Inc.
AI can supercharge a business, but only if the basics are solid first: data, architecture, governance and strategy.
How To Build a Successful AI-Powered Company
Author: Joe Procopio
Over the last two weekends, I’ve been writing up informal “weekend” guides for anyone wanting to get their feet wet with AI and start working towards the goal of building or expanding their business.
The results have been amazing! It’s like freakin’ Cape Canaveral with all the launches around here. The emails and messages have been pouring in from people, just like you, who are finally getting their brains wrapped around the “why” of AI.
Which is so much more valuable than the “how.” Hell, Claude can give you the “how.”
This weekend, I want to talk to the people who are already underway building, or catch folks right before they get started. Maybe you already have a business running, or you have one in mind, or you want to explore ways of implementing AI that can accelerate your business without screwing it up.
We’ve seen AI do both of those things. It can be a catalyst for a satisfying success or a dizzying screw-up. And either outcome can happen to all kinds of businesses, at all kinds of sizes, and at all levels of technical maturity.
Yeah, even those tech billionaires have a tendency to chase tech that’s not quite ready to execute the vision that’s in their heads.
So let’s take five minutes to make sure you’re ready for your AI-powered business, even if you’re already underway.
And it’s the weekend, so let’s make it fun.
Ready, Fire, Aim!
Now, these aren’t “how to” guides, and they don’t get super technical. And this is because there’s no need to do technical deep dives for smart people who all have an always-on “search and explain” partner at the ready.
Instead, in these guides, I do the same thing I’ve been doing with cutting edge technology for 30 years, and AI for 20 of those years – talking about the “why” of the tech behind every feature, build, and business idea that separates success from screw-ups.
In consulting on AI, including the last two years full time advising on AI readiness, AI strategy, and actually building AI-driven products, I can say without hesitation that the vast majority of “AI implementation screw ups” happen long before the first line of code is generated.
It’s in the preparation. Or lack thereof.
Maybe you’ve already implemented AI, maybe a bunch of it. In fact, in the last two weekend installments, I begged you to go forward, digital guns blazing, and have your chatbot coding sidekick and a team of assassin agents help you solve a thorny problem that might very well serve as the foundation of an eventual business success.
The time is now. The tools are there. Get started.
But this weekend I’m going to go all Buzz Killington on you, especially if you’ve already had some success.
Because it’s the lack of preparation, foundation, governance, and strategy that destroys most any successful new business. I’m not just talking about AI businesses or AI-built businesses either. The only thing AI brings to that equation is that its ease and ubiquity go a long way to making that careful preparation seem less than required, and that thoughtful strategy feel more like an afterthought.
Then all your data gets eaten. Or leaked. Or stolen and held for ransom.
So now that you’ve got something good and potentially valuable going, or even if you don’t yet, you’ll want to spend some time on these preparation requirements. By the end of the weekend, I want you to be well on your way to making sure you not only know the answers to these questions, but also “why” the right answers to these questions are important.
You Need To Know Your Architecture
Data is the most important factor in any technical success, for real this time!
Yeah, it’s something we keep saying, but it’s never been not true. In fact, the need to have proprietary, perpetually growing, properly structured, and securely accessible data is exponentially more critical for any kind of AI project.
First of all, you need to know what your data is, and why it’s the foundation of the value of what you’re building. You’ll need to know how and where it’s stored, and how it flows through your system. Want to create a data architecture diagram? Great! You can look one up and copy the format or you can literally sketch one out on a napkin, as long as you have a reference.
That reference should help you understand where the risks are in your architecture or structure, what’s broken or will break, and how to mitigate those risks and fix those problems. You need to know who owns your data (ideally, you) and who within your organization “owns” that data – the ones responsible for keeping the data clean, up-to-date, protected, and used properly and ethically (probably also you).
You need to identify what data are important, Critical Data Elements (CDEs), and what data are sensitive, like Personally Identifiable Information (PII). You’ll need to offer the right access to the former and add the right controls to access for the latter.
There. I just did data governance. In three paragraphs. Hopefully, you’re still awake. I dozed off a little in paragraph two.
