Exa raised $17 million from Lightspeed, Nvidia, and Y Combinator to construct a Google for AIs

Although the goal is to replace Google Starts using AI -led search (we are looking for you, confusing) is the beginning of -UP called EXA Different ideas: Ai Google.

EXA founders believe people are not urgent people who need a new type of search engine. On the contrary, since this was increasingly understood by the company and consumer life, the AI ​​platform itself has to experience the Internet regularly entering the Internet to search for information and return to hallucinations. And they can't enter a request on the keyboard. The EXA constructs a tool that allows the AI ​​model to perform a function similar to web search, but has a native distortion.

Co -Founder purchased a $ 1 million GPU (it was easier to get) and used a vector database and embedded (not the classic transformer LLM). Well -taught. Instead of speech and sentences. “Transformers normally predict the next word. We train our search engine to predict the next link,” Bryk says. "So people share links online, we use that data as a dataset for the models we train. We train the models to predict the next link. So this is a new search algorithm.

So, just as an LLM would complete a sentence with the most likely next word, the Exa system would complete a sentence with the most likely link (or ten), but perhaps minus the SEO spam and (ironically) AI-generated friends that clog every popular search engine these days.

According to an exclusive TechCrunch report, the startup announced a $17 million Series A round on Monday led by Lightspeed Guru Chahal, with participation from Nvidia's venture capital arm NVentures and Y Combinator. Exa has now raised a total of $22 million, including an earlier $5 million in seed funding. (Exa is in the summer 2021 YC cohort.) "It's a very ambitious vision," Chahal said. "Google is made for both humans and artificial intelligence."

The team was formed about a year before starting ChatGPT by two good friends who met during their first year at Harvard: CEO Will Brack (now 27) and co-founder Jeff Wang (26).

“We started before ChatGPT. Our original goal of our company is not at all. Yes: How are you using AI for better search? "Wang said. After Chatgpt attacked the technology industry, AI began asking Exas API version of its search engine, which can insert it into the model. Exa is located in San Francisco, which is part of the comfortable Brain Valley AI startup. In fact, as TechCrunch previously reported, a tweet by Wang went viral when he was looking for other companies that wanted to go in on an order of office nap pods and the response was overwhelming. (The work-nap-repeat culture is alive and well in this part of the tech industry.)

With AI companies now as its primary customers, use cases for Exa’s search engine span everything from an AI chatbot looking up info on the internet while answering customers’ questions to companies looking to curate training data. For example, DataBricks is an EXA Marquee customer using it to find large training kits for their model training initiatives, the founders say.

The API version was started about a year ago. "It's gotten a lot of attention since then," Wang said. Today, Exa says it serves thousands of developers, although it's worth pointing out that Exa has a free tier that allows anyone to try out its search engine in a limited way. It also has several tiered fee levels. The founders won't disclose revenue, other than to say they have some, and the numbers are growing. (Interestingly, in addition to running its own GPU clusters, Exa also hosts its products on AWS, not the AI-powered Google Cloud.)

The team isn't particularly focused on being a search startup that disrupts Google. Even if AI becomes everything the tech industry thinks about, AI-powered robotic search engines could pose an unexpected threat to search supremacy.