Lingelbach: Media AI apps cannot yet rely purely on foundation models
Michael Lingelbach · His Video AI app hit $10M+ ARR in Months—with 0 outbound sales. | Michael Lingelbach, Founder of ... · PMF Show · Aug 18, 2025 · at 4:00
Hedra founder Michael Lingelbach compares the maturity of text LLMs to visual media generation models.
0:00 / 0:26exact quote · 26.6s
▶ Watch the full episode on YouTube → 720p mp4 · rendered on demand · StarZero watermark
“LLMs are just, like, much more farther along, and they have, like, a very well-defined, like, path, and it's really just about scaling them at this point, and, like, You know, working on reasoning, improving general intelligence. I think for, like, media generation, it's still too early for that. and, like, we're seeing things like fundamental architecture changes happening even now. so that's why I'd say, like, we're not in the stage where we can just build on top of a general purpose media generation model just yet.”
quote is from the automated transcript, cleaned for reading: filler sounds and stutters are removed, nothing is rephrased. names can be misheard (the analysis reads context, assessments check outside sources). how →
More from Michael Lingelbach
Lingelbach: AI models will be sufficient for most use cases in six months
“I think in six months the capabilities of these models are going to be good enough for, like, most of the use cases that you'll need, and, like, a lot of it is going to be, can you build vertical specific workflows that, like, solve, like, specific pain points…”
Lingelbach: Hedra's models support infinite recursion for long video generation
“One of the advantages of our technology and our architecture is our models support infinite recursion so we don't have an issue with long video generation like other providers.”
Lingelbach: Free users often provide negative product signal to startups
“Free users are often, like, not, they're often negative signal, in my opinion. People, the people who are willing to pay for software are sometimes a totally different group than the people who are willing to use it for free.”
Lingelbach: Clever academic AI techniques were ultimately ineffective against scale
“A lot of the clever approaches were, that we were doing in academia were ultimately, like, nothing in the face of scale, and that, like, if I wanted to build something impactful, I felt like I needed to do it in, like, an industry setting at that time.”