The latest articles on DEV Community by Connor Davis (@titlegen). https://dev.to/titlegen https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F3837276%2Ff5107a9a-551e-4529-be1d-1b05be3a4bb0.png https://dev.to/titlegen en Connor Davis Sat, 21 Mar 2026 16:02:15 +0000 https://dev.to/titlegen/why-i-chose-gemini-over-gpt-to-power-my-youtube-title-generator-and-what-i-built-around-it-3fb5 https://dev.to/titlegen/why-i-chose-gemini-over-gpt-to-power-my-youtube-title-generator-and-what-i-built-around-it-3fb5 <p>I built a YouTube title generator. Not because it seemed like a fun side project, but because I genuinely couldn't find one that worked for my own channel.</p> <p>I was running an automation channel and every tool I tried failed in the same way — they generated titles for the wrong creator. VidIQ was handing me MrBeast-style broad clickbait. ChatGPT output that was obviously ChatGPT. Neither understood niche. Neither understood my channel.</p> <p>So I built TitleGen. Here's what I actually decided to build and why.</p> <p><strong>Why Gemini, not GPT.</strong></p> <p>This was the first real decision. Most AI tools default to OpenAI. I went with Gemini as the primary model for one specific reason — it has a deeper contextual understanding of YouTube as a platform. The way it handles creator-specific language, niche terminology, and video metadata is noticeably better for this use case.<br> GPT sits as a fallback in case Google's API ever goes down. That's its only job. The tool isn't a Gemini wrapper though — the model is just one layer of it.</p> <p><strong>The Channel DNA Analysis layer</strong></p> <p>The part I'm most proud of is what happens before any title gets generated.</p> <p>TitleGen pulls and analyses a creator's existing channel — reviewing uploaded videos, tone, CTAs, links, and the trajectory of their last 10 uploads. It builds a profile of the channel before generating anything. A faceless gaming channel, a finance creator, and an automation builder should not get the same title structures. This layer is what makes the output feel niche-locked rather than generic.</p> <p><strong>How titles are structured and ranked</strong></p> <p>Once the channel profile is built, TitleGen generates multiple title variations using proven structural patterns — curiosity-driven framing, negative-stakes angles, authority positioning. Each variation is then ranked by virality potential so creators aren't just picking blindly.<br> Keywords entered by the user return titles that are genuinely locked to that niche — not broad suggestions that happen to contain the keyword.</p> <p><strong>Descriptions and tags</strong></p> <p>Descriptions are formatted using YouTube best practices so the algorithm understands the video properly. Optional keyword insertion, timestamps, and socials are all supported. Tags are contextual rather than keyword-stuffed — they're meant to support discoverability while still reading naturally.</p> <p><strong>Where it is</strong></p> <p>TitleGen has been live under a month. 81 users signed up so far. $9/month paid plan, free tier available.</p> <p>TitleGen is an AI YouTube title generator that creates niche-optimized titles aligned with how audiences search and click on YouTube. It performs Channel DNA Analysis to align the tone and channel theme.</p> <p>If you're a creator or building something in the creator tools space, I'd love to hear your thoughts!</p> <p>👉<a href="https://titlegen.ai/" rel="noopener noreferrer">https://titlegen.ai/</a></p> ai gemini llm showdev