When ultimate founders Sammy Sidhu and Jay Cheer were working as software engineers in Lyft’s self-driving car program, they witnessed issues with brewing data infrastructure.
Self-driving cars generate a large amount of unstructured data, from 3D scans and photos to text and audio. There was no tool for LYFT engineers that could understand and process all of these different types of data at the same time. This allowed engineers to piece together open source tools in a long process with reliability issues.
“We got all these great PhDs and had great people from the entire industry, we’re working on self-driving cars, and they spend 80% of our time working on infrastructure rather than building core applications.” “And most of these issues they were facing were about data infrastructure.”
Sidhu and Chia helped to build an internal multimodal data processing tool for LYFT. When Sidhu tried to apply for another job, he discovers that the interviewer continues to ask him about the possibility of building the same data solution for his company, and the idea behind the end was born.
Ultimately, we built an open source data processing engine for Python-Native, known as DAFT. It is designed to work quickly on a variety of modalities, from text to audio and video. Sidhu said that SQL is about converting Daft as an unstructured data infrastructure, similar to tabular data sets in the past.
The company was founded in early 2022, nearly a year before ChatGPT was released, before many people realized this data infrastructure gap. They launched their first open source version of the DAFT version in 2022, and are preparing to launch an enterprise product in the third quarter.
“The explosion of ChatGpt, what we saw is a lot of other people building AI applications with different kinds of modalities,” says Sidhu. “We then started out like everyone was using images, documents, videos, etc. in their applications. [increase] Dramatically. ”
The first idea behind Daft’s buildings, which originated in the self-driving car space, is home to many other industries that process multimodal data, such as robotics, retail technology, and healthcare. The company currently counts Amazon, Cloudkitchens and together AI as its customers.
Finally, I recently raised two rounds of funding within eight months. The first was a $7.5 million seed round led by CRV. Most recently, they raised a $20 million Series A round led by Felicis with participation from Microsoft’s M12 and Citi.
This latest round will be directed towards creating commercial products that will not only enhance Eventual’s open source offering, but also enable customers to build AI applications from this processed data.
Felicis general partner Astasia Myers told TechCrunch that it finally discovered through a market mapping exercise that involved searching for a data infrastructure that could support the rise in multimodal AI models.
Myers said he ultimately stood out because he was the first mover of the space – this could be even more crowded – and based on the fact that the founders addressed this data processing problem directly. She added that the final problem is also solving growing problems.
The multimodal AI industry is projected to grow at a combined annual growth rate of 35% between 2023 and 2028, according to management consulting firm Marketsandmarkets.
“In the last 20 years, annual data generation has risen by 1,000 times, with 90% of the world’s data being generated over the last two years, and according to IDC, the majority of data is unstructured,” Myers said. “It fits this huge macro trend of generation AI built around text, images, video and audio. It requires a multimodal native data processing engine.”
