A hands-on introduction to building your own agentic AI team
Adam Lloyd · Heidi Egan · Cindy Goodwill — BYU-Idaho
Welcome! This is a hands-on introduction to building your own “AI team” — a small set of AI helpers (agents) you set up to work together on your real projects. In the next few minutes you’ll meet the three of us, see real examples we actually use, hear why a team of agents beats a single chatbot, and watch one get built live. Good news for newcomers: this all starts from plain files and plain-language instructions — no coding required. As you watch, ask yourself: what would you want an AI team to help you with?
Introductions
Cindy GoodwillWWO Curriculum & Instruction Director
Heidi EganProgram Curriculum Designer
Adam LloydProgram Curriculum Designer
We’re three curriculum people from BYU-Idaho Online Programs — Cindy Goodwill, Heidi Egan, and Adam Lloyd. Each of us built an AI team for our own day-to-day work, so the examples coming up aren’t polished demos — they’re tools we really use.
Show & Tell
Example 1 — Tim: Meal Planning
Menu planning for the week
Creating the shopping list from the menu
Tracking items bought and their prices
Tracking the cost of each meal
Planning for food storage
Cindy’s meal-planning team — its assistant is nicknamed “Tim” — runs the whole food routine: it plans the week’s menu, turns that into a shopping list, tracks what was bought and what it cost, adds up the cost of each meal, and plans for food storage. The image is a sample weekly menu it produced, complete with real per-meal prices. What’s a weekly chore at home you’d love to hand off?
Show & Tell
Example 2 — Heidi: Zettelkasten (Smart Notes)
A “Zettelkasten” is a note-taking method where every idea becomes its own small note and you link related notes together, so patterns show up over time. Heidi’s AI team automates the whole thing: she drops her notes into an inbox file, one agent files each note into a database on her own computer, another breaks them into single-idea “atomic” notes and links them, and a third builds the dashboard you see here — surfacing the connections and the reminders she left herself. The best part: it works with notes she already had, in the tools she already uses. Do you have a pile of notes you’ve never had time to connect?
Show & Tell
Example 3 — Adam: Personal Task Manager
Adam had his AI team build a personal task manager. One agent designed the underlying database, and a “personal assistant” agent takes in his tasks, sorts them by priority, and schedules them. What you’re seeing is his real dashboard. What’s the to-do list that always seems to slip through the cracks for you?
Show & Tell
Task Manager — Day View
This is the same task manager, zoomed in to a single day. The assistant lays out what to work on and when, so Adam can just follow the plan instead of re-deciding all day long.
Show & Tell
Task Manager — Week View
And here’s the whole week at a glance. A nice detail: it’s a real database with a manual interface, so Adam can add tasks or check them off directly — with or without the AI. He leans on the AI team most, but likes having the hands-on controls too.
Show & Tell
Example 4 — Cindy: Physics (Backwards Design)
Backwards design: start with the exams, then build the assignments and lessons.
Hired a faculty of experts: assessment, lab design, physics, math, instructional design, and lesson-standard specialists.
The team defined the rules for a good lesson, then wrote lessons — with labs.
It even suggested spiritual thoughts to pair with the lessons.
Cindy’s team used “backwards design” — start from the final exams, then build the assignments and lessons that get students there. She “hired” a faculty of specialist agents (assessment, lab design, physics, math, instructional design) that first agreed on what makes a good lesson, then wrote the lessons — labs included, and even suggested spiritual thoughts to pair with them. The screenshot is the PH 121 “Big Picture” site the team produced: an interactive map of the whole course. If you teach, what would you build first if you started from the exam and worked backward?
Why agent teams, rather than out-of-the-box AI?
SimplePlain text files — no coding needed
DurableTools change; your text files don't
PortableMoves with the file's owner
Tool-agnosticWorks across different tools
OwnershipYour files, your control
Why bother with a team of agents instead of just chatting with one AI? Because the team is built on plain text files, and that changes everything. It’s simple (no coding), durable (tools come and go, but text files stay), portable (it moves with you), tool-agnostic (use whichever AI you like), and fully yours (you own the files). Those five qualities are what make this approach last.
Basic Concepts
Know what the team will do
Work with it as a consultant
Get its plan before it builds
Balance model level vs. tokens
Brain-dump; let it ask questions
A few habits that make working with an AI team go smoothly: have a rough idea of what you want the team to do; treat it like a consultant and think out loud with it; ask it to explain its plan before it starts building (that saves time and money); pick a stronger model for hard work and a lighter one for simple tasks; and don’t be shy about brain-dumping — let it ask you questions until it truly understands. Which of these would you try first?
Advanced Concepts
What it may connect to on your computer
Where you store the team folder
Choose your tool for its strengths
Sharing a team
Google / Apple / OneDrive
Connecting on your phone
Privacy and security
Once you’re comfortable, these are the next things worth thinking through: what you let the team reach on your computer, where you keep its folder, when to pick one AI tool over another for its strengths, how to share a team with someone else, storing it in the cloud (Google, Apple, or OneDrive), using it from your phone, and keeping everything private and secure. Happy to dig into any of these — which one matters most to you?
Advanced Applications (if time permits)
Building a website (with a database)
Scheduled actions
A personal dashboard
Connecting to a notes tool
Connecting to Gmail
API calls
Here’s where it can go once the basics click: building a website (even with a database behind it), scheduling actions to run on their own, spinning up a personal dashboard, connecting to your notes app or Gmail, and calling other services through their APIs. We’ll cover these if time allows — is there anything here you’d love to see?
AI was used in the making of this presentation… along with a human. 🤝
Thanks for spending this time with us! We’d love your questions — about anything you saw today, or about getting started with your own AI team. And yes: AI helped build this presentation… alongside a human. What’s the first thing you’d want to try?