Nearly every site these days uses some kind of recommendation service to personalize their offerings, whether it's an online marketplace, a store, or a social media platform. Shaped, which announced an $8 million Series A funding round today, wants to make it easy for companies of all sizes to combine their existing data with large-scale language models and other recommendation systems to deliver personalized user experiences.
It's worth noting that Shaped is a developer-first platform, which gives customers a high level of flexibility, including choice of data sources, integration methods, language models (Llama, CLIP, BERT, etc.), and scoring mechanisms for recommendations and search results.
Image credit: Shaped
The company was founded by CEO Tully Murrell, CPO Daniel Camilleri, and Shai Bruhis (who has since moved on to other ventures). Murrell previously worked at Meta/Facebook, where he created the PyTorchVideo project and worked on other video-centric AI projects at Facebook AI Research. Camilleri previously worked at Uber and Afterpay. The two first met nearly 10 years ago while attending hackathons as students in Australia.
“How can we help small businesses build the same hyper-personalized experiences that you see from big tech companies? They should be able to do this without having to hire thousands of machine learning engineers,” Murrell explained when I asked about the team's initial vision for Shaped.
Shaped's current clients include brands such as outdoor travel marketplace Outdoorsy, Brazilian grocery delivery app Trela, and QVC, which uses Shaped for real-time video recommendations in its mobile app.
Even though Shaped launched before the generative AI boom, Murrell’s background in AI research meant that it was already leveraging many of these cutting-edge techniques.
Image credit: Shaped
Initially, Shaped was primarily focused on video personalization, but after going through the Y Combinator program in 2022, the team decided to branch out into other areas. [YC]”As we talked to our customers and looked at the market size, we realized, 'Video personalization isn't really what's important here. What if we sold a horizontal personalization platform that worked across all types of media: language, video, audio,'” Murrell said.
Currently, the service can ingest data from a variety of sources, ranging from data warehouses like Databricks, Google's BigQuery, and Snowflake, to PostgreSQL, MySQL, and MongoDB databases, and even services like Amplitude, Segment, Google Analytics, Mixpanel, etc. With all this data, and especially the large amount of user interaction data, developers can build their own custom recommendation systems.
Image credit: Shaped
The team emphasized its focus on developer experience: Shaped isn't necessarily aimed at marketers, but rather aims to give developers the tools and data to build and test these systems. And while the service is primarily code-first, it also includes a dashboard that lets you test your models and explore why the system makes certain recommendations.
More recently, Shaped has also moved deeper into search. “We have all this semantic understanding of the user and the content, so it was actually very easy to add a search package. That's actually a big focus for us over the next year, to become a complete search platform,” Murrell said.
Shaped's Series A round was led by Madrona Ventures, with participation from Y-Combinator and executives and founders from Clickhouse, Docusign, Okta, Rippling and StitchFix.
“We first met Shaped CEO and co-founder Tullie Murrell over a year ago and were incredibly impressed with his understanding of recommendation systems, having built and studied many such systems during his time at Meta,” Madrona's Karan Mehandru and Sabrina Wu wrote in a blog post today. “He has perfect founder-market fit, deep empathy for customers, and a nuanced understanding of the market and opportunity. We've also had the pleasure of getting to know Dan, Ben, and the rest of the Shaped team, who bring extensive experience from Uber, Apple, and other well-known companies.”