You Need To Know Where the Value Is and How To Expose It
That’s a lot of words and, unfortunately, the bad news is it’s also a lot of work.
I had a conversation this morning with a board member of a company who told me his company is enriching data that becomes extraordinarily valuable after enrichment – he went on to explain why – but the problem was, like a dog chasing a car, the company didn’t know what to do with it now that they’d caught it.
Neither did he. Which is why he came to me.
Now, enrichment is simple. It’s taking data that you own and adding other data to it that you either also own or have rights to use, and making your data more valuable.
Quick example: You got 85% of the questions on my test correct. Nice job. Now let me add that the average person got 73% of the questions correct. Excellent job! Now I’ll tell you that based on all your test taking compared to average test takers, you should get 98% of the questions on the next test correct. I just sold you an hour of time back that you don’t have to study. How much is your time worth?
The good news is that this is the “fun” part. In all my consulting, founding, working, and advising, all the governance stuff I described in that last section creates the playground for all the cool things you can do.
But here’s the buzzkill part. Cool things don’t sell themselves.
This is where you and your data discover opportunities – how what you’re building solves the problem that perpetuates your business and revenue. You then decide which of these opportunities, they can be features, services, even granular functions, alerts, and random actions, provides the most value to your customers: Impact. And then you determine how much work, cost, and complexity is required to deliver that impact: Effort.
You score the impact and effort (ask Claude), put the opportunities on a quadrant chart (ask Claude), and then, based on the level of impact versus the level of effort, put them on a timeline, or a roadmap (ask Claude).
There. I just did product management. Two paragraphs.
By the way, you can start to see here how this flushes out a lot of product ideas that looked killer on paper. Like social networks, for example. None of the math I just did works for social networks until you make the customers the product and the advertisers the customers.
Then all that “YOU are the product” stuff isn’t so silly anymore.
You Need To Know Your Growth Vector
Because standing still in business is dying.
See? Buzz. Kill.
The reason why the entire SaaS sector is so spooked by AI is because they know that people like you and me can spend a weekend not only building something comparable to some of what they offer, but can now do it in a way that’s built on a stable foundation not unlike what they offer both their consumer and the enterprise customers.
Give me a weekend and I can build a Salesforce, a Hubspot, or a Facebook. And what’s more is I can build a version of those things that are built for the very disillusioned customer bases of those companies that have been turned off by the changes growth has brought to those ecosystems.
It’s a total Catch-22. You grow because you’re awesome. Then you’re no longer awesome because you’ve grown.
That’s the reason why those megacap SaaS companies haven’t taken their ball and gone home. They believe you’ll either die out after a little while or become just as enshitified as them.
So prove them wrong.
If building a business is hard, building a successful business is harder, and building a growing successful business is the hardest. The moment you become successful at one thing, you should already have been planning for the next. This is your growth vector, moving to the closest extension to what you were doing, and offering that new thing to the closest cohort of customers to the ones who love what you do the best.
The good news is, if you’ve built your architecture right, you can go back to that napkin reference again and start from step one. Although at this point it really shouldn’t be a napkin anymore. Napkins get wet.
This Isn’t Just An AI Story
You might notice that, except for an even heavier reliance on data, none of this stuff is really AI specific.
Shh. That’s the secret. As much as all of this business and product and growth stuff has changed dramatically over the last few years, none of this stuff has really changed all that much. The rules are still the rules. Value is still value. Data is still king. Access to the tools and the markets have just become available to way more people, including you.
And like I said at the beginning of each and every one of these weekend “guides,” the differentiator is still the “why.”
So spend the weekend figuring out why your customers need your product, why they’ll pay for it, and why your implementation of AI is going to be more than a bolt-on experiment, but rather a solution grounded in value, powered by AI, and built on a foundation of proprietary, enriched data.
Most entrepreneurs started out messing around with the “why” on the weekends. Please join my email list to get sensible tech and entrepreneurial guidance peppered with jokes and obscure references.
Credits: TCA, LLC.