The pace of AI’s evolution this year has been incredible.
Co-coding with AI, auto-generated proposals, writing support, workflow automation — things that felt like “the future” just a year ago are now part of everyday work.
Still, many people say, “AI can’t do my job yet.”
But honestly, that’s not about AI’s ability — it’s usually about how we use it.
AI’s answers completely change depending on how you ask. If your question is vague, the result will be vague too. But when you give clear instructions and explain your goal, AI can surprise you with how sharp and helpful it gets.
So when people say “AI isn’t useful,” it often just means they don’t yet know how to ask the right questions. Today’s AI can reason like a college professor in many fields — the real question is whether we can use that brain effectively.
It’s a bit like leadership. Some leaders bring out the best in their teams, while others struggle to delegate. The same goes for AI — it all depends on how you guide it, how you communicate, and how you build that partnership.
Here are 7 practical tips I’ve picked up working as a PM — things that actually changed how useful AI became for me.
1. Be specific about your goal
Vague prompt = vague output. This is the most common reason people walk away from AI feeling unimpressed.
Instead of “write a project summary,” try “write a 3-paragraph project summary for a non-technical stakeholder, covering what we built, why it matters, and what’s next.” The more you spell out your intent, the closer the output lands to what you actually wanted.
Think of it like briefing a new team member. The more context you give upfront, the less back-and-forth you need later.
2. Set a role or persona
One of the quickest ways to improve AI output is to tell it who to be. “Act as a senior UX designer reviewing this flow” produces something very different from just “review this flow.”
I use this constantly in PM work. “Act as a skeptical client” before a proposal review. “Act as a junior developer who has never seen this codebase” when writing onboarding docs. Giving AI a role shapes both the tone and the level of depth it goes into.
It sounds simple, but it consistently produces better results than firing a prompt cold.
3. Give context and examples
AI doesn’t know your project, your team, or your company culture unless you tell it. Background information matters — a lot.
Before asking for something, drop in a few lines of context: who the audience is, what’s already been decided, what constraints you’re working with. Even better, paste in a short example of the kind of output you’re looking for. “Something like this, but for our situation” goes a long way.
The more relevant context you provide, the less time you spend editing the response afterward.
4. Specify the output format
Do you want bullet points? A table? A paragraph? A numbered list? Tell it explicitly.
“Give me this as a 5-row table with columns for task, owner, and deadline” is far more useful than “organize this information.” When you’re preparing something to share with your team or a client, format matters just as much as content.
This also saves you the time of reformatting output that came back in the wrong shape. Ask for what you need, exactly.
5. Iterate — don’t expect perfection on the first try
A lot of people try AI once, get an okay-but-not-great result, and conclude it’s not useful. The thing is, the first response is almost never the final answer — and that’s fine.
Treat it like a conversation. “That’s close, but make it shorter.” “This part is too formal — can you make it sound more casual?” “Add a section on risks.” Each round gets you closer to what you need.
In my experience, 2-3 rounds of back-and-forth usually produces something I’m genuinely happy with. That’s not a failure of AI — that’s just how collaboration works.
6. Use AI as a partner, not a replacement
There’s a version of AI adoption where people hand everything off and disengage. That doesn’t really work — at least not yet, and probably not ever for complex work.
The better frame is: AI handles the parts that slow you down, and you handle the judgment calls. AI drafts, you decide. AI structures, you refine. AI checks, you review. Your domain knowledge and your relationships with your team and clients — those still matter enormously.
When I use AI in project management, I’m not trying to automate the job. I’m trying to get more time back for the parts where I’m actually irreplaceable.
7. Experiment without fear
The fastest way to get good at working with AI is to just try things. There’s no wrong prompt. Nothing breaks. You can always start over.
Try asking it to critique your own work. Ask it to argue the opposite side of a decision you’ve made. Give it a messy problem and see what it says. Some of the most useful outputs I’ve gotten came from experiments I wasn’t sure would work.
Trial and error is the skill here. The people who get the most out of AI are the ones who aren’t precious about their prompts — they just keep trying different approaches until something clicks.
Closing thoughts
The winners in the coming years won’t be the ones who “beat AI” — they’ll be the ones who work well with it — much like how our IT director describes his own shift.
AI isn’t here to replace us. It’s here to collaborate with us. Those who treat AI as a teammate, not a threat, will shape the next era of work.
AI is evolving. The question is — are we evolving with it?
Working out where AI fits your team?
I run my own company on AI every day — the wins and the misfires included. If you’re deciding what to hand to AI and where people stay in the loop, a free 15-minute call is a good place to start. Not a pitch — just thinking it through together.
Book a 15-minute call →
Shin Uchida内田 伸
COO / AI Consultant / System Development Producer · Linnoedge Inc.
Shin has over 35 years of experience in the IT industry. He began his career in medical systems and later worked on large-scale system development projects for major corporations. Today, he produces business design and system development initiatives that integrate AI into practical operations.