AI really is a “boring” problem once you get close to it. Boring and exciting.
As a part of this series, we had the pleasure to interview Rob Collie.
Rob Collie is the founder and CEO of P3 Adaptive, a Microsoft Solutions Partner for Data and AI that serves hundreds of mid-market and Fortune 1000 clients. A former Microsoft engineering leader on the Excel, Bing, and Power BI teams, Rob led the Power BI wave after leaving Microsoft and has authored three previous business-technology books (92,000+ copies sold). He also hosts the Raw Data with Rob Collie podcast. His fourth book, Fair Game: Customizing AI to Your Business Is Easier Than You Think (August 2026), turns that practitioner’s credibility on the AI moment.
We had the pleasure of talking with Rob Collie. Rob is the founder and CEO of P3 Adaptive, a Microsoft Solutions Partner for Data and AI that serves hundreds of mid-market and Fortune 1000 clients. A former Microsoft engineering leader on the Excel, Bing, and Power BI teams, Rob led the Power BI wave after leaving Microsoft and has authored three previous business-technology books (92,000+ copies sold). He also hosts the Raw Data with Rob Collie podcast. His fourth book, Fair Game: Customizing AI to Your Business Is Easier Than You Think (August 2026), turns that practitioner’s credibility on the AI moment.
Thank you so much for joining us in this interview series. Before we dive into our discussion, our readers would love to “get to know you” a bit better. Can you share with us the backstory about what brought you to your specific career path in AI?
The biggest thing that brought me to an AI career was necessity, just like I think it’s going to eventually play out for everyone. I have a background in software and data stretching back three decades, but AI was still every bit as new to me as it was to the average person. I had no idea how it worked, what it was good for in a business context, or how it was going to disrupt my profession. All I had was an increasing sense of anxiety with no clear answers on “what should we be doing to survive this change?”
We’re a small company (50 people), and I’m CEO, so if we were going to get answers, they had to come from me. About eighteen months ago, I eventually forced myself to confront the anxiety and dive deep into answering those questions. At some point along the way, I realized that AI was going to be our company’s future, and that it was a natural extension from the existing “data and dashboards” work we’ve done for our clients since 2013.
Can you share the most interesting story that happened to you since you started working with artificial intelligence?
The two most surprising things for me as a human being are one — how simple AI is to understand (if someone had just bothered to explain it to us better) and two — how excited I’ve become. Yes, there’s still some anxiety for sure, but that’s receded a bit. Excitement is now the stronger emotion, and I was not excited at all — just afraid — before I began my own little personal journey of discovery.
As a professional, the most interesting thing was when I realized that when you’re building AI for your business, no one needs to become an expert LLM researcher, or really even understand how LLMs work. Building effective AI systems and agents for your business is about “boring” stuff like regular traditional software (which we all grew up with) and about data (which we interact with in business every day). When you’re building AI for your business, the one thing you’re actually never building is the AI itself. You’re building around AI that others built. You’re plugging into AI like it’s a wall outlet. AI really is a “boring” problem once you get close to it. Boring and exciting.
You are a successful leader in the AI space. Which three character traits do you think were most instrumental to your success? Can you please share a story or example for each?
First, I believe that things change much more than they stay the same. The status quo holds no “home field advantage” when change starts to happen, and what feels normal and stable today is always just an accidental and temporary condition. This helped me internalize — sooner than most people — that AI wasn’t just hype, and that it really was going to change everything. Even as long as two years ago, I was watching other consulting firms — ones with deep pockets — rushing to outsource technical work overseas and cut costs, and I was scratching my head a bit. Long before I understood how to build AI agents, it was pretty clear to me that LLMs were going to be “the new offshore,” and that we needed to be getting closer to our customers, not farther away. I think we’re now seeing that prediction proven correct. And that’s not because I’m smart enough to see the future. I’m not. I just don’t give the status quo as much weight as others do.
Second, I really believe in the power of incentives. Whether with other companies or with your own employees, if you can engineer the relationship so that you both win at the same time, you tend to win a LOT more often. For example, for many years our company charged by the hour simply because that’s what our industry expected. But with AI tools accelerating development timelines, hourly pricing suddenly incentivized us to not embrace those tools. That’s bad for our clients, because it deprives them of higher ROI, and ultimately bad for us, because eventually we’d go out of business. So we’ve been re-engineering our engagement models around this idea: AI provides a huge productivity boost. How do we share those benefits, so that our company and our client both win when we do it right?
