Kosi Uzodinma · Lead Data Architect, General Mills
I teach people how to think about AI so they can build with it themselves. No technical background required.
I am a Lead Data Architect at General Mills and an AI educator. I have spent years making AI and data tools work inside a large, complex organization. What I learned is simple: the tools are not the problem. People do not know how to think about them. That is what I teach.
AI feels like the internet did in the late nineties — a once-in-a-generation shift where most people are not sure where they fit in what is coming. Some think it will take everything. Others think it is overhyped. Most are somewhere in the middle: curious, a little overwhelmed, not sure how to start. I built my work for those people.
Teaching has been true of me long before AI was the subject. I spent two years as a Young Adult Pastor, preaching and leading discussions with people at very different starting points. At General Mills, I built and ran a two-year AI and data curriculum for 50+ engineers and scientists. I know how to meet people where they are. That is the work.
Outside of work, I build for real clients. A real estate investor writing letters of intent manually for every deal now generates them in under two minutes. A marriage counselor without a development budget built her entire practice website without a developer and owns it outright. I build for myself too: a personal operating system that remembers my financial goals, training protocol, and the decisions I have already made, so I can show up and work without re-explaining my life. I have tried most of the tools that exist, not to collect them, but to understand what each one is actually designed to do. That is the work that makes me a useful guide, not a theorist.
The Clarity Method
The underlying models are impressive. That is not the gap. The gap is between what a model can do and what the application built around it was designed to do. Every company building on top of AI is solving a different problem. If you do not know which problem a tool was built for, you will keep reaching for the wrong one.
Think of it like trades. A handyman can do a lot: hang a door, patch drywall, fix a cabinet. But there comes a job where they hit a wall. Rewiring a panel. Running new plumbing. Not because the job is too hard. Because it requires a specialist. You would not call a handyman to wire a new circuit. You would call an electrician. AI works the same way. There are tools built for conversation. Tools built to write and run code. Tools built to retrieve from your documents without losing context. Tools built to generate images. Each one was designed with a specific job in mind. The skill is knowing which one to call.
AI does not know what you need. You have to. Most people open a tool and start typing without knowing what they are asking for. They get something that sounds right but does not help. Your industry knowledge, your experience, your judgment: that is the raw material these tools run on. AI multiplies what you bring. The people who get results are not smarter. They are more specific. Getting clear about what you want, before you open any tool, is the skill that changes everything.
“The tools will do almost anything. Knowing what to ask for is the whole job.”
Bring this framework to your audience.
Book a Speaking InquiryEach of these started with a real problem. The build is the lesson: what the right tool was, why it was the right call, and what you can take from it.
A real estate investor client was hand-writing letters of intent for every property offer. Same format. Different numbers. Every single time.
A tool he describes a deal to in plain language and receives a complete, properly formatted letter of intent in under two minutes. He kept the judgment. I removed the friction.
The friction was not in the decision. It was in the document. Identifying that distinction is the whole job.
A marriage counselor client was building her practice with her own curriculum and methodology. She needed a client-facing website but did not want recurring platform fees or a developer she would have to call every time something needed to change.
A full website where she can present her services, share her approach, and collect client inquiries. Built entirely in Lovable. She owns it outright, saving over $1,000 in development costs and replacing a Google Docs intake process with a professional client-facing site. No platform dependency, no ongoing fees, no developer required.
The goal was not a website. The goal was independence. Knowing which tool gives you that matters as much as knowing how to use it.
Every major area of my life was siloed: career, finances, training, health, nutrition, relationships, business. I had tools for some of it and notes scattered across the rest. Nothing held the full picture, and nothing connected the dots between them. Every time I sat down to think through something real, I was starting from scratch.
A Personal OS in Claude Cowork with nine workstations covering every area of my life. Each one has its own memory and rules. A master constitution governs how they interact when a decision touches more than one area. It knows my salary, my financial philosophy, my training protocol, and the calls I have already made. When I was deciding whether to sell my rental property, it already held the financial context. The answer was clear: I was profiting and did not know it. I open it and pick up exactly where I left off.
Without context, you cannot get the right answer. You are guessing, or the tool is making assumptions based on generalizations that have nothing to do with your actual situation. The system works because it holds the context. Walk in with the right context and the right answer is usually obvious.
