Give your AI a proper onboarding

At some point over the last couple of years you have likely heard the advice that prompting with context gives you better results from AI tools. For most people, prompts look something like "Pretend you are an expert in [X], I'm building [Y], I like responses that are [Z]." The AI gives you something decent and you refine the responses with better prompting. But then you often have to repeat the same context over and over after every conversation.

Although AI tools increasingly have persistent memory between chats, most people never set it up intentionally. It either stays empty, or it fills up with random things the AI decided to remember on its own.

This week, we're fixing that. Just like sitting down for a proper 1:1 with a new co-worker, giving your AI tool a formal brain dump will change what it can do for you, making it more useful permanently.

The Challenge:

You're going to teach AI who you are so it can show up like a colleague who already knows you, not a contractor you have to brief from scratch every time.

Most AI tools now have a way to store your context permanently. It might be called Memory, Custom Instructions, or Personalization depending on the tool you're using. In the prompt below, we'll ask it to help you figure out how to make this stick in your tool of choice.

The prompt to get you started

First, we're going to figure out what's actually worth telling AI about you. Drop this into any AI tool and let it interview you:

"I want to set up a personal context document so AI tools can give me more relevant, useful help without me re-explaining myself every time. Interview me to gather everything you'd need. Ask me one question at a time to cover what I do, what I'm currently working on, how I like to communicate, what I'm trying to get better at, any personal preferences about formatting, or sample documents that can help you better understand my work, and anything else you think would help. When you have enough, write it up as a short, clean summary I can use to configure my tool's memory or custom instructions. First we'll do the interview – then give me step by step instructions on how to add this to your memory so you can reference it across all future conversations."

Let it ask you questions. Answer honestly. At the end you'll have a summary of your own context. 

Note: if typing the responses feels like too much, this is a good moment to try Voice mode. You can experience having a conversation with your AI using your own voice.

Once you've set it up

Open a brand new chat. Don't explain yourself or add any context. Just ask for help with something from your day to day and see how the output compares to previous conversations.

Too easy?

If your memory is already configured, Build a Project (Claude) or a custom GPT (ChatGPT) for a specific recurring task in your life. If you're on another tool, search for "Projects" or "Notebooks". Most platforms have their own version of localized memory.

The difference between general memory and a Project is focus. This generalist new hire vs. a specialist. General memory makes every conversation slightly more personal. A Project gives one specific workflow its own dedicated context, instructions, and files.

Drop this prompt into your tool of choice to get it set up:

"I want to create a dedicated space for support with [recurring task or project]. Based on what you already know about me, help me configure it. What context should live here specifically? What instructions would make your responses in this space more useful than your general defaults? What files or information should I upload to give you better grounding?"

Not sure what to build it around? Try something you do on a repeat basis like planning your weekly team meeting, meal prepping for the week, managing a job search, tracking a side project, prepping for a recurring client. The more you'd normally have to re-explain the setup, the better the candidate.

The goal is a version of your AI tool that feels like it was built for one specific job and you can call on it regularly. 

Why this matters?

Your AI tool is already building a picture of you. But without intention, you don't control what it remembers or how it uses that context. The gap between a generic AI response and a genuinely useful one is almost always context. 

The tools are capable, and the more they know about you, the more they can help. The people getting the most out of AI aren't necessarily using better prompts; they've done the work upfront to make the tool understand who they're talking to and what they're trying to accomplish. This set up can change the quality of the work you are getting and how much time you have to spend giving back and forth instructions. 

Think of this challenge as creating a 1:1 onboarding for a new employee.

That’s it for this week.

Once you're done, reply and tell me: did the AI ask you something you weren't expecting? What was the biggest change you noticed in your output? I read every response, and your experiments help shape these challenges.