Can we leave it at two? I gave longer answers than they’re expecting. But I really like those answers and would prefer not to dilute them too much.
Let’s jump to the primary focus of our interview. Can you share a specific example of how you or your organization used AI to solve a major business challenge? What was the problem, and how did AI help address it?
So many examples. One is that we used AI to cut our SaaS bill. A lot of SaaS systems have crept up in price over the years, and are now very expensive relative to what they should be. But customers paid it because they were trapped. Suddenly, AI-powered development turns the tables on SaaS vendors, and you can build your own replacements. Even if you pay to have them built, the ongoing savings can be significant — and the customized versions fit your company better than the original off-the-shelf versions. At our company, we haven’t replaced all of our SaaS systems, but we’re saving five figures a month — at a 50-person company. And honestly if SaaS costs were a primary focus for us — as opposed to building things for our clients — I think we’d have replaced a lot more of it already.
For sharp contrast, our consultants need to acquire a particular new soft skill (blank page, human factors software design) in order to deliver max value to our clients in a world where suddenly the sky is the limit. That’s a skill that took me years to learn at Microsoft, and I could teach them to the team, but I would need to do it one-on-one and spend probably a month or more with each person. Instead, I worked with our developers to create an interactive “coach” (named Archie) who patiently teaches these lessons to our team — in private, one on one chats that actually feel fun for the consultant. We’re scaling mentorship of a soft skill. Which is wild.
What are some of the common misconceptions you’ve encountered about using AI in business? How do you address those misconceptions?
The biggest misconception is that we should wait and find out what happens. AI is now good enough — and understandable enough — that you absolutely can be moving today. Which is the whole reason I’m writing my book.
The second one is that you can just give ChatGPT or Copilot subscriptions to your teams and your AI strategy will just emerge organically. That is a dead end. Off-the-shelf AI is impressive, for sure, but it’s mostly impressive in our personal lives and individual professional use. As soon as we bring it into a business context, it’s like a new hire with a PhD in everything. It knows about physics, and it knows about Roman History, but it doesn’t know any of the important details of your business. So you start “teaching it,” like you might teach a new hire, but as soon as you start a new chat, it’s forgotten most everything. So you quickly stop investing in teaching it. But you can build customized versions, and the things you teach them, they remember. This is the big unlock moment: successful business AI is customized AI, and customized AI is not nearly as hard as it might sound.
Another one is that one person can now manage fifty. That’s never going to happen. The human touch points and random crises alone will bury a manager of twenty people, much less fifty. I think AI makes most management jobs more efficient, but only by a fraction — maybe 30%, depending on the role.. If a manager’s job is such that they can manage fifty people with AI, what we’re really saying is that we should just replace the manager with AI. I think upper management at certain companies wanted to have it both ways — be hyper efficient, cut costs to the bone, but still have a human to blame if things go wrong. That’s not going to work.
In your opinion, what is the most significant way AI can make a positive impact on businesses today?
The biggest way to make an impact today is just to get started. Once you’ve seen what a few customized AI solutions (agents) can do — even modest ones — the lights REALLY come on. So pick some workflow at your company. One that could benefit from more human attention than it currently gets. There are a lot of workflows that resemble “making the rounds” — checking in with colleagues, customers, vendors, and/or many different software systems. These workflows are usually not someone’s primary job, and they’re time consuming and tedious, so the workflows get neglected — sometimes to the extent that no one’s even imagined it could be a workflow. Now it can! Customized AI is really good at stuff like that, and suddenly you have “someone” who is tirelessly making sure everything is on track — and who lets a human being know (or maybe even takes action) when something isn’t.
Ok, let’s dive deeper. Based on your experience and research, can you please share “5 Ways AI Can Solve Complex Business Problems”? These can be strategies, insights, or tools that companies can use to make the most of AI in addressing their challenges. If possible, please share examples or stories for each.
