Lamina Labs raises $4.8M to make AI explain with video, not just text
Lamina Labs raised $4.8 million led by Foundation Capital to build video infrastructure that lets AI models generate explanations as narrated video rather than paragraphs of text. Its first product, Simi, turns a single prompt into a one-minute narrated explainer in as little as eight seconds across more than 80 languages.
Context from: Laminalabs | Ycombinator
The decision it puts on your desk
companies building educational products, documentation platforms, or customer-facing AI agents should evaluate video-as-output infrastructure now rather than waiting for it to become a commodity. Request early access to Simi at laminalabs.ai and test whether a narrated explanation converts better than a text wall for your highest-value user question this month.
Knowledge is not the same as understanding.
That sentence, from Lamina Labs' launch announcement Monday, is the thesis behind one of the more unusual AI infrastructure bets of the year.
Lamina Labs raised $4.8 million led by Foundation Capital, with participation from Link Ventures and Y Combinator. The founders, Kartikesh and his co-founder, are the first all-Nepalese founding team in Y Combinator history. They met at MIT. Neither grew up knowing what MIT, Y Combinator, or venture capital were.
The company's first product is Simi, and it works like ChatGPT except the answer is a video. Give it a single prompt, document, or script, and it generates a complete one-minute narrated explainer in as little as eight seconds in more than 80 languages.
Lamina is not building another cinematic AI video tool. The company is building fast, precise, and controllable video infrastructure for LLMs. The goal is to let any model, agent, or application generate video at the moment an explanation is needed.
The use cases the founders describe are not about making videos that look good. They are about making explanations that teach:
A student should be able to turn a difficult question into a visual lesson. A company should be able to turn documentation into a walkthrough. An AI agent should be able to show how it reached a conclusion instead of returning another wall of text.
The timing is worth paying attention to. LLMs changed access to knowledge. Anyone can ask a question and receive an answer instantly. But the interface for that answer has been almost entirely text, not because text is always the best way to teach, but because text has been the easiest medium for computers to generate.
Lamina is betting that changes, and that the companies that build the infrastructure for AI-generated video explanations now will own a layer of the stack that becomes obvious in hindsight.
I think the ambition here is bigger than it reads on a first pass. The company is not selling a video generation tool. It is building infrastructure that treats video as a native output format for AI, the same way text is today. If that happens, the explainer video market is the visible beachhead. The real target is everything downstream of every AI response that would be better understood as a visual explanation.