Founder Snapshot
- Founder: Orestis Lykouropoulos, Co-Founder & CTO
- Company: AI Video
- Stage: Post-YC, $1.3M ARR
- Investors: Y Combinator
- Founded: 2023
- Previous: Amazon (4+ years), film music composer
Most founders can’t tell you the exact moment they decided to pivot. Orestis can tell you three.
The first pivot came when people stopped wanting to read personalized finance newsletters and started asking for an app. The second came when users skipped the reading altogether and just wanted audio or video. The third came when video editors started asking what tools they were using to make the videos in the first place.
“We bought the domain aivideo.com very early on for a good price because nobody had noticed yet,” Orestis says. “We were like, okay, if this doesn’t work, we’re just going to keep this in the background and turn our business into a video product.”
That was January 2024. By then, Orestis and his co-founder Justin had already been through Y Combinator with a different idea. They’d built a system that synthesized financial data into newsletters tailored to each recipient. It grew fast, but not fast enough. They tried to scale it to all of news. People wanted an app, not a newsletter. Then people wanted to watch, not read.
So they built a video automation system. No video models existed yet. Just generated images, titles, subtitles, background music, and LLM-written scripts. Some video editors saw what they were doing and asked if they could use it too. That question changed everything.
From Violin to Video Production
Orestis didn’t start out thinking about video. He started thinking about music. He picked up a violin at six years old and spent his childhood playing classical music. Computer science didn’t show up until college, where he double-majored in music and computer science before landing at Amazon for four years.
After Amazon, he took a break to work in film music. That’s where video production entered the picture.
“I realized that to write good music for film, you really have to think like a filmmaker,” he says. “You have to understand video and by extension video production as well, but especially video editing. You have to kind of get the beats and express them in music.”
That understanding of rhythm, timing, and storytelling became useful later. When he and Justin started hacking on AI projects in the early LLM days, Orestis had a sense for how music and video fit together. They didn’t plan to build a video company. They were just trying to figure out what LLMs could do.
The finance newsletter grew fast. Y Combinator accepted them. Then the pivots started stacking up. By the time they launched the Discord bot version of AI Video in January 2024, they’d already spent months building video automation infrastructure for a news product that wasn’t working.
“The first version was actually just a Discord bot that you could talk to and make videos,” Orestis says. “And since then we’ve built a basically fully featured video editor and we’ve turned the Discord bot into a bot that lives on the web.”
The Niche Strategy
AI Video doesn’t try to do every kind of video. They focus on niches they can understand deeply. Music videos came first. Musicians and small indie labels could make videos that used to cost $10,000 with a basic subscription. That grew fast.
“When we’re really, really targeted in a smaller sub segment of video production, we can understand that really, really well and actually provide the most value,” Orestis explains. “So that’s when we did that, that’s when we saw the most growth.”
The playbook works like this: pick a niche, collect the right information for that niche, feed it into the LLM agent, and let the system produce videos that understand the context. For music videos, they analyze the audio file to extract beats, sections, and lyrics. The agent uses all of that to make better videos than competitors who treat music as generic background noise.
“We feed all that into the LLM that runs the agent and then it can make a better video than our competitors can because it understands the music.”
They just launched product ads yesterday. Same infrastructure, different information collection. They want to understand brand values, color schemes, and tone. The video production doesn’t change, just what goes into it.
“I think the most exciting thing is basically that we think we found a playbook that’s repeatable and can be launched in a very similar way across different niches,” Orestis says. “Now we have basically the tech side enables that as well. We have the agent and the infrastructure that can basically make any video as long as you give it the right information.”
The competition question surprises them. There’s intense competition in the clip generation market, which they avoid. But for full video production, the field is less crowded than they expected.
“We’ve been kind of surprised that there’s not as much competition as we expected,” he says. “For video production, yeah, this has been great and I’m sure there’s going to be more competition, but we think we have the best product out there right now.”
Hiring for the Prompt-to-Production Era
Orestis has strong opinions about how engineers should work in 2026. The conversation around LLMs replacing engineers misses the point. They don’t replace engineers. They change how engineers work. And if you’re not willing to work that way, you won’t fit at AI Video.
“You can basically build full features without ever reading the code,” he says. “You can just prompt to production. I mean, obviously you’re going to test and everything, but they are writing really usable and maintainable code.”
He calls it being “AI native.” Three or four months ago, you could be skeptical about LLMs changing coding. Now you can’t. They’re too good. AI Video only wants engineers who embrace that shift.
“I only want engineers who want to work this way. I don’t think it’s going to replace engineers. It just changes how we work and if someone’s not willing to work in that paradigm, I think one, they’re not going to work in our company and I don’t think it’s going to be sustainable for them in general.”
The interview process reflects this. Traditional algorithm questions can be one-shotted with an LLM. Even if that still shows intelligence, it doesn’t prove you can work in the new paradigm. So they’ve switched to project-based interviews. Give the candidate a project, have them record their screen for an hour or two, and watch how they work with AI.
But technical capability isn’t enough. The team is small. Features often come directly from engineers who built a demo and asked “what do you think?” Many of those demos turn into production features.
“Very often an engineer will be like hey, I built this little demo, what do you think? And very often we’ll turn it into an actual feature that goes production,” Orestis says. “So we’re looking for people with product sense, very independent and yeah, an eye for design definitely helps.”
The team is remote-first, spread across San Francisco, LA, Columbia, India, and Louisiana. That widens the talent pool, which brings benefits and challenges. But it also enables their niche strategy. Each new vertical will be led by a different person, and so far they’ve only hired former founders for those roles.
“The internet has enabled people from anywhere to launch a small business and if we find a person that’s done really, really well bootstrapping something related to what we’re doing, we can tell them hey, you can keep your independence, you can stay where you are and you can just lead this product for us from where you are.”
Building Something You Care About
When asked why someone should join AI Video instead of any other startup, Orestis doesn’t oversell it.
“I mean, I think for many people maybe the right answer is they should join somebody else,” he says. “We’re looking for people who are specifically interested in video production, which is a pretty wide field.”
The current team reflects that filter. One engineer did video production for fun before joining. Another produces music. The connection to what they’re building matters.
“If you have that passion for what we do in addition to building that, that’s probably the best reason. We think we’re gonna own a good chunk of the video production market, basically.”
From finance newsletters to news apps to a Discord bot to a full video production platform doing $1.3M in annual revenue, the path wasn’t linear. But the pivots made sense. Each one moved closer to what users actually wanted. And Orestis’s background in music gave him a way to think about video production that turned into a technical advantage.
He still plays violin sometimes. The timing and rhythm that made him a good film composer now inform how AI Video thinks about editing, beats, and storytelling. It’s not the obvious path from classical violinist to YC founder to CTO of an AI video company. But it works.
AI Video is an AI-powered video production platform that automates video creation for music videos, product ads, and other verticals.
- Website: https://aivideo.com
- Connect with Orestis: https://www.linkedin.com/in/olykos/