1. Make data access simpler — coming from the BI world, we have this illusion that dashboards are simple. And they are — when you know where the dashboard is and are familiar with using it. But most business questions start with “do we even have a dashboard that answers this? If so, where? How do I find it?” Sadly, there often isn’t a dashboard for that question, so the search is fruitless. People learn not to even go looking. Your careful BI investments fail to meet the humans where they are, and so they don’t pay off. But it’s 100% possible today to create “chat with data” agents — trained chatbots who can go look up answers to any question you have — whether there’s a dashboard for it or not. Just ask a question, get an accurate and certified answer. It’s a gamechanger.
2. Start looking into “semantic models” — AI agents of all flavors — not just “chat with data” — will need access to your company’s structured data in order to be effective assistants. And you cannot just let AI “figure out” the definitions of key metrics, or how to identify the same customer across systems, because they will make expensive mistakes. You need to give your agents “decoder rings” which unambiguously answer those sorts of questions for the agent. If your company has been using Power BI, you’ve already been building semantic model decoder rings without knowing it, so you’ve got a big head start here. Other BI vendors are starting to offer semantic models, too, because of the AI impetus. Regardless, it’s not too late — at all — to get started. Lately we’ve been helping a fair number of clients adopt Power BI for the first time — because they see it as a stepping stone to AI.
3. Start looking for your “Crafters” — I talk about this group a lot in the book, and they are going to be a big and unexpected part of your AI strategy. These are the “shadow IT” people. The spreadsheet gurus. The SharePoint and WordPress admins. Once they discover AI-assisted software development, and understand how custom AI systems work, you will be amazed at how valuable some of these people are. I spend several chapters on this process in my book and that won’t fit neatly into our conversation here, but for now, just start looking around. Start noticing these people and filing them away in your head for later.
4. Avoid “we’ll figure it out” layoffs — there is a lot of groupthink pressure coming from corporate boards and absentee ownership to just lay a bunch of people off, like that will then force the organization to compensate by adopting AI. I do believe that jobs are going to change — some will go away and others will open up — and that necessarily means some people will get laid off eventually (and then go find some of those new jobs, hopefully). But your existing “leaderboards” of who is most valuable at your company — both the roles and the individuals — are going to get scrambled. Some formerly-secondary roles and people will become more important tomorrow, and vice versa. I believe you need to start developing a strategy, and evaluating your existing roles and individuals through the new lens. Otherwise you might be unknowingly jettisoning the people you most needed to make the AI transition.
5. Try out Claude Cowork — this is one of the easiest, quickest ways to start turning on the “AI lightbulb” in your head. It won’t be your company’s “answer” to custom AI, but it’s a great way to get the gist of it. Download and install it. Pay the $20 subscription. Create a folder on your computer called “My First Project,” then open that folder in Cowork. Tell Cowork about a research or work project in your life — my daughter used it as her job-hunt assistant and coach. I used it to edit my book. My wife is using it to manage a complex medical issue. Drop any background documents, or things you’re working on, in the folder. Tell Cowork to look at those. And as you teach Cowork about this project and what what kinds of help you need, ask Cowork to “update claude.md” (precise words) with those new rules and procedures you just provided. It will remember those across chats. It’s like an employee handbook. My book editor’s handbook is over 5,000 words, built up slowly over time as we worked together.
How can smaller businesses or startups, with limited budgets, begin to integrate AI into their operations effectively?
I think small-to-midsized companies are actually in the most advantageous spot here. They have the least inertia to overcome — fewer levels of management to navigate, less political baggage and overhead. Plus, everyone at smaller orgs tends to wear so many hats that they immediately know what isn’t getting enough attention. Good ways to use AI are easier to spot when they are problems you yourself — or your immediate colleagues — live every day. It’s not expensive to get started with custom AI, but you need to understand what it is first, and who on your team can help pull it off. Again, I wrote my book to address both of those — giving you the confident understanding of what it is, how to build it, and who should do what in the process.
What advice would you give to business leaders who are hesitant to adopt AI because of fear, misconceptions, or lack of understanding?
First, I would tell them, “I get it.” I’ve been there — even though my entire career has been in tech. So first, don’t feel bad. Virtually every business leader — outside of the Big Tech C-Suite crowd and Fortune 500 CIOs — is in exactly the same boat as you. I talk to people like you every day. You are just like everyone else, despite what the LinkedIn influencers would have you believe.
The second thing I’d say is that you feel like you’re late to the party, but you can actually still be early. Which again, is also how I feel. I was almost late, but I ended up being early. That can happen for you, too, which is why I am so anxious to get my book on shelves.
