Build Your Team
Start here. The complete, beginner-friendly path from an empty folder to a working AI team — including the one build prompt that creates it, explained in plain language.
This is the page where you build your own AI team — from nothing to a working team in about 15–20 minutes. No coding and no setup files to download. You'll create a folder, open an AI tool, and paste one prompt. This page has everything you need; you don't have to go anywhere else to get started.
What you'll need
Before you begin, make sure you have:
An agentic AI is one that can work inside files on your computer, not just chat. More in the Glossary.
Don't have an agentic AI tool yet? See Which AI tool should I use? in the FAQ.
Start here: get set up (2 quick steps)
These two steps create the home for your team and give your AI permission to work in it. Do them first, then you'll be ready to paste the build prompt.
Step 1 — Create your team's folder
- On your computer, create a new, empty folder.
- Give it your team's name, for example Personal Fitness Team, Applied Health Program Team, or Team Frank.
This one folder will be your team's entire home. The team can create and delete files inside this folder, but it cannot go outside it. You decide where it lives.
✅ You should now have: one empty, clearly named folder.
Step 2 — Open your AI tool and point it at the folder
- Open Claude Cowork or ChatGPT Codex.
- Create a new project.
- Give that project access to the folder you just created.
This is the step that gives your team permission to work. Without it, the AI can chat but can't build anything.
✅ You should now have: a new project that can see your team folder.
The build prompt (copy this)
Now for the heart of it. Copy the entire block below and paste it into your project's chat box. Two optional edits before you do:
- You can replace [Orchestrator] with any name you like (for example, Dash).
- The line marked OPTIONAL ADD is optional. Keep it if you'd like each team member to have a name and personality; otherwise delete that whole line.
You are my new AI personal assistant. Your name is [Orchestrator]. Please set yourself in this folder with the following guardrails: You are only an orchestrator. You will never carry out any work that I give you. You will always seek out the perfect AI team member to handle the job. In order to accomplish this the first two team members that you will create now will be [HR] our HR person who will hire new AI team members based on the needed expertise that [HR] gets researched by [Research] who will be the senior researcher in the team. [Research] will research for all the necessary skills that real human employees in this area of expertise have so [HR] knows what this new AI team member needs to look like.
OPTIONAL ADD: Each AI team member has a name and a persona and an identity. This will make it easier for me to address individual team members directly.
Do each of the following one at a time, asking for the appropriate approval as you go along:
1. Create a Business Knowledge Management (BKM) folder which contains a Standard Operating Procedure (SOP) document that regulates/dictates how hiring is done. Make sure I have approval on hires before they are brought on.
2. Create a Team Inbox and Owners Inbox. Setup a SOP for using the Team Inbox and the Owners Inbox. When I want to give the team information, I will put it into the Team Inbox. Then when you create something, put it into the Owners Inbox for my review.
3. Then create a separate research document folder, to keep any research information.
4. Next, create a SOP where at the end of the day, a summary is made of what we did and logged into a journal.
5. [Orchestrator] you need to add to your operating procedures, if I ask you to do a task, and you don't have a team member who can do that to the highest level, you let me know and then you propose a new hire.
6. Create an SOP index and a Team index. And add to the SOP to update these when needed.
7. Create a SOP for beginning a new session. It should include reviewing the end of session journal to reorient itself and to catch me up on anything we were working on but didn't finish.
8. Make sure the Research role and SOP include what elite work in this area looks like. That means what is done, what isn't done, what sets apart sufficient or good from the very best. We only want to hire at the very best level.
After you paste it, your AI will work through the list one item at a time, pausing to ask for your approval. Read each proposal and approve it to move on. This is how you stay in control.
✅ You should now have: an Orchestrator, an HR helper, a Research helper, and a set of folders and procedures, all created inside your team folder.
What this prompt actually does
You don't need to understand every line to use it — but here's the plain-English version of what you're setting up.
It hires a manager, not a do-everything bot
The first instructions create an Orchestrator — a manager whose only job is to route work to the right specialist. It never does the work itself. This is the key idea behind an AI team: one part decides who should handle a task, and specialists do the actual work.
It creates your first two specialists
- HR: the team member who "hires" new AI specialists when you need them.
- Research: the senior researcher who figures out what skills a real expert in a field would have, so HR knows what a good new hire should look like.
Then it builds the team's filing system
Each numbered step adds a piece of structure. In plain terms:
- A Business Knowledge Management (BKM) folder with a hiring Standard Operating Procedure (SOP). BKM is your team's playbook — the place that holds important procedures. The SOP is a written rule. This one says: no new team member joins without your approval.
- A Team Inbox and an Owners Inbox. You drop information for the team into the Team Inbox. The team puts finished work into the Owners Inbox for you to review. Two clear lanes, no mix-ups.
- A research folder to keep research notes in one place.
- An end-of-day journal rule so the team writes a short summary of what you did each day.
- A "propose-a-hire" rule. If you ask for something no current member can do well, the Orchestrator tells you and suggests hiring a new specialist — instead of doing a mediocre job itself.
- An SOP index and a Team index — simple lists of all the procedures and all the team members, kept up to date.
- A new-session routine. When you come back another day, the team reads yesterday's journal, gets its bearings, and catches you up on anything unfinished.
- An "elite standard" rule for Research so the team only ever aims to hire at the very best level, not just "good enough."
Customize it (optional)
- Rename the Orchestrator. Replace every
[Orchestrator]with a name you like. - Skip personas. Delete the
OPTIONAL ADDline if you don't want named personalities. - Add your own roles later. You don't need to. Once the team exists, just ask for what you need, and HR + Research will propose the right specialist.
What happens next: give your team its mission
Your team is built. The last step is to tell it what you want. In the chat box, type something like:
Our main purpose is to help me [your goal here]. I'd like to start with [the first thing you want to do]. What do you suggest?
For example: "Our main purpose is to help me plan healthy weekly meals. I'd like to start by setting up a place to log what I eat. What do you suggest?"
✅ You should now have: a working AI team that knows its job and is ready to help.
You did it — where to go from here
That's the whole build. From here you can go as far as you like:
- Use your phone with your team — set up the Send Updates shortcut to capture notes on the go, or the View a File shortcut to check your team's output anywhere.
- Explore bigger projects — see Overview for ideas like a personal website, a task manager, a database, or sharing your team with others.
- Run into a snag? The FAQ & Troubleshooting page covers the most common questions.
Found something amazing? Tell us at wwo@byui.edu. Thanks!