Were you unable to attend Transform 2022? Check out all of the summit sessions in our on-demand library now! Watch here.
Osmos, a company designing a platform to help enterprises adopt machine learning technologies, today emerged from stealth with $13 million in funding led by Lightspeed Venture Partners, with participation from CRV, Pear, and SV Angel. Cofounder and CEO Kirat Pandya says that the funds will be used to invest in Osmos’ product and grow its team, supporting R&D to advance the company’s services for data synthesis.
Companies have a growing number of data relationships, where they need to share data with customers, suppliers, partners, and others. The volume and variety of data are increasing, moreover, while the number of data sources and destinations ticks correspondingly upward. In 2016, IDG found that 7% of organizations are managing more than a petabyte of information across their data stores. And in a recent Forbes survey, 95% of companies cited the need to manage unstructured data as a problem for their business.
Osmos’ goal is to ease the analytics burden by eliminating data silos, allowing apps to talk directly to each. The company’s no-code platform provides users a way to bring in data from their systems or build pipelines to data, leveraging a real-time data transformation engine that automatically learns how to reconcile data.
Pandya and Naresh Venkat founded Osmos in late 2019 after a stint at Google Cloud, where they led AI and machine learning initiatives. Prior to Google, Pandya headed the Azure Cosmos DB rollout at Microsoft and worked on wireless mesh technology.
“During our time at Google, we noticed a few key trends,” Pandya told VentureBeat in an email Q&A. “Cross-company relationships are becoming more important than ever … [But] a lot of ‘automation’ companies [are] focused on connecting systems, automating workflows, and moving data within the four walls of an organization. When it [comes] to working with external data and systems … [t]here [are] no end-to-end solution to simplify data collaboration for the modern era … [M]ost solutions in the market that deal with data primarily target technical users — the non-technical frontline teams that actually deal with customers, partners, and vendors make do with Excel or nothing.”
Transforming data with AI
Unlike most AI-based systems, Osmos’ technology doesn’t require huge amounts of training data from customers, Pandya says. Companies only need to provide a couple of example rows of what the output should look like, and then Cosmos generates a debuggable data transformation program. This explainability is a key differentiator, Pandya believes, because it confers confidence that the system will do the right thing rather than make guesses based on arbitrary confidence values.
“Using this technology, Osmos is able to automatically learn complex transforms including conditionals, complex multi-column joins, and splits. It also ensures that the data that’s uploaded is not just schematically valid, but also passes any business rules [a users defines],” Pandya explained. “More importantly, we have turned [the platform] into a seamless full-feedback loop. [Users train] the system to learn transformations and clean up data. New data shows up, and if something breaks, the system notifies the relevant users. The users fix the errors with more examples and formulas, and the system relearns the transformation logic to handle such exceptions in the future.”
Osmos counts Bluecore, Mosaic, and Blissfully among its customers. One brand, an ecommerce company, is using the platform to automate the ingestion of catalog data from multiple distributors and vendors. Another client, a hospitality brand, is leveraging Osmos to bring property listings into its operational systems.
“Nowhere is this data silo problem more painful than when onboarding customers. One of the biggest challenges is populating your product with your customer’s live data as quickly as possible, so that they can start using it, have that ‘aha’ moment and see the value,” Pandya said. “Companies are moving more infrastructure to the cloud and investing in more efficient systems to make their teams more productive, and Osmos helps streamline one of the most manual and inefficient processes for engineering teams — dealing with external data on an ongoing basis. We are excited about the breadth of customers we have across multiple industries including manufacturing, ad tech, martech, analytics, IT service management, logistics, and supply chain.”
Osmos’ team currently stands at 15 employees. Pandya expects that number to double by the end of the year, with growth focused on the company’s engineering and go-to-market teams.
VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Learn more about membership.