As audiences flock to TikTok, Instagram Reels and YouTube Shorts, companies are struggling to maintain up with the demand for short-form video.
Munch, a startup that makes use of AI to show long-form video content material into short-form video clips for social media, introduced $7.2 million in seed funding on Tuesday.
Enterprise capital agency A* Capital is main the spherical, with participation from Liquid2 and pre-seed buyers Cardumen Capital and Reimagine Ventures.
Bake up a storm
Though Munch depends on some commercially obtainable know-how, together with for facial recognition and automated transcription, its important algorithm is proprietary and constructed principally in-house, mentioned CEO and Co-Founder Oren Kandel.
However, up to now a minimum of, Munch isn’t drawing on publicly obtainable massive language fashions as a result of the obtainable tech “isn’t correct sufficient and nonetheless doesn’t get the depth of the multimodality of video.”
Munch made a strategic choice to repurpose present long-form content material that somebody has already filmed, edited and produced into short-form video, quite than producing content material from nothing, Kandel mentioned.
Excessive-quality, human-made content material is “the place you get the true gems, the true engagement,” he mentioned. “We simply do the sprinkle on prime and make it extra accessible to new audiences on different platforms.”
As for the way the AI tech behind the Munch platform works, Kandel used the metaphor of a layer cake.
The underside layer is machine studying (ML) capabilities, which parse by what’s occurring inside a video and extracts hundreds of knowledge factors. For example, the ML instruments decide up on key phrases (celebrities, model names, places), audio cues indicating that individuals are enthusiastic about what they’re saying (talking sooner, laughing, interrupting one another) and different visuals (acquainted faces and scenes).
The following layer is what Kandel calls a “coherence engine.” The AI software crunches the movies to find out the very best size and whether or not the clips comprise sufficient context that they’re coherent as standalone movies.
The highest layer of Munch’s AI gobbles up knowledge from Google Traits and website positioning companies that reveals the types of subjects individuals are at present trying to find, Kandel mentioned. This AI-powered pattern analysis ensures that individuals’s factors of ardour intersect with what’s trending.
However nearly any long-form supply video holds many of the knowledge Munch wants. When you perceive the way to pull insights from that video, “you’re 90% of the way in which there,” Kandel mentioned. “The remaining is simply fine-tuning.”
Piece of cake
When shoppers begin a brand new mission both by way of the Munch cellular app or on desktop, they will select the platform they need to create clips for.
They then add a file or hyperlink to the long-form video, which Munch turns into 30 totally different clips, starting from 15 to 90 seconds. Purchasers have a selection of video facet ratio, overlays and elegance of mechanically generated captions.
Since Munch integrates with YouTube, TikTok and Instagram, shoppers can publish short-form movies on to these platforms. However content material isn’t printed mechanically, and although Munch suggests hashtags, they aren’t included in posts until the consumer approves them.
“It’s extra of a copilot, quite than an autopilot,” Kandel mentioned.
Munch extra
Headquartered in Tel Aviv, Munch launched roughly two years in the past and now has 12 workers working out of Israel, Kandel mentioned. Munch plans to spend its funding on rising the corporate by hiring extra builders, gross sales and advertising and marketing workers and investing in R&D.
The startup’s enterprise mannequin is 99% self-service. Purchasers can buy month-to-month or annual subscriptions, which vary from $49 for 200 minutes of video uploads monthly to $220 a month for 1,000 minutes of video uploads monthly.
Munch at present works with greater than 3,000 small companies, together with solopreneurs, and a handful of mid-market and bigger enterprise shoppers.
Businesses that concentrate on producing video content material for social platforms are amongst Munch’s shoppers, too. Though Kandel considers businesses to be Munch’s important competitor, he claims they will use Munch to “eliminate the chores,” aka the hours of repetitive human labor that Munch is automating.
Round 65% of Munch’s shoppers are within the US, however it additionally has shoppers in Western Europe, South America, Africa and Asia.