AI’s ability to reinvent web search remains murky, but the technology’s impact on everyday tools seems to be more promising. Case in point: A startup called Capsule has been putting AI to use in video editing software to improve the speed and efficiency of post-production edits. After launching its AI-powered editor into beta, the company closed on $4.75 million in seed funding to commercialize the product.
Longer term, Capsule says such technology could allow anyone to be creative with video, even if they aren’t professional video editors.
The company hasn’t always been into AI technology. Founded in 2020, Capsule emerged from the same team that built the animated GIF capture tool and social network Phhhoto, which eventually lost out to Instagram’s clone, Boomerang. After shutting down their app in 2017, they moved on to an experiential marketing business for live events called Hypno. But they soon had to pivot when the COVID-19 pandemic put an end to the need for Hypno’s in-person photo booths and other interactive experiences.
That led to the creation of Capsule, a platform that began as a way for brands to reach their communities in the post-COVID era using online Q&As and video stories. In 2021, the company raised $2 million in pre-seed funding for its collaborative video platform from Array Ventures, Bloomberg Beta, and various angels.
The company isn’t necessarily trying to use AI to take over the work of video editors. It says 90% of its revenue comes from the enterprise but, specifically, an underserved market of enterprise teams without video expertise that still need consistency in branding. Its video platform has been used at companies like Snowflake, TED, Salesforce, and The Wall Street Journal, among others.
More recently, Capsule began to explore how new AI models could improve its product.
Citing data from HubSpot, the company points out that short-form video is set to grow faster than any other format in 2023, and over 90% of marketers said they plan to either maintain or increase their investment in video creation. But the demand for video exceeds the supply of professional video editors, noted Capsule co-founder and CEO Champ Bennett.
“Despite the large number of video tools in the market, the needs of enterprise teams have largely been ignored,” he said in the company’s funding announcement. “Ask anyone in marketing, comms, sales, or success and they’ll tell you video outperforms all other formats, but they’ll also tell you they don’t get to use it often enough because of how costly and complex it is to make.”
To address these challenges, Capsule built AI Studio, which focuses on AI-driven, post-production video edits.
The company showed off a demo of the technology in December (see below), which uses AI and machine learning in a variety of models, including an ASR (automatic speech recognition) model for transcribing the video’s audio into text.
It also offers a diffusion model for generating B-roll images from the transcript as well as a generative LLM (large language model) that summarizes text from the transcript.
The AI Studio software runs in the browser, without requiring an app or extension to work, the company says.
After uploading a video to the platform, Capsule creates the transcript, which is placed off to the side of the video for use in edits. In the demo, the company showed how a user could select a block of text and then click a button to have the text automatically summarized and turned into a title card, using AI and its video markup language. There are different card styles available, including an animated title card page and a title card that appears below the video, each of which can be selected with a click.
It also showed off how you could select a block of text and then have AI generate an image automatically based on the subject matter identified in the highlighted text. Plus, you could click into the text prompt field and adjust the text for more precise control over the end results.
Another feature lets you select a line of text to have it appear as one of several available caption styles, like full-screen text, animated captions, or even a tweet-style caption.
“What we do for video is similar to what companies like Jasper are doing for copywriting or Replit is doing for coding,” Bennett tells TechCrunch. “We don’t own the models. Instead, we leverage the best foundational models to make video creators 10–100x more productive, while simultaneously lowering the barrier to entry so that teams in marketing, sales, success, and leadership can create compelling on-brand videos on their own.”
The edits themselves are powered by Capsule’s video scripting language, CapsuleScript, built over the past couple of years and designed to work in the browser. All of the AI model outputs are fed as inputs into CapsuleScript.
“Think what HTML/CSS are to websites, CapsuleScript is to video. It can render video dynamically, both at create-time and run-time –– allowing for the creation of personalized video at scale for the first time,” Bennett says. But he clarifies Capsule’s customers aren’t looking for a fully automated “one-click” solution, even though CapsuleScript would be capable of that.
“In reality what customers actually want is 80% automation, and 20% customization so they can tell a unique, creative story without a ton of friction,” he explains.
After the demo was posted, the company had to put access to AI Studio behind a waitlist due to demand, the company said.
With the additional funding, Capsule says it aims to make key hires across engineering, product design, and marketing teams to help it more quickly commercialize its AI Studio product. It’s looking to add a dozen full-time employees, including an ML engineer, front-end engineer, head of video and marketing, and product designer.
Investors backing the company in the new seed round that closed late January include Human Ventures, Swift Ventures, InVision founder Clark Valberg’s Tiferes Ventures, Behind Genius Ventures, plus its pre-seed investors Array Ventures and Bloomberg Beta.
Angel investors include Replit CEO Amjad Masad, Dropbox CTO Arash Ferdowsi, Figma head of sales Kyle Parrish, former head of audio & video at Spotify/Anchor founder Mike Mignano, Chorus.ai co-founder Roy Ranani, and Gumroad founder Sahil Lavingia.
The company declined to share its metrics related to revenues or the total number of customers but said it plans to announce the latter “soon.”
Including the new funding, the New York–based startup has raised $6.75 million since its founding.
Source @TechCrunch
