The 30-Day AI Champion Bootcamp · Lesson 1 of 4
Week 1 — Identify the Wedge
> 30 min/day · 6 days · One-page problem brief by the end of the week
This is the week most aspiring AI champions get wrong. They skip it. They're so excited to build something they pick the first "cool" use case that comes to mind — *"let's build an AI meeting summarizer!"* — and end up with a shiny prototype that solves a problem nobody is losing sleep over.
Your goal this week is the opposite: find the problem at your company that is so expensive and so annoying that solving it is unignorable. That's the wedge. Once it's in, the rest of your AI program follows through the hole it makes.
A good wedge problem has four features:
- A specific, boring human suffers — not "improved team efficiency." Joe in accounts payable spends 3 hours a week chasing invoice approvals. *Joe* is the wedge.
- You can measure the cost in hours or dollars — not "productivity boost." Actual math: 3 hours × $32/hour × 52 weeks = $5,000/year, one person.
- AI is a plausible fit — reading, summarizing, classifying, drafting, lookup, extraction. If the problem is "we need a new warehouse," AI is not the answer.
- You have access to the data and the person — you can see the emails, spreadsheets, tickets, whatever. If it's locked behind security you don't have, pick a different problem.
Day 1 — Observe (30 min)
Open a blank doc. At the top write: *"Things at my company that take too long."*
Spend 20 minutes writing down everything you've seen someone complain about in the last month. Ignore AI. Don't filter. Examples:
- "Jenna spends every Monday morning re-typing contractor timesheets into the payroll system"
- "Our sales team writes the same five follow-up emails 20 times a day"
- "Customer support has to dig through a 400-page PDF manual to answer warranty questions"
- "Finance takes 4 days to close the month because someone has to reconcile two spreadsheets by hand"
- "Onboarding a new hire means IT sends a 12-email thread to get them all their accounts"
Aim for 10–15 items. Don't edit. Volume first.
Spend the last 10 minutes looking at your list and circling the 3 that look like they involve *reading, writing, extracting, or summarizing information* — those are the AI-shaped ones.
Day 2 — Interview (30 min)
Pick ONE of your three candidates — the one where the affected person is most accessible to you. Message them today. Ask for 15 minutes. Something like:
> *"Hey — I've been poking at some AI tools on my own time and wanted to understand the [invoice approval / weekly timesheets / whatever] process better. Do you have 15 minutes this week to walk me through how you actually do it? I'm not trying to automate you out of a job, I'm genuinely curious where the painful parts are."*
That second sentence is not optional. Say it every time.
While you wait for them to reply, do the same thing with candidate #2. Get two interviews on the calendar.
Day 3 — Conduct interview #1 (30 min)
Run the interview using the Coworker Interview Script template (linked at the bottom of this lesson). It's 8 questions, takes 15 minutes, and focuses on:
- *Walk me through the last time you did this — what did you open, what did you type, what did you copy?*
- *Where do you get stuck? What makes it take longer than it should?*
- *If this was magic, what would you want it to do for you?*
- *Who else does this? How many hours total per week, across everyone?*
Take notes. Don't pitch AI. Just listen.
Day 4 — Conduct interview #2 (30 min)
Same process, second candidate. Between the two interviews you'll notice your gut has picked a favorite. That's fine — but force yourself to finish both before deciding.
Day 5 — Do the math (30 min)
Open the ROI Worksheet template. Fill it in for your favorite candidate.
The math is intentionally basic:
``` (Hours per week saved × loaded hourly cost × 52 weeks) minus (Tool cost per year + build time at your hourly cost) = Annual savings ```
Don't game it. If the honest number isn't compelling, your instinct was wrong — pick the other candidate. If the honest number is good, write it down and move on.
A typical first wedge lands between $5,000 and $30,000/year in hours saved. That's perfect — it's big enough to matter and small enough that nobody has already tried to solve it.
Day 6 — Write the brief (30 min)
Use the Problem Brief Template. One page. It has five sections:
- The problem, in one sentence — "Joe in AP spends 3 hours a week chasing invoice approval signoffs via email."
- Who suffers, and how much — the person, the hours, the dollars
- Why it's an AI problem — what specifically does AI need to read/write/classify to help?
- What I'm proposing (vague is fine) — "A small tool that reads invoice emails and auto-drafts approval reminders."
- The first experiment — "I'll build a working demo in Week 2 using [Copilot Studio / Claude API / Power Automate]."
You are done with Week 1 when you have that one page written.
You'll find all of these linked in the templates folder below this lesson — copy them to your own docs, fill them in, keep them:
- Coworker Interview Script — 8 questions + what to listen for
- ROI Worksheet — simple annual-savings math (Excel or Google Sheet)
- Problem Brief Template — the one-pager you'll walk into leadership with in 3 weeks
Most "AI initiatives" fail because they start with the tool ("let's use Copilot!") instead of the problem. By forcing yourself to find a real human suffering a measurable amount, you've already done more than 90% of the "AI champions" at your company's competitors.
Move to Week 2 — we're going to build something that actually solves the problem you just identified.