The data should conduct each decision that a modern business takes. But most companies have a massive blind spot: they do not know what is happening in their visual data.
Coactive works to change this. The company, founded by Cody Coleman '13, Meng '15 and William Gaviria Rojas '13 has created a platform fed with artificial intelligence that can give meaning to data such as images, audio and video to unlock new information.
The coactive platform can instantly seek, organize and analyze unstructured visual content to help companies make faster and better decisions.
“In the first megadroned revolution, companies have improved to draw the value of their structured data,” explains Coleman, referring to the data of tables and spreadsheets. “But now, around 80 to 90% of the world's data is not structured. In the next chapter of the big data, companies will have to process data such as images, video and audio on a large scale, and AI is a key element in unlocking this capacity.”
Coactive is already working with several major media and retail companies to help them understand their visual content without counting on manual sorting and marking. This helps them get the right content for users faster, delete explicit content from their platforms and discover how specific content influences user behavior.
More broadly, the founders believe that the coactive serves as an example of how AI can allow humans to work more efficiently and solve new problems.
“The word coactive means working together simultaneously, and it is our great vision: helping humans and work -working machines,” explains Coleman. “We believe that vision is more important than ever because AI can separate us or bring us together. We want the coacse to be an agent that brings us together and gives human beings a new set of superpowers. ”
Give a vision of computers
Coleman met Gaviria Rojas in summer before their first year in the MIT Interphase Edge. The two would continue to specialize in electrical and computer engineering and work to bring it MIT OPENCOURSEWARE Content for Mexican universities, among other projects.
“It was an excellent example of entrepreneurship,” recalls Coleman of the OpenCoursware project. “It was truly empowering to be responsible for the company and the development of software. This led me to start my own small web development companies afterwards and follow the founder's journey (the MIT course). ”
Coleman first explored the power of AI to MIT while working as a graduate researcher with the Digital Learning Office (now MIT Open Learning), where he used automatic learning to study how humans learn on MITX, which hosts massive and open online courses created by MIT teachers and educators.
“It was really incredible for me that you could democratize this transformational route that I went through at MIT with digital learning – and that you could apply AI and automatic learning to create adaptive systems that not only help us to understand how humans learn, but also to offer more personalized learning experiences to people around the world,” explains Coleman de Mitx. “It was also the first time that I could explore video content and apply AI.”
After the MIT, Coleman went to the University of Stanford for his doctorate, where he worked on the reduction of barriers to the use of AI. Research led him to work with companies like Pinterest and Meta on AI and Automatic Learning Applications.
“This is where I could see at the corner of the future of what people wanted to do with AI and their content,” recalls Coleman. “I saw how the main companies used AI to stimulate commercial value, and this is where the initial spark of coactive comes from. I said to myself: “And if we create a business quality operating system for multimodal content and AI to make this easily? »»
Meanwhile, Gaviria Rojas moved to the Bay region in 2020 and started working as A scientific data at Ebay. As part of the move, he needed help transporting his sofa, and Coleman was the lucky friend he called.
“On the car trip, we realized that we both had seen an explosion occurring around the data and the AI,” explains Gaviria Rojas. “At the MIT, we obtained a seat at the forefront of the Megadone Revolution, and we saw people inventing technologies to unlock the value from this large -scale data. Cody and I realized that we had another powder on the verge of exploding with companies collecting a huge amount of data, but this time, there were multimodal data like images, a video, an audio and the text. missing to unlock it on a large scale.
The platform that the founders have built – what Coleman describes as an “AI operating system” – is an agnostic model, which means that the company can exchange AI systems under the hood while the models continue to improve. The coactive platform includes predefined applications that commercial customers can use to do things like looking for in their content, generate metadata and carry out analyzes to extract information.
“Before AI, computers would see the world through bytes, while humans would see the world through vision,” explains Coleman. “Now, with AI, machines can finally see the world like us, and that will make the digital and physical worlds blur.”
Improvement of the human-computer interface
Reuters' images database provides journalists from the world millions of photos. Before the coactive, the company relied on journalists manually entering tags with each photo so that the right images appear when journalists have looked for certain subjects.
“It was incredible slow and expensive to go through all these raw active ingredients, so people simply did not add labels,” explains Coleman. “This meant that when you looked for things, there were limited results even if relevant photos were in the database.”
From now on, when journalists on the Reuters website select “Activate research on AI”, Coactive can lay down the vigure content on the compreprense on the Detail Comparison System in each image and video.
“This considerably improves the quality of the results for journalists, which allows them to tell stories better and more precise than ever,” explains Coleman.
Reuters is not the only one to fight to manage all its content. The management of digital assets is a huge component of many media and retail companies, which often rely on metadata entered manually for sorting and research in this content.
Another coactive customer is Fandom, which is one of the world's largest information on television shows, video games and films with more than 300 million monthly active users. Fandom uses a coactive to understand visual data in their online communities and helps delete excessive gore and sexualized content.
“He was taking 24 to 48 hours for the Fandom to review each new content,” explains Coleman. “Now, with a coactive, they have codified their community guidelines and can generate more thin information on average around 500 milliseconds.”
With each use case, the founders see coactive as allowing a new paradigm in the way humans work with machines.
“Throughout the history of human-computer interaction, we had to look at a keyboard and a mouse to grasp information in a way that machines could understand,” explains Coleman. “Now, for the first time, we can simply speak naturally, we can share images and videos with AI, and it can understand this content. It is a fundamental change in the way we think of human-computer interactions. The central vision of coactive is because of this change, we need a new operating system and a new way of working with content and AI.”
