I've Used AI Every Day for Three Years. Here's What Nobody Tells You About Making It Work.
The secret to AI leverage: treating it like a slightly overeager junior analyst, not a crystal ball.
I’ve been using AI every day for more than three years now across revops, law, valuation work, content, app development, all of it. Long enough to get past the honeymoon phase. Long enough to be wrong a few times, too.
If you’re a senior exec trying to figure out how this fits into your world, or just a normal person opening an AI app on your phone at night after the kids are in bed, this is for you. These are the lessons that stuck after the experiments, the misfires, and the “wow, this actually works” moments.
It starts with a conversation, not a magic prompt
People keep hunting for “the perfect prompt.” The secret string of words that unlocks everything.
That has not been my experience at all.
What actually works looks a lot more like coaching. You give the model a clear brief, it gives you something back, you react, you go again. The magic is not in the first prompt; it is in the back and forth.
When I work with AI on a keynote or a long article, the first draft is rarely good. Sometimes it’s flat. Sometimes it sounds like a corporate press release. Sometimes it just misses the point. And the most important thing I’ve learned is, that is fine. The first draft is just the first pass from a very fast junior teammate.
So I talk to it the way I would talk to a junior analyst.
No, this sounds too polished. I want it a little messy and human.
You missed the emotional hook. Start with the moment when the room goes quiet.
This section is clever but useless for my audience; cut it and focus on what a CFO would actually care about.
You can feel the quality jump after two or three of those cycles. It starts to pick up your rhythm. It leans into your priorities. It drops the generic filler. The writing is still assisted by AI, but it actually sounds like you.
Same thing on the legal side. I’ve used AI to help draft clauses in speaking contracts, fee language, and cancellation terms. The first version is usually too soft or too aggressive. So I send it back. Tighten this. Make the obligations clearer. Remove the weird formality that sounds like a 1950s law textbook. Again, the power is not that it “nailed it” on the first try. The power is that I can move from nothing to a decent working draft in minutes, then I can spend my actual energy on judgment.
If you remember nothing else, remember this: you are allowed to tell the AI that its answer is not good enough. In fact, you should. That is where the good stuff starts.
Stop asking for answers, start building assets
In the beginning, I mostly asked AI questions. Explain this concept. Summarize this article. Draft an email.
Useful, yes. But not life-changing.
The real shift came when I started asking AI to build things I could reuse. Dashboards. Templates. Playbooks. Full-blown assets that live outside the chat window.
In revops, I’ll give it a scenario. Series A SaaS company. Here are the rough ARR numbers, win rates, sales stages, and conversion assumptions. I’ll ask it to design an executive dashboard, think through what the CEO needs to see weekly, what the head of revenue needs, and what an enterprise AE needs. Then I’ll have it generate the actual HTML or React for that dashboard, or the structure for a Google Sheet with formulas.
The first version is not final. But now I’m not staring at a blank page, I’m editing a working dashboard. I’ll say, make this less noisy. Group these charts differently. Put financials at the top, funnel in the middle, and risk indicators off to the side. Within an hour, I have something a board would recognize as a real management tool.
Same thing with law firms. Instead of “write me some prompts for lawyers,” I started asking, help me design a Legal AI Playbook. Give me a table of contents. Governance chapter. Risk and confidentiality chapter. Concrete workflows for things like drafting client emails, summarizing discovery, building a first pass on a brief, and building custody schedule options. Then I had it draft sections, not as a finished product, but as clay.
Now, that playbook is something I can refine, brand, deliver, and update. It is an asset. It is also a system, because it encodes how a firm will actually use AI Monday to Friday, not just what the tech can theoretically do.
Most people stop at “AI answered my question.” The leverage shows up when you say, no, help me build the dashboard that will answer this question every week. Or the template that every associate can use. Or the recurring summary I can send my exec team on Fridays.
That is when it stops saving you five minutes on a memo and starts compounding.
You are the architect, the models are your tools
The longer I use AI, the less I think of it as “one big brain in the sky” and the more I think of it like a workshop full of very capable tools.
Different models are good at different things. Some are better at long-form writing that sounds like an actual human. Some are better at code and data. Some hook more cleanly into your email, calendar, or CRM. Some are just less annoying to talk to for hours.
On top of that, you have features. Projects. Skills. Memory. Canvas. Connectors into Salesforce, Outlook, and Google Drive. All that plumbing matters. It shapes what is easy and what is painful.
So now I think of myself less as “the user” and more as an architect.
For a revops engagement, I might use one model to explore ideas, another to generate the chart code, and another to sanity check the math or test edge cases. For a law firm training, I might use one tool to design the curriculum, another to simulate client scenarios, and another to generate visual aids and worksheets. The client just sees a clean experience. Under the hood, it is me choosing the right tools in the right order.
But, and this is the key part, I never outsource judgment.
When we draft a cancellation clause for a speaking agreement with AI, a human still decides what is fair. When we generate a list of AI use cases for actuaries or litigators, a human still picks the ones that are actually worth trying this quarter. When I build a dashboard for a CEO, I’m the one who says, no, if you show it this way, they will chase the wrong number.
AI can analyze, write, mock up, refactor, and translate. It cannot care. It cannot own the consequence of being wrong in front of a client, or a judge, or an employee whose comp plan you just changed.
That part stays with you.
And strangely, that is what makes this whole thing exciting instead of scary. You are not handing your job to a machine. You are putting a ridiculous amount of leverage in the hands of the person who is willing to think like an architect, to learn a few tools, to ask better questions, to edit hard, and to be accountable for the final call.
So where does that leave you
If you are a senior executive, the big question is not “what can AI do.” You already know the answer is “a lot.” The question is, how do you personally want to work with it.
Are you willing to talk to it like a collaborator, not a search bar? Are you willing to build assets with it, not just get one-time answers? Are you willing to stay an architect, to learn enough about the tools that you can point them at the right problems and still own the outcome?
And if you are just getting started, you do not need a masterclass in prompt engineering. You need to open a chat and treat it like a smart, slightly overeager assistant. Ask for a draft. Tell it what you do not like. Ask again. Then, when you find something that works, save it, turn it into a little system. A pattern you can run tomorrow, next week, and the week after.
That is what I wish more people knew. Not that AI is perfect, or safe, or scary. Just that, used well, it lets you spend more of your time on the part that is still unmistakably human: judgment, taste, responsibility, the way you explain the world to other people. All the stuff we secretly signed up for when we took these jobs in the first place.
Business leaders are drowning in AI hype but starving for answers about what actually works for their companies. We translate AI complexity into clear, business-specific strategies with proven ROI, so you know exactly what to implement, how to train your team, and what results to expect.
Contact: steve@intelligencebyintent.com