In a recent LinkedIn post, Kobi Omenaka discusses a significant challenge faced by many content creators: the difficulty of producing compelling visuals, particularly YouTube thumbnails, when lacking design expertise. Omenaka, who identifies as a builder rather than a designer, shares how he overcame this hurdle by integrating a new AI tool into his workflow.
Omenaka highlights the critical role of thumbnails in video performance, stating:
“Everyone judges a YouTube video by its cover. Terrible for me as my thumbnails were always…weak. I’m not a designer. Never have been.”
He elaborates on his own design limitations, humorously noting, “I can build an app, edit a podcast, set up a full marketing stack, … but ask me to make something that looks good and I’ll give you something that looks like it was made in PowerPoint… in 2014.” This personal struggle sets the stage for his discovery and implementation of Luma Labs’ Uni-1 API.
Automating Thumbnail Generation with AI
The core of Omenaka’s post details how he utilized the Uni-1 API, powered by Claude Code, to automate the creation of multiple thumbnail variations. He emphasizes the speed and ease of integration, noting that it took him mere seconds to wire the API into his existing stack.
YouTube’s native A/B testing feature for thumbnails was a key motivator for Omenaka. He explains the power of this feature:
“One thing I LOVE about YouTube is that it lets you upload THREE different thumbnails for the same video. It tests them all and tells you which one wins. The difference between three good options and one OK one can be the difference between failure and massive success for the exact same video.”
Previously, generating three distinct thumbnails would have been a time-consuming design ordeal. However, Omenaka reports that with the new tool, he can now generate three unique thumbnails in approximately 80 seconds after providing his video transcript.
Diverse Thumbnail Strategies
Omenaka describes the distinct approaches his AI-generated thumbnails take:
- One variation features his face in a typical “YouTube Face” reaction style.
- Another showcases the tool he built, demonstrating the innovation itself.
- The third option uses bold, prominent text, forgoing any facial imagery.
This variety allows the YouTube algorithm to naturally determine which visual style resonates most effectively with the target audience, a process Omenaka describes as letting “the algorithm fight and tell you which one wins.”
The Power of AI for Non-Designers
Omenaka underscores the significance of the Uni-1 API, particularly its ability to accurately replicate his likeness from a profile picture, a feature he has been using for a couple of weeks. He views this as a major breakthrough, especially for individuals who, like himself, struggle with design tasks.
“It’s wild that I gave Luma my LinkedIn Profile Picture a couple of weeks ago. It nails the likeness every single time. … I could never do “YouTube Face” myself 🤣. So that’s been a huge unlock for me!”
As Kobi Omenaka concludes, the potential of this technology is vast: “Anywhere code runs, you can pipe Uni-1 into your stack. If you can describe the output, you can build the tool.” This sentiment positions AI not just as a tool for automation, but as an enabler for creators to overcome personal limitations and achieve greater success on platforms like YouTube.
📝 About This Content
This article is based on insights shared by Kobi Omenaka on LinkedIn.
📅 Originally posted on May 5, 2026 | View original post on LinkedIn →