An AI tool can draft a course in an afternoon. Learners still stop in week two.
I review AI-drafted courses for creators and training teams. The draft usually has the content. What it's missing is the design that gets people to the end:
1. Outcomes a learner can do and show, not "understand"
2. Practice before the quiz, not only after the video
3. Feedback that tells the learner what to change next
4. A reason to come back for lesson four
I took a typical AI-generated course outline and redesigned it. The free sample shows the before, the after, and the reasoning behind every change.
Closing the books shouldn't take days of detective work.
The problem: Every month, finance teams spend days explaining why the numbers moved. They dig through invoices, POs and ledger lines to justify each variance against budget, then chase approvals and write board commentary. It's slow, manual and hard to audit.
The solution: Closeroom is an AI agent for month-end close. You ask in plain words, and it explains every variance, cites the source document behind each number, ties the totals out to the trial balance, and flags anything that doesn't add up for a human decision.
The impact: The work moves from investigation to review. Teams close faster, reviewers sign off with evidence in hand, and every number can be traced for audit.
AI Safety Training Video with an AI Presenter (Hurricane Demo)
This is a safety training demo we made with AI. A presenter walks through what a hurricane actually does while the studio around him slowly floods, with debris in the water and a car drifting past.
People remember training like this because they see the danger instead of reading a list about it. The same format works for any safety topic, and the script can change without a reshoot.
The calm presenter against a rising flood is the smart part. The contrast does the warning for you, no list needed. You mentioned the script can change without a reshoot: have you tried the same presenter in other languages? For safety training across a mixed workforce, that could be the biggest win.
Your AI-assisted product is supposed to deduct ₦50,000.
It deducts ₦500,000.
Who owns that mistake?
Would "AI built it" be an acceptable explanation to your customer?
I use AI in my design process.
For research, it helps me organize information and explore questions.
For design thinking, it helps me challenge assumptions and explore different directions.
For inspiration, it helps me get unstuck.
But there's something I won't outsource:
My judgment.
I still have to understand the problem, question the suggestions, consider the user, make the design decisions and validate the experience.
AI can help generate the work faster.
But the people who approve and ship a product must own what reaches the user.
AI can assist the process. It cannot replace the responsibility behind the product.
Founders building digital products: what's your human-review process before an AI-assisted feature reaches your customers?
I'm also open to remote UI/UX and Product Design opportunities and selected projects. If you're building something and need thoughtful design alongside AI-assisted execution, let's connect.