In your opinion, how will AI continue to shape the business world over the next 5–10 years? Are there any trends or emerging innovations you’re particularly excited about?
No one knows where this is headed, really. I hesitate to predict two years into the future, much less 5–10. But no matter how fast the tech moves, people still need to digest it before it changes the world. Right now the big AI and tech companies are racing so hard against each other that they are making next to no effort to educate the broader public on what it all means.
We’ve reached a point already where the capability of these LLMs is not remotely being harnessed by the business world. Adoption is lagging far behind the LLMs’ level of capability because of the education gap. And still the AI firms race forward.
So I’m most looking forward to business starting to understand what this all really means and how to use it. We might discover that we don’t even need the AI companies to keep pushing the envelope. (They will anyway, of course).
How do you think the use of AI to solve business problems influences relationships with customers, employees, and the broader community?
Do NOT let AI speak for you, because it will not take long before you get burned. When we’re generating opinions the old fashioned way — without AI — there’s a built-in safety mechanism. A filter. And its name is “yes this idea actually came from my brain, I thought it through.”
If someone asks you for an opinion, or to come up with a plan, and your first move is to ask AI to help you generate your response, you are bypassing that safety mechanism without knowing it. When the LLM gives you its response, you just scan it for obvious mistakes. And when you don’t see any, you will then send the response to other people, representing it as your opinion.
The problem is, the LLM will have said things that you would not have said, if you had written it yourself. The scan for obvious mistakes isn’t going to remove those things, because they sound reasonable to you on a quick read. But they are sometimes not reasonable to the person who receives the response. I’ve seen professional relationships badly damaged by this dynamic — where the recipient is very upset by something that the other person never actually said — the LLM said it, but when you sign your name on it, now it’s yours.
In contrast, when you’re having an interactive conversation with AI — like asking it for advice on how to approach a problem — you are present and representing yourself in that conversation. If the LLM suggests something to you that’s off-base, you instinctively push back — just like you would with another human being. You explain “yeah, that’s normally a good idea, but…” and the conversation takes the turn that it needs to. There’s a better built-insafety mechanism for using AI like that. But there isn’t one for the “help me generate an opinion” case.
Bottom line: When you let AI speak for you, you’re then just reviewing a conversation between two other people — the LLM and the recipient. “You” have been removed from the equation, but you will bear the consequences. Generate your own opinions first, and then it’s fine to get AI assistance on refining or troubleshooting them. But the raw material always needs to be yours.
You are a person of great influence. If you could start a movement that would bring the most amount of good to the most amount of people through AI, what would that be? You never know what your idea can trigger. 🙂
I really like helping people “catch waves.” Helping them realize a potential they didn’t know they had, capitalize on new opportunities, and avoid becoming irrelevant. It’s in my core wiring, and it’s the thing that gives me the most satisfaction. It’s what I’m doing as leader of my company right now. I could be the CEO who says “everyone figure this out or you’re fired,” or the CEO who sees a threat coming and jumps ship to safer ground. But I love the challenge — and the payoff — of taking people with me on journeys. It’s not a moral high ground, it’s just what I enjoy doing the most.
Same thing with the book, but different. Frankly, writing a book is hard on me and my family. It’s like having a second full time job for several months. But I felt like I had to do it. In 2012, when I saw what Power BI (then Power Pivot) could do, and what it was going to do for a certain kind of person, there were no books on the market explaining those things. So it honestly felt like an obligation to write my Power BI books, and they helped a lot of people.
I felt the same way about this book. I didn’t go on my AI journey so that I could write a book, I went on that journey to plot a course for our company. The book is about sharing what I found, because I think so many people need to know it, too.
So in a way, I’m already trying to start a movement — one local (our company) one broader (the book). A “now you can” type of movement. So we can all stop feeling a bit less fearful and a bit more confident and empowered.
How can our readers further follow you online?
Linkedin primarily: https://www.linkedin.com/in/robcollie/
Soon: FairGameBook.ai
This was great. Thank you so much for the time you spent sharing with us.
Rob Collie of P3 Adaptive On How Artificial Intelligence Can Solve Business Problems was originally published in Authority Magazine on Medium, where people are continuing the conversation by highlighting and responding to this story.