If it holds something lasting, add it to your brief or your project.

Where these steps run out

All five steps work inside one app, and only as long as you keep them up. Stop updating the brief and it goes stale, and the AI keeps trusting it.

If you use ChatGPT, Claude and a coding tool, you now have three copies of your brief, three sets of projects and three memories. You're the one keeping them in step.

The way around this is a memory that sits outside the apps. ChatGPT and Claude can both connect to outside tools through an open standard called the Model Context Protocol, or MCP. A memory service that speaks MCP gives every connected app the same notes to read and write. Tell one app something, and the others can look it up. Check your plan first, because both apps limit this feature by plan.

Before you pick one, ask three questions. Who holds the encryption keys? Can you see, fix and delete what's stored? Can you export everything and leave?

Where Penny fits

Penny is a memory service of this kind, built so the upkeep isn't your job. Your AI writes notes as it works with you, so there's no brief to keep current by hand. Penny keeps different kinds of things as what they are: notes, to-dos, the things you track, and how you like things done. When your AI needs a detail, Penny's search finds it and brings it into the chat.

It works in Claude, ChatGPT, Perplexity, Cursor, Codex, and Claude Code, so the same memory follows you from one tool to the next.

On the three questions: Penny holds the encryption keys, because your AI needs to read memory in plain text to use it. Every time Penny's service decrypts your memory, it's on record in your dashboard and sealed into a public, tamper-evident log. You can read, fix or delete anything Penny keeps, and export everything from your dashboard anytime.

Free to start: 14 days, no card. After that it's $59 a year for founding members, locked for life, or $99 a year on the standard plan. Connecting Claude or ChatGPT requires a paid plan with them.

A saved Penny profile supplies the business details and writing preferences for a new conversation. Demonstration with sample data.

Frequently asked questions

Why does my AI forget what I told it yesterday? Each new chat starts fresh. The AI only knows what's in the current chat, plus whatever your instructions, project or memory settings bring in.

Are custom instructions and memory the same thing? No. Instructions are text you write and control, and the AI reads them in every chat. Memory is what the app saves as you talk. Use instructions for anything that must always be there.

Will this work across ChatGPT and Claude? Steps 1 to 5 work in each app, but you have to set them up and update them in each one. To keep a single copy, connect both apps to one shared memory.