Exa Secures $17M to Develop AI-Native Search Engine

While numerous startups aim to replace Google with AI-powered search engines, Exa has a different approach: creating a Google for AI. Exa's founders believe that AI platforms, not humans, need a new kind of search engine. As AI becomes more prevalent in corporate and consumer life, these platforms must search the internet to find reliable information without hallucinations.

Exa's Unique Approach

Exa is developing a tool that allows AI models to perform web searches with an AI-native twist. Instead of relying on classic transformer-based language models, Exa uses a vector database and embeddings. The company invested a million dollars in GPUs to build a machine learning model trained to understand links rather than words and sentences.

"Transformers normally predict the next word. We train our search engine to predict the next link," says CEO Will Bryk. This novel search algorithm uses web links as a dataset, training the model to predict the most probable next link, minus the SEO spam and AI-generated content clogging traditional search engines.

Recent Funding and Vision

On Monday, Exa announced a $17 million Series A funding round led by Lightspeed's Guru Chahal, with participation from Nvidia's venture arm NVentures and Y Combinator. This brings Exa's total funding to $22 million, including a previous $5 million seed round.

"This is a very ambitious vision," says Chahal. "What Google is to humans, they are building for AI." Founded a year before ChatGPT's launch, Exa's initial goal was to use AI to build better search. However, after ChatGPT's success, AI companies began requesting an API version of Exa's search engine for integration into their models.

Exa's Evolution and Customer Base

Located in San Francisco's Cerebral Valley AI startup community, Exa has gained traction among AI companies. Its search engine is used by AI chatbots for internet lookup and companies for curating training data. Databricks, one of Exa's prominent customers, uses it to find large training sets for model training initiatives.

The API version of Exa's product, launched a year ago, has attracted thousands of developers. While Exa offers a free tier for limited use, it also has multiple fee levels. Although the founders did not disclose revenue figures, they indicated that revenue is increasing.

Exa's product is hosted on AWS, not Google Cloud, despite running its own GPU cluster. The team is not focused on being the startup that upends Google but acknowledges that AI search engines could challenge search giants if AI becomes central to the tech industry.