How to build your JARVIS

Before you write a single word, decide what JARVIS runs. There are two lanes, and the whole build is cleaner if you commit to one first:

• Business brain — it thinks about your company: sales, leads, cash, customers, tasks. Best if you run or work in a business and want a chief of staff.
• Life brain — it thinks about you: money, health, projects, family, goals. Best if you want a personal operating system that keeps your life on track.

You can build both later. For now, pick one. Paste the little self-interview below into your AI and answer out loud in the chat — it drags the raw material out of your head so Step 2 is just tidying up.

This is the heart of JARVIS. Open a blank doc, title it CONTEXT.md, and fill in the template below with the summary you just generated. Be specific — real names, real numbers. The AI is only as smart as what it can see, so give it eyes.

Don't chase perfect. A rough, honest CONTEXT beats a polished, vague one. You'll add to it every week.

Now the personality and rules. Start a fresh chat, paste the prompt below first, then paste your entire CONTEXT.md right after it. From that moment the AI stops being a generic bot and starts answering as your operator.

On Claude and ChatGPT you can make this permanent: drop the prompt into Custom Instructions / Project instructions / a Claude Project or a Custom GPT so you never paste it again — just open the chat and go.

One brain that knows everything is powerful. But your best thinking happens when JARVIS goes deep on one area — a "sub-brain" that only wears one hat. In a business that's a Sales brain, an Ops brain, a Finance brain. In a life that's a Money brain, a Health brain, a Projects brain.

A sub-brain is just the system prompt above plus a role focus and the slice of data for that area. Here's a worked example you can adapt — swap the role and it becomes any area you want.

This is the highest-ROI habit of the whole system. Once a day you hand JARVIS what changed and it hands you a sharp briefing back. Do this for five mornings and you'll never run blind again.

The prompt is short because your system prompt already defined what "briefing" means. You just feed it the fresh stuff:

How to make it actually happen every day — pick the lightest option you'll stick to:

• Phone shortcut (easiest): save the prompt as a text-replacement snippet or a home-screen note. Every morning: open JARVIS, paste, fill the blanks. 60 seconds.
• Calendar nudge: a recurring 7:00am reminder titled "JARVIS briefing" with the prompt in the notes field. The reminder is the ritual.
• Scheduled task: in ChatGPT use Tasks (or a Claude scheduled routine) to auto-run a briefing at a set time so it's waiting for you. Paste yesterday's data the night before and wake up to the answer.

You now have a working JARVIS. Two upgrades make it feel like the movie:

• Talk to it. Both the ChatGPT and Claude apps have voice mode. Open your JARVIS chat, tap the mic, and just ask — "what's my day look like, what am I forgetting?" Hearing your briefing out loud while you make coffee is the moment it clicks.
• Stop pasting — go live. The manual version has you paste data in. The next level connects it so each brain updates itself. The cheat below turns any messy export into clean context in one shot.