Seltz Raises $12.5 Million to Build Search Engine Optimized for AI Agents

Exclusive: Seltz, a startup trying to reinvent web search for AI agents, raises $12.5 seed round | Fortune

The rise of AI has rekindled the long-dormant search wars. Chatbots and AI agents need to surface timely, relevant information about news and all kinds of products and services. AI startup Seltz is among the players lining up to take on Google, betting that AI agents and chatbots demand a new kind of search engine. Today Seltz announced that it has raised $12.5 million in seed funding. The round was led by the European venture firm Speedinvest and the global investor B Capital, with participation from the Italian Founders Fund, United Ventures, and Future Back Ventures, the venture arm of Bain & Company. Seltz founder and CEO Antonio Mallia said traditional search engines were designed for people typing short keyword-based queries and then skimming a list of ranked links. AI agents work differently. They fire off long, precise queries -- some "research agents" send dozens or hundreds of queries in parallel -- and they need machine-ready information they can cite, not a snippet designed to entice a human to click through. "The old search methods don't work because they were architected for humans," Mallia told Fortune. "The information [the AI agent needs] is actually not in the snippet. It's in the body of the web page, it's in things like tables, images, and other forms of representation that can be useful for an LLM or for an agent." Mallia has been preparing for this moment for much of his career. His PhD. work in computer science at New York University focused on information retrieval, and he worked as an applied scientist on Amazon's artificial general intelligence team and as a research scientist at the vector-database company Pinecone before deciding to found Seltz. He told Fortune the current moment reminds him of the early 2000s, when Google's PageRank upended how search worked. "The revolution is back again," he said -- this time driven by transformer models and the AI workflows that increasingly do the searching themselves. What sets Seltz apart, Mallia argues, is that it owns the entire search stack -- web crawler, search index, retrieval models, and ranking -- rather than wrapping someone else's search engine. Many AI search products are built on top of Google, Bing, or Brave's APIs. Some AI companies, such as OpenAI and Perplexity, have also reportedly been working to build their own search indexes and use their own web crawlers to scrape information from the web to respond to queries that demand current information. Still, Google is considered to have the best search index because its massive scale, with billions of daily users, enables it to see far more of the web than any rival product. In December, Google sued SerpApi -- a service that scraped Google's results and counted OpenAI among its reported customers -- for allegedly circumventing its anti-bot protections. Anthropic and Mistral, meanwhile, have been reported to lean on Brave's index to power web search in their chatbots. Mallia points to such arrangements as evidence that even the largest labs have not built genuinely independent web retrieval, leaving room for a purpose-built alternative. Seltz's system crawls hundreds of millions of pages a day, and returns results in under 200 milliseconds. Instead of always handing back a full page or a summary, Mallia said, it scores individual passages and extracts the specific table, text, or image an agent actually needs -- an exercise in what he calls context engineering. The startup is entering a crowded field with a number of better-funded competitors. Parallel, founded by former Twitter CEO Parag Agrawal, recently raised $100 million at a $2 billion valuation; another AI native search company, Exa, pulled in $85 million; and another called Tavily was acquired by the AI-cloud company Nebius for up to $400 million earlier this year. But Mallia said he is confident that controlling the entire search stack and building a better product will prove decisive. Seltz, incorporated in the U.S. and founded last October, is a lean operation: it currently employs just 15 people, only a half dozen of whom are full time employees. The team, which works fully remote, is split between the San Francisco Bay Area and hubs near European universities in Pisa, Italy, and Leipzig, Germany. Mallia said many on the team hold Ph.D.s in information retrieval and he has recruited other veterans of Amazon's AI efforts. Seltz's advisers and angel investors include executives from Google, Ramp, Cohere, Synthesia, and Databricks, along with academics from information-retrieval labs at NYU and the University of Glasgow. Seltz said the funding it has raised will go towards continuing to develop its search stack, hiring, and the start of an enterprise sales effort.

Agentic infrastructure startup Seltz raises $12.5M to help AI agents search the web for answers

Agentic infrastructure startup Seltz raises $12.5M to help AI agents search the web for answers Agentic search startup Seltz Inc. said it has bagged $12.5 million in seed funding today to build a more optimal infrastructure so that artificial intelligence agents can find their way around the web. The round was led by Speedinvest and B Capital and saw participation from Italian Founders Fund, Future Back Ventures, futurepresent, Arc Investors, Vento Ventures, Mango Capital, 2100 Ventures and United Ventures, plus angel investors from Google LLC, Hugging Face Inc. and Ramp Network Inc. The startup was founded by its Chief Executive Officer Antonio Mallia, who told Fortune that he's not just trying to build another AI answer engine like Perplexity. Rather, he's looking lower down the stack, building search infrastructure that's optimized for AI algorithms that generate long and detailed queries and run them in parallel to surface structured evidence, rather than a list of links. "The old search methods don't work because they were architected for humans," Mallia told Fortune in an interview. He explained that the most useful information needed by AI agents often sits beyond where traditional search engines such as Google can reach. For instance, it might be inside the main body of text, or it might be embedded in tables, images, snippets or other page-level material that human-focused search engines cannot easily dig up. Mallia is an expert when it comes to search. His studies at the University of Pisa were focused on information retrieval, and he later earned a PhD in computer science at New York University before working as an applied scientist on Amazon.com Inc.'s artificial general intelligence team. He also worked as a research scientist at Pinecone Systems Inc., the creator of a specialized vector database that helps large language models to search through unstructured data. In a blog post in April, Mallia wrote that he realized the need for a specialized agentic search platform while working at Amazon as part of the team that developed Alexa's question answering engine. It was then that it first struck him that the consumer of search was no longer a human, but a machine using what it surfaces to inform its own answers. Unlike other search engines that often route their queries through Google Search, Bing or another major search provider, Seltz has built its entire search stack, including the crawlers, index, retrieval models and ranking systems. It's a big part of the company's plan, though he conceded in a blog post announcing today's round that web-scale search is "one of the most capital-intensive problems in software." He added that Seltz sought funding to scale its platform to "tens of billions of documents." Mallia said he started with a news index and was able to ship its platform within eight months of starting to build it. The company has also created its own Dynamic News Search benchmark, which reveals that it delivers 89% accuracy and returns its results in less than 250 milliseconds. It should be noted that those numbers are not an independent assessment, but Fortune said Seltz can crawl "hundreds of millions of pages a day," and generally returns its results in under 200 milliseconds. Mallia explained that Seltz's platform works by searching, scoring passages and extracting the exact text, table or image that an agent needs. Seltz is attacking a genuine problem, but it's not the only startup trying to do this and it has a distinct disadvantage in terms of the funding it has been able to attract. In April, Parallel Web Systems Inc., led by former Twitter CEO Parag Agrawal, raised $100 million in a Series B round that valued it at $2 billion, while Exa Labs Inc. nabbed $250 million just last month. The data center infrastructure giant Nebius Group N.V. is also making an agentic search play, having bought the Israeli startup Tavily Inc., which has developed a specialized search layer for autonomous AI agents, in February. Mallia's company is also much smaller than those rivals, with just 15 staff on its books at present, with only half of those working for it full time. However, many of its employees hold PhDs in information retrieval, and others previously worked with Mallia on Amazon's AI research teams. Seltz already has a foundational lab under contract and is running multiple pilots with companies building agentic workflows. The funding will be spent on engineering, hiring more staff and launching enterprise sales, Mallia said.