"The right tool for the right problem. The wrong tool confidently applied is worse than no tool."
What This Guide Is
This is not a strategy guide about AI.
It will not tell you how to "leverage AI across your organization" or "implement AI-driven workflows at scale." There are hundreds of those documents. You've probably read some of them. They contain real ideas executed at the level of the boardroom — useful for deciding what to fund, useless for deciding what to do with your own time.
This guide does two things:
- Teaches you to use AI with your own hands — open a tool, describe what you need, produce something real.
- Teaches you where each tool doesn't belong — so you use your development team, your writers, your analysts for the right problems and AI for the right problems.
Getting both right is what makes this a capability rather than a novelty.
Why This Is Different
There are two ways an executive can engage with intelligence work — research, writing, building, analysis:
Delegation: You describe what you need. Someone interprets the description. The interpretation passes through layers. Something comes back that resembles what you asked for, weeks later, after three rounds of revision.
Direct work: You do it yourself, with a tool that makes it fast enough to be worth doing. The only interpretation is yours. What comes back is what you meant.
Both are real. Both have their place. A 50-person research project requires delegation. A competitive analysis you need to understand your own market doesn't.
The mistake most executives make is applying the delegation model to every problem — including the ones that are small, specific, and so close to their own judgment that nobody else could possibly understand the requirements as well as they do.
AI tools exist for those problems. Not for all problems.
What AI Actually Is Good At (and What It Isn't)
| AI tools ARE for | AI tools are NOT for |
|---|---|
| Research and synthesis across large bodies of information | Replacing your judgment about what the information means |
| First drafts of things you know well | Writing anything where your authentic voice is the product |
| Personal tools and automations | Production systems other people depend on |
| Meeting capture and follow-through | Decisions that require presence and relationship |
| Prototyping ideas before they involve your team | Validating whether the idea is right |
| Processing messy data into readable form | Guaranteeing the data is accurate |
| Tasks you do repetitively that follow a pattern | Tasks where every case is genuinely different |
The last column matters more than the first. AI is not dangerous because it fails obviously. It's dangerous because it fails plausibly — producing outputs that look correct until someone with judgment checks them. That person has to be you.
The Three Modes
This guide covers AI use in three modes:
Building — using AI to create software tools with your own hands. Dashboards, scripts, automations, prototypes. This requires a terminal and some patience. The payoff is tools that exist nowhere else because nobody else knew exactly what you needed.
Operating — using AI for the daily knowledge work of running a company. Research, writing, meeting capture. This requires no technical background — just knowing which tool to reach for and how to evaluate what comes back.
Directing — using AI to handle tasks autonomously while you're doing other things. This is the emerging edge: agents that can file documents, run reports, manage workflows. This guide covers where it's mature enough to rely on and where it isn't yet.
Most executives start in the Operating mode (it requires the least setup) and gradually build a Building habit. The Directing mode becomes useful once the first two are stable.
What You Need
For the Building section:
- A Mac or Linux computer (Windows works via Git Bash or WSL)
- $20/month (Claude.ai Pro) or $100/month (Claude Code Max)
- Two hours for your first session
- A specific, small problem you want to solve
For the Operating section:
- A web browser
- A free Perplexity account, a free Fathom account
- A specific research question or document you're working with
The last item in each list is the most important. Do not start without a real problem.
Generic exploration produces nothing useful. The executive who opens any AI tool and says "let me see what this can do" will spend an hour impressed and leave with nothing built. The executive who opens it with a specific problem will leave with that problem solved.
Start with the problem.
What This Guide Is Not
Not a coding tutorial. You will not learn Python or JavaScript. You don't need to. The building tools write the code. You direct the build and evaluate the output.
Not a guarantee. Some projects will fail on the first attempt. Some research will be wrong. That is normal. The guide teaches you how to handle it.
Not a case for replacing your team. AI tools are for your problems. Your engineering team is for your product. Your analysts are for the work that requires institutional knowledge and accountability. AI accelerates the in-between — the things that fall through the cracks because they're too small to brief someone on but too important to let slip.
The Decision
You're reading this because something prompted you to ask whether you could do something yourself.
That instinct is correct — for the right class of problems.
This guide will show you what that class looks like, which tools fit it, and when to step back and hand off. That last part is as important as the first.
Next: The Stack