Reading Scripture without historical and cultural context means missing most of what is actually there. Ten open browser tabs later, you still are not sure what to trust.
A personal Bible study companion that gives historical context, original audience perspective, word studies, and cross-references for any verse. Not a replacement for study. A research assistant that handles retrieval so I can focus on the real work: reflection.
AI is not a rabbi. It is a research assistant. Knowing the difference before you start is the whole point.
Most podcast episodes are not worth 40 minutes to find out if they are worth 40 minutes. The queue piles up and the guilt sets in.
A Gemini gem that extracts main points and creates analogies for each takeaway from any YouTube link. Gemini was the deliberate choice: YouTube is Google, and the integration is architecturally tighter than any workaround.
Infrastructure matters. The right tool often is the one with the least resistance.
Every gym session was tracked in a notes app or not tracked at all. No history, no progression data, no way to look back and see if the work was actually working.
A native iOS gym logging app built with Xcode and Codex. Log workouts, track sets and reps, and build a personal training history that persists. The data belongs to the athlete, not a subscription.
The best tracking system is the one you keep using.
I write to teach you how to think about AI: which problems are worth solving, which tools are built for them, and why the build worked. Published on LinkedIn.
A real estate investor was drafting Letters of Intent by hand for every offer. The task was simple. He just had not built the system to do it for him. That is what we built.
Read on LinkedIn →My son had read every book we owned. The library had gaps. So I built a tool that generates illustrated stories where he is the main character. In under five minutes, every night.
Read on LinkedIn →Tracking nutrition works until the friction kills the habit. I stopped weighing food and started talking to my tracker instead. Here is what I built and why it actually stuck.
Read on LinkedIn →Context matters. Culture matters. I built a personal Bible study companion that gives me historical context, original audience perspective, and word studies, so I can do the real work: reflection.
Read on LinkedIn →Read by 237+ professionals on LinkedIn.
Subscribe on LinkedIn4.5 / 5
average rating across all training programs
“You did a great job and helped demonstrate what adoption looks like at your company. I can tell the participants enjoyed it.”
“You are great at moving people towards adoption and communicating ideas.”
— Senior Manager of Data Architecture, General Mills
“I never had a curriculum or clear pathway to understand what I should know as a data engineer. This changed that.”
— Data Engineer, General Mills
“This is exactly what we need to start automating our business to free us up to do more high value work.”
— Real Estate Investor
“I had an idea and didn't know it was possible and after talking to Kosi about my needs he created a platform that is perfect for my practice.”
— Marriage Counselor
I give audiences a framework they can put to work the next day. I speak at conferences, professional associations, masterminds, schools, and universities. If your audience includes business owners, leaders, or professionals who know AI matters and have not found a practical path into it yet, that is who I am built for.
The Two Skills That Separate AI-Effective Professionals from Everyone Else
Most professionals who struggle with AI blame the tools. The tools are fine. This talk identifies the two skills that determine whether AI works for you: knowing which tool is designed for what, and being specific enough about what you want that the tool can deliver it. Audiences leave with a framework they can apply the next day.
45–60 min
A Practical AI Workshop for Business Owners and Teams
A hands-on session for business owners and professionals who have tried AI, gotten mediocre results, and are not sure why. We map the AI ecosystem by what each tool is designed to do, identify where your current workflow has the most friction, and leave with a specific tool tied to a specific problem. No technical background required.
90 min – half day
An Honest Conversation for Leaders Who Need to Make Decisions
A direct conversation about the current state of AI tools: what they are genuinely good at, where they fail, and how leaders should think about bringing them into their organizations. Designed for decision-makers who need to act on AI without becoming technical experts.
30–45 min
What You Need to Know Before You Graduate
A session for students on how to think clearly about AI as a skill set, not a threat. Covers the AI tool ecosystem, the clarity required to use these tools effectively, and what it actually looks like to build with AI in the real world. Designed for any major, any background.
45–60 min
| Event | Organizer | Date | Format |
|---|---|---|---|
| DevColor Member Led Professional Development Workshop | dev/color | October 29, 2026 | Virtual |