I built a real-time highlight detection Twitch bot for my roommate
+ my startup experience in college
Command Palette
Search for a command to run...
Search for a command to run...
+ my startup experience in college
I’m Akash Moorching (left) —co-founder of Saved.gg, I’m still a sophomore at the University of Maryland!
Last year, as a freshman at the University of Maryland (UMD), I spent a lot of time hanging out in my friend’s dorm. He was a Twitch streamer who occasionally posted clips on TikTok—but not nearly enough. After hundreds of hours of streaming, he was only getting a fraction of the viewers he deserved. I watched him manually edit short clips every time he wanted to share something new, and it struck me as absurdly inefficient.
At the same time, I joined an entrepreneurship club at UMD called Startup Shell, where I met Aayush Gupta. We had a funny, roundabout connection through mutual friends on Instagram (including his then-ex-girlfriend’s friend—don’t ask). We quickly realized we were both interested in tackling the same problem: simplifying the entire content creation process for live streamers. And so, Saved.gg was born—or at least, the seed for it was planted.
I never really saw myself as a founder. Honestly, I was a “big tech gunner,” laser-focused on the typical career path of trying to get those fancy internships. But seeing my roommate grind out those Twitch-to-TikTok edits by hand left me thinking,
“wtf, there’s got to be a better way.”
As someone who’d been following Twitch culture since COVID (and watching the likes of HasanAbi, JasonTheWeen, and other pop culture icons), I got sucked into the behind-the-scenes world of streaming. When you see how much time and energy streamers pour into their craft—only to face huge bottlenecks in content repurposing—you realize there’s a massive gap waiting to be filled.
Initially, though, it wasn’t all smooth sailing. The first two or three months felt like stumbling in the dark. Aayush and I started collaborating around late February 2024, but by spring break, we had to ask ourselves: “Is this truly worth our time?” There was a real tug-of-war between prioritizing classes (and GPAs— both cooked) and going all-in on Saved.gg.
By summer, we were four months into development and staring at zero traction. No paying users, no validation. We were grinding away at our MVP, trying to make it both functional and polished. And we kept questioning a universal startup dilemma: “Are we building the right thing for the right people?” Because even if you code up the most robust, beautiful product, it means nothing if no one wants it.
Through this grind, two big lessons emerged:
The 80/20 Rule – We realized we needed to spend 80% of our early effort figuring out what users actually wanted, so we’d only have to do 20% of the work later to fine-tune it.
“Ship Ship Ship” – No matter how tempted you are to perfect every detail, you have to get your product in front of real users, collect feedback, and iterate. Otherwise, you’re building in a vacuum.
On the practical side, our parents were covering tuition, so we could at least manage small infrastructure costs (like hosting, cloud storage, and workspace subscriptions) out of our own pockets. But the real sacrifice was time. We spent countless hours in the shared Startup Shell space—code-slinging, brainstorming, skipping classes, and basically living off the hustle.
Everything changed in late September of 2024 when we landed our first-ever sale: the University of Maryland’s esports team paid us around $70 a month for our platform. Sure, it wasn’t a huge sum, but it was a massive emotional win that proved we were onto something. It validated that short-form content creation wasn’t just a personal annoyance—it was a real pain point for streamers and editing teams.
Not long after, Boom.tv—a major tournament organizer in the esports space—also came on board, reinforcing our conviction that Saved.gg could serve both the B2C and B2B markets.
Initially, we were just a website offering AI-driven short clip generation. But we soon realized there was a better way to reach creators: a Twitch bot integrated directly into their channel.
Fiscally Smart – Switching to a custom transformer model for highlight detection drastically cut our processing costs.
Native Environment – Meeting streamers on their turf (Twitch chat) gave us an advantage; they didn’t need to leave the platform or learn a new workflow.
Uniqueness – The bot detects live highlights as they happen, a feature that’s practically unheard of among our competitors.
We made tons of mistakes. Here are two big ones:
Not Shipping Early Enough – Our initial instinct was to perfect the product. In reality, the best move is to launch a “good enough” version and improve it based on user feedback. Nobody cares if your UI is clunky if you solve a real problem effectively.
Mismanaging Hiring – At one point, our team ballooned to seven people—mostly other college students trying to help. But that added an unproductive layer of bureaucracy. We learned to hire (and fire) quickly, making sure each person is a net benefit, truly passionate about the problem, and able to keep pace.
As for criticism? We get loads. People have complained about everything from “shit highlight detection” to “ugly caption fonts.” But we love it because it helps us keep improving. If there’s one trait you absolutely need as a founder, it’s the ability to take feedback on the chin and use it to make your product better.
AI sits at the heart of Saved.gg. Every time LLMs advance or new AI breakthroughs come along, it means we can offer more powerful or efficient tools to our users. Plus, AI development helps reduce the learning curve. I’m far from the world’s most “cracked” developer, but being able to leverage machine learning frameworks and advanced language models has drastically boosted our ability to prototype, test, and deploy new features.
Over the past year, Saved.gg has become my life. There’s a thrilling rush in solving people’s real problems and hearing “This saved me hours of work!” from a satisfied streamer. Whenever I feel myself slacking or losing motivation, it’s usually because I’m not distributing and testing enough. That’s my cue to get back out there, talk to creators, and see what else they need.
Our overarching mission is simple: “To be an AI content repurposing tool for live-streamed events—one that automates short-form content creation without sacrificing quality.”
No journey is solo. Special thanks to:
Amogh (twitch.tv/amoghiehoagie), for inspiring me (ty pookie)
Aayush Singh, Eric Chen, Victor Suski, and Tariq Abdelkahlek, for believing in the vision, coding into the night, arranging our first TO contracts, and putting up with last-minute pivots.
In the next 5 to 10 years, I see Saved.gg becoming the go-to platform that streamers (and broader event organizers) trust for AI-powered content repurposing. We’re not just helping with clips; we’re transforming how people view, consume, and share live-streamed moments—whether in gaming, sports, or beyond.
If I could give any advice to my past self or other would-be founders, it’d be this:
99% of people have ideas. 1% can actually execute on them.
Keep shipping, keep iterating, and keep your users in the loop. That’s how you turn a dorm-room spark into a real, sustainable venture.
And if you read till here— use BLOG20 when buying a Saved.gg subscription for a discount!