AI Foundations
No slides.
No lectures.
You drive, I ride shotgun.
By the end of this hour, you'll have solved a real problem with AI.
First — select your industry
All examples and exercises will be tailored to your world.
Today's Class Plan
Here's what we'll cover in the next hour.
How This Works
This isn't a course you watch.
It's a skill you build.
You Bring
A real problem from your business.
Something you'd actually use.
The messier the better — we're not here for toy examples.
We Build
By the end of each session, you have finished work.
Something that works.
Something you'll use after this call ends.
The Mindset Shift
Google way: Type keywords → scan results → click links → maybe find something useful
AI way: Ask like you'd ask a smart colleague → get an answer → push back until it's exactly right
Your First Real AI Conversation
Time to actually use it.
Open ChatGPT, Claude, or Gemini in another tab and work through this with a real problem from your life or business.
Here's what to do
Pick a problem below — something real you'd actually use the answer for.
Copy the starter prompt into ChatGPT or Claude.
Fill in the blanks with your details.
Save what works.
Paste your best prompt and output into the notes below.
New Venture Research
You're launching something new — a product, service, or initiative.
Use AI to accelerate the research and positioning.
Improve What's Working
Something in your business works but it's manual or messy.
Get AI to build a tool or document you can use tomorrow.
Clear the Backlog
What's been sitting on your to-do list?
A vendor email?
Product descriptions?
A process to document?
Paste in the prompts that worked, the outputs you liked, and what you pushed back on.
This is your reference.
Push Back
That first answer?
It's a draft.
The skill is knowing how to make it better.
Do at least 3 rounds.
Common Push-Backs
"Too generic." → Make it specific to premium products, not budget.
"Too long." → Give me the top 3, not a wall of text.
"Wrong tone." → Write it like I'd explain to a business partner.
"Missing context." → I'm selling to camp directors who buy in bulk.
Click to mark each round complete.
Don't settle for the first answer.
Notice the gap between round 1 and round 3?
That's the entire skill this class teaches.
Basecamp Foundations is 5 classes of exactly this (Alpine Academy has 22 in all) — hands-on, one real problem at a time — until it's building assistants that do this automatically instead of you re-prompting from scratch every time.
See the program →
Research vs. Reasoning
This is the thing most people get wrong on day one.
Understanding this changes everything.
The Hard Truth
When you ask ChatGPT a question, it might go find the answer, or it might reason its way to something plausible using patterns it learned during training.
Mainstream AI now searches by default a lot of the time — but not always, and you often can't tell which one happened just by reading the answer.
Neither mode makes it "intelligent" the way people assume.
One mode finds real sources.
The other finds the quickest way to stitch together a few things that make sense and hand you an answer that sounds right.
The risk: assuming every answer is backed by a real source, or assuming it's all made up.
Either way, you're guessing instead of checking.
You ask: "What regions should I target for children's bedding?"
What happens: It thinks "cold + rich = buy bedding" and builds an answer from that logic.
No search happened here — sounds plausible, might be wrong.
You ask: "Search recent Amazon sales data and social media trends for children's bedding by US region."
What happens: It goes and looks at actual sources, pulls real data, and reports what it found.
Specific.
Verifiable.
Useful.
The Three Layers of "AI"
Everything called "AI" today falls into one of these buckets.
Knowing which one you're using changes what you should expect.
You ask, it answers.
It's not connected to anything, not doing anything automatically, not watching your inbox.
You pull every lever.
Input triggers output.
Send this email, update this spreadsheet, file this document.
Predictable, repeatable, no reasoning involved.
Your morning briefing, draft emails, route-aware reminders.
This is the building part.
before you ask the question.
Not a real gap.
Just pattern-matching.
Think of it like briefing a new hire: who are you, what's the job, what should you avoid, and what does the deliverable look like?
Spot the Difference
Click each response to reveal whether the AI researched it or reasoned it.
The Habit
Every time AI gives you a factual claim, ask yourself: did it look this up, or did it figure it out?
If you can't tell, ask it: "Where did you get that?"
Before & After
Watch a basic prompt transform into one that actually works.
Click an example below.
A structured prompt you don't want to rebuild from scratch every time is a template waiting to happen.
Save it, and that's the first rung of the ladder — template, then assistant, then automation, then system.
Here's What You Just Built
One real problem, three rounds of push-back, one usable output.
That's the whole method.
The gap: you just did this by hand, once, for one problem.
Basecamp Foundations is where you build the version that runs on its own — on your real data, for every problem like it — instead of re-prompting from scratch each time.
Optional, but this is how the skill sticks.
Solve another real problem
Pick something from your business.
Screenshot what worked.
Push back 3+ times
Refine until it's actually good.
Document the progression.
Catch it reasoning vs. researching
Ask it something factual.
Can you tell if it looked it up or made it up?
Push back: "Where did you get that?"
Use AI for 5 real tasks
Emails, research, descriptions — whatever comes up this week.
The habit to build:
