Sudip Rokaya (@sudiprokaya) and Kartikesh Mishra (@mk314k) have raised $4.8 million for Lamina Labs, their two-person San Francisco AI-video operation, Rokaya said in an X thread on Thursday. The founders also introduced Simi, a product that converts prompts, documents and scripts into narrated whiteboard-style explainer videos.
https://x.com/sudiprokaya/status/2082878413703196874?s=46 The financing backs a specific bet on how people will consume AI output: models and agents will need to present some answers as generated visual explanations instead of returning paragraphs of text. Lamina Labs is packaging that capability as both a web application and infrastructure that developers can call from their own software.
Rokaya grew up in Simikot, a remote village in northwestern Nepal, and named Simi after it. He studied computer science and mathematics at MIT and conducted machine-learning research at MIT's McGovern Institute before taking leave to build Lamina Labs. Mishra, also from Nepal, completed an MIT bachelor's degree in electrical engineering and computer science and mathematics in 2024, followed by a master's in engineering in 2025. The two met at MIT and began building together there, according to Y Combinator's profile of Lamina Labs. (ycombinator.com)
From a learning problem to video infrastructure
Rokaya began working on the idea after struggling to find an AI tool that could explain material from an MIT biology course through video, he told Nepali Times in June. An earlier attempt involved automating Manim, the Python animation library created by the educational YouTube channel 3Blue1Brown. Rokaya and Mishra abandoned that approach after concluding it would be difficult to make reliable enough for broad use. (nepalitimes.com)
That origin shaped Simi's format. Lamina Labs is concentrating on structured teaching videos rather than cinematic clips, synthetic presenters or general-purpose video generation. A user can upload a PDF, Word document, PowerPoint, Markdown file or plain text, or enter a prompt. Simi writes the script, produces and animates illustrations, adds narration and returns a finished video, according to Lamina Labs.
Rokaya said Thursday that Simi can generate a one-minute video in as little as 20 seconds and supports over 80 languages. Lamina Labs' current product page gives a broader benchmark of about 40 seconds for a one-minute explainer, indicating that the 20-second figure applies to its fastest runs rather than typical generation. Both figures are Lamina Labs' own performance claims. (laminalabs.ai)
Lamina Labs initially focused on education, where the founders had experienced the problem themselves. Selling into schools proved slow, pushing them toward small businesses, training teams, onboarding and product documentation. Nepali Times reported that Simi had attracted 4,000 users and 187 paying customers within 12 days of its initial launch, figures attributed to the founders. Lamina Labs has not published revenue or retention figures alongside Thursday's financing announcement. (nepalitimes.com)
Making video callable by software
The larger opportunity sits behind the consumer-facing generator. Lamina Labs has released an HTTP API, Python and Node SDKs, and an MCP server, allowing software and AI agents to request videos directly. The API can ground an explainer in a supplied document and currently supports jobs of up to five minutes through its standard endpoint. (laminalabs.ai)
That changes the economics Lamina Labs is pursuing. Training departments and software businesses typically treat each explainer as a production task involving scripts, storyboards, recording and editing. Lamina Labs wants videos to be generated whenever documentation, onboarding material or an AI answer changes. The API is the route to recurring, high-volume usage if Simi's output proves accurate and consistent enough to publish without extensive human review.
Lamina Labs currently charges $19.99 a month for 6,000 credits, $49.99 for 18,000 credits and $99.99 for 45,000 credits, with custom business plans. One credit covers roughly one second of balanced-quality video. API access is included with the Max and Business plans, while API generation consumes credits at three times the standard rate. (laminalabs.ai)
The founders are entering a crowded AI-video market that includes avatar platforms, generative-video labs, template editors and Google's NotebookLM video overviews. Lamina Labs is narrowing the job to explanation: sequencing information, matching narration to drawings and producing localized versions quickly. The focus gives Simi a clearer use case than a general video model, while leaving Lamina Labs exposed to larger model providers adding similar structured-output tools.
Y Combinator lists Lamina Labs in its Spring 2026 batch. Nepali Times reported in June that the founders had secured a further $3 million commitment after YC's backing. Rokaya's Thursday post gave a new total of $4.8 million without naming a lead investor, valuation or round structure. Lamina Labs plans to use the capital to expand beyond its founding pair and build the generation layer that sits beneath Simi. (ycombinator.com)