build your own magic wand

The useful bit is how this fits your life. Keep what helps, change what does not, and tell it when it has misunderstood you.

You might want a quieter listener, a cheekier thinking partner, or help getting one practical thing done. You do not need to explain your entire life first.

Help me shape this support around my life. Start with what I have brought today and ask one useful question at a time. Reflect what you think you understand, make your assumptions visible, and let my corrections win. Ask what kind of support and tone would help. Keep work and money in view alongside care, relationships, health, creativity, play and rest. Do not turn everything I mention into a task. Help me find one useful thing we can try with the information I choose to share. There is no need to build a system or produce a set of documents first.

If you want something to carry into another conversation, ask for a few lines about what matters to you, what helps and what to avoid. Review them before saving or sharing. Memory works differently in different AI tools: check your tool’s settings rather than relying on a promise in a chat.

A good conversation may be enough. If you keep finding the same friction, you can gradually give your AI a more useful job. Let the need lead.

Begin with the life

Ask what a better ordinary week would feel like. Notice what needs relief now, what deserves steady attention over the next few months, and what you hope to grow into over the next few years. A distant possibility does not have to become today’s task.

Choose one piece of help

Try something small enough to understand and useful enough to feel: preparing one meeting, making sense of notes you supply, or untangling one household routine. Say what relief would look like and what must not happen. Let it prepare something you can inspect before giving it permission to act.

Keep the machinery small

Start with the tools you have. If you begin tracking commitments, choose one place for them. Keep wishes, reference material and things you are waiting for distinct from promises you have actually made. Save a short, corrected account of what matters rather than an archive of everybody’s lives.

Be clear about permission

Decide what it may do by itself and what needs your say-so. Begin with conversation and drafts. If a useful job needs access to a tool, give only the access that job needs. Check sending, spending, publishing, deletion and commitments carefully before allowing them.

Writing a rule in a chat does not make an app obey it. Check what the app is allowed to do in its settings. If you cannot safely limit an action, keep doing that part yourself.

Protect what is personal

Share only what the current job needs. Leave out other people’s private information unless you have a proper reason and their permission where needed. Check the AI provider’s data and memory settings. Reading this guide and choosing to use a chat service are different things; that service’s privacy terms apply to what you tell it.

Check what actually happened

A draft is not a sent message. An attempted action is not proof that it worked. Ask the AI to say what changed, what it checked and what remains unknown. For anything consequential, look at the result in the place it was meant to happen.

Notice whether life gets better

After trying it, ask whether it brought clarity, relief or room to be more yourself. Notice the effort you spent supervising it as well as the help you received. Keep, adjust, simplify or remove it accordingly. A system creating more work than it saves has not earned a bigger job.

If one part fails, pause that part and keep the useful parts. You do not need to start your whole life again.

A small brief for your next experiment

Help me improve one recurring frustration: [describe it]. What would feel better is [the difference I want]. What must not happen is [my boundary]. Use only the information I choose to give you. First prepare something I can inspect. Do not connect accounts or act outside this conversation without the permission we have agreed. If a capability is unavailable or a boundary cannot be enforced, say so and suggest a manual way forward. Keep the experiment small, ask only questions that change the next useful step, and help me judge the result by its effect on my life.

The point is more room to live. Your magic wand should earn its place, one useful piece of help at a time.

I call mine Lila. She is an experiment in support that knows me, holds the thread and helps things happen. Here is what I’m learning, including the bits that still get stuck.

Snapshot: 15 September 2026. This is an evolving example, not a finished product. Take what helps and leave the rest.

I call mine Lila. I want to be able to say something once, go for a walk, and come back to something useful having happened. Some days we get closer. Other days I find myself managing the thing that was supposed to spare me the managing.

You do not need to copy my setup. Start with the AI you already use and one small thing you would like help with. Here is what I would borrow from the way I am building mine.

Help me make one ordinary part of my week easier. I will tell you what keeps getting in the way.

Use Marc's example below as something to learn from, not a setup to copy. Ask one useful question at a time. Use only what I choose to share, and keep my relationships, rest and room to be myself in view.

Help me choose one small thing we can try with the tools I already have. Show me what you can prepare now, what I would need to do, and what you cannot do. Do not assume you can remember things or work after this conversation ends.

Agree with me what you may do by yourself and when you should ask first. Explain any app settings I need to check in plain words. A promise in a chat is not a permission setting.

After we try it, help me check what actually happened and whether it made life easier. Keep what helps and simplify what does not.