Let me guess.
You have finally accepted that AI is not just hype. Your board is asking about it. Your competitors are talking about it. And now you are sitting there, maybe with a half-empty coffee, staring at a whiteboard that still says “AI Strategy” in big letters.
The question keeping you up at night? Where do I start with AI?
I have been in that chair. So has every senior leader I know who is not just pretending to have it figured out.
Turns out, the answer is both simpler and harder than you think. Simpler because the first move costs less than a bad hire. Harder because it means ignoring the flashy demos and taking a long, honest look at your own messy data.
Let me save you the next 18 months of recovering from a failed pilot. Here is where you actually begin.
What You Are Really Asking
Here is something the McLean Forrester whitepaper nails. When you search “where do I start with AI for my business,” you are not really asking for a tool list. You are admitting three things without realizing it.
One, you know AI matters. Two, you have no idea what to actually do about it. And three, you are exhausted by articles that say “start with strategy” but never explain how.
Sound about right?
Most AI advice is written for people who already have a team of PhDs on staff. The rest of us? We have got spreadsheets, legacy systems, and that nagging feeling that someone in marketing is already feeding customer data into a free ChatGPT account.
Be Honest: Where Are You Really?
Let me give you a quick gut check from the paper.
Most organizations are sitting at Stage 1, what they call “ad-hoc literacy.” That means people across your company are using AI tools. Some are paying for them personally. Some are not. Productivity gains are happening, but nobody is measuring them. And yes, data is leaking into consumer tools without anyone tracking it.
Sound familiar? Do not feel bad. That is just reality.
The real mistake is trying to jump straight from this controlled chaos to a custom AI model that runs your whole supply chain. That is like learning to swim by jumping off a cruise ship. What you actually need first is Stage 2: an enterprise license with proper controls, basic training, and a simple acceptable-use policy.
Boring? Yep. Absolutely necessary? Also yep.
Here Is Where I Push Back a Little
The paper cites that 95% of generative AI pilots fail to deliver measurable value. And look, that number is worth paying attention to. But I think we are measuring the wrong thing.
Here is what actually happens.
Your marketing writer finishes a first draft in 30 minutes instead of two hours. Great, right? Except then she spends that “saved” time on three more rounds of revisions. The final output looks exactly the same as before. Your metrics show zero improvement.
Does that mean the AI failed? Not really. It means the workflow did not change.
And this is where the paper gets really good. It says organizations that redesigned their work around AI were nearly three times more likely to see real business value. That is the hidden variable. Not the fancy model. Not the perfect data architecture. Just the messy, human process of changing how things actually get done.
Two Places to Go That Will Actually Help
Instead of giving you more generic advice, let me point you to two specific spots on the McLean Forrester website.
First, go look at their homepage. Find the AI Learning Path, Tier 1. It launched June 10. This is not theory. It is built for exactly where you are right now. And here is a detail I love: the CEO, Heather McLean, spent 20 years in the Air Force. That means she cares about practical results, not buzzwords. If you are tired of consultants who have never run a real operation, that is your signal.
Second, go back to the full whitepaper itself. There is a self-assessment in there, Stage 0 through Stage 5. Print it out. Pass it around the leadership team. Be honest about where you actually sit. That single exercise will save you from the most expensive mistake out there: trying to build an advanced solution when you have not even finished the basics.
The Three Levels, Plain and Simple
Let me translate the paper’s technical framework into something you can explain over lunch.
Level 1 is commercial AI. Think ChatGPT Enterprise, Microsoft Copilot, that whole category. You are looking at $20 to $60 per seat monthly, plus maybe $50,000 to $200,000 to roll it out properly. It will not transform your business overnight. But it will teach your people what AI is actually good at and where it lies. Skip this level, and you will be the CEO who approves a $2 million project without really understanding it.
Level 2 grounds AI in your data. Think customer service bots that actually know your products or internal search that finds the right contract. This is where things get interesting. It is also where they get expensive, roughly $75,000 to $400,000 for a single pilot, and messy. The paper admits openly that “RAG sounds clean in a slide deck and is messy in practice.” I love that honesty.
Level 3 is proprietary AI. These are custom models built on your secret sauce. We are talking millions of dollars. Most of you should not even think about this for years.
Here is the pattern I see over and over. Companies at Stage 1 try to build Level 3. Or they buy Level 2 without doing Level 1 first. Or they invest millions in technology without redesigning a single workflow. Do not be that company.
What to Actually Do This Quarter
Enough theory. Here is my opinion, shaped by the paper but pushed into action.
First, authorize an enterprise AI license within the next 90 days. Not because it is transformational. Because your people are already using consumer tools, and the risk of data leakage is higher than the cost of the subscription. Pick one vendor. Deploy it. Train everyone. Write a one-page policy. Done.
Second, codify the knowledge in your top performers’ heads. Your best account manager knows how to triage complaints. Your senior engineer knows which suppliers to escalate. Most companies have never written this down. AI forces you to do it. And that work pays off forever, with or without the technology.
Third, start where your data is cleanest, not where the headlines are. For manufacturers and financial firms, that is the back office. For retail and media, it might be customer-facing personalization. Ignore the generic playbooks. Look at your own spreadsheets instead.
The Risks Nobody Warns You About
Everyone talks about hallucinations. And yes, those happen. But that is not the main risk anymore.
Data leakage is the real problem. Every time an employee pastes customer information into a free AI tool, you are gambling with your reputation. Vendor lock-in is another sleeper. The model you build on today might be gone in 18 months. Quality drift means systems that worked last quarter suddenly behave differently this quarter.
The paper puts it perfectly: “The risks that get publicized are not the risks that most often realize.” Stop worrying about Skynet. Start worrying about your data governance.
Here Is the Bottom Line
You do not need a perfect strategy. You need basic literacy this quarter, a targeted pilot next quarter, and a real commitment to redesigning workflows, not just buying software.
The McLean Forrester whitepaper is worth your time because it is written for someone with a budget, a board, and a burning desire to stop feeling behind. It will not give you easy answers. It will give you honest ones.
And honestly? That is where every successful AI journey starts. Not with a grand vision. But with the courage to admit you are at Stage 1, buy the enterprise license, and do the unglamorous work of documenting how your business actually runs.
The AI revolution is not coming. It is already here, probably in the personal accounts your employees are not telling you about. Your only real choice is whether you lead it or just clean up the mess afterward.
Start smart. Start now. Head over to McLean Forrester and begin.
Top comments (0)