The race for web search just got a brand new chapter, and this time the ones calling the shots aren’t humans anymore. Funding, cutting-edge technology, and a completely different vision for how information should be delivered are at the heart of this story.
Seltz is one of the startups that decided to tackle this challenge head-on, betting that Google and traditional search engines simply weren’t built for what’s coming next.
And what’s coming are AI agents 🤖
These agents don’t work like you or me when we type a question into Google. They fire off long, precise queries, and some of these research agents send dozens or even hundreds of queries in parallel, looking for structured information ready to be processed by a machine.
The problem is that web search as it exists today was built to get humans to click on links, not to feed artificial intelligence systems with clean, contextualized data.
That’s exactly the gap Seltz wants to fill, and it just raised $12.5 million in a seed round to make it happen. 💡
The round was led by European venture capital firm Speedinvest and global investor B Capital, with participation from Italian Founders Fund, United Ventures, and Future Back Ventures, the investment arm of Bain & Company. It’s not every day you see a lineup like that backing such a specific thesis.
So what is Seltz actually building?
Seltz‘s pitch might sound simple at first glance, but it hides some seriously relevant technical complexity. The startup is developing a web search infrastructure built specifically to be consumed by AI agents, not by human eyes.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
According to Antonio Mallia, founder and CEO of Seltz, traditional search engines were designed for people who type short keyword-based queries and then scan through a ranked list of links. But AI agents work in a completely different way.
In his words during an interview with Fortune: the old search methods don’t work because they were architected for humans. Mallia explains that the information an AI agent actually needs isn’t in the snippet that shows up in search results. It’s in the body of the page, in things like tables, images, and other forms of representation that can be useful for an LLM or an agent.
That changes everything. When a traditional search engine returns results, it’s thinking about clicks, time spent on page, ads that need to be seen. But an artificial intelligence agent doesn’t click, doesn’t watch an ad, and doesn’t need a pretty interface to navigate through results.
What an AI agent needs is different: fast answers, well-structured data, enough context to make decisions, and the ability to process multiple sources at the same time without freezing up or losing information along the way. The infrastructure Seltz is developing prioritizes low latency, high precision in data retrieval, and an architecture that can handle the massive volume of requests that modern AI agents generate during a single complex task.
The differentiator: owning the entire search stack
Seltz’s biggest advantage, according to Mallia, is that the company controls the entire search stack: the web crawler, the search index, the retrieval models, and the ranking. In other words, they’re not just slapping a layer on top of another company’s search engine.
This matters more than it might seem. Many AI-powered search products are built on top of APIs from Google, Bing, or Brave. Some companies, like OpenAI and Perplexity, have also reportedly started building their own search indexes and using proprietary crawlers to collect information from the web. Still, Google is considered the owner of the best index, precisely because its massive scale, with billions of daily users, allows it to see far more of the web than any rival.
Seltz‘s system crawls hundreds of millions of pages per day and returns results in under 200 milliseconds. Instead of always delivering an entire page or a summary, it scores individual passages and extracts the specific table, text, or image the agent actually needs. Mallia calls this context engineering.
Who is Antonio Mallia, and why does he know this stuff
Seltz’s founder didn’t stumble into this market by accident. Mallia has been preparing for this moment throughout much of his career. His PhD in computer science at New York University focused specifically on information retrieval. After that, he worked as an applied scientist on Amazon’s artificial general intelligence team and as a research scientist at Pinecone, a company specializing in vector databases, before deciding to found Seltz.
He compares the current moment to the early 2000s, when Google’s PageRank turned the way search worked completely upside down. For Mallia, the revolution is back, and this time it’s driven by transformer models and AI workflows that increasingly handle search on their own.
Why does this funding matter so much?
The $12.5 million seed round funding might not seem astronomically large when you compare it to the billions some artificial intelligence companies are moving around out there. But context matters a lot here. A seed round at this level for a startup tackling an infrastructure problem, which historically is harder to monetize in the short term, says a lot about the confidence investors are placing in Seltz‘s vision.
Infrastructure is expensive, takes time to scale, and demands top-tier engineering, but when it works out, it tends to become practically irreplaceable within a tech ecosystem. And the timing couldn’t be better. The AI agent market is exploding, with companies like OpenAI, Anthropic, and Google launching or refining autonomous agent solutions that need to access external information to work well.
Another interesting angle is how this landscape exposes a vulnerability of the big labs. In December, Google sued SerpApi, a service that scraped Google’s own results and reportedly counted OpenAI among its clients, alleging the company circumvented its anti-bot protections. Meanwhile, Anthropic and Mistral have reportedly turned to Brave’s index to power web search in their chatbots. Mallia points to these arrangements as proof that even the biggest labs haven’t built genuinely independent web information retrieval, which leaves room for a purpose-built alternative.
A crowded playing field
There’s no pretending Seltz is alone on this journey. Far from it — the company is entering a competitive market with several better-funded rivals. 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 a third, called Tavily, was acquired by AI cloud company Nebius for up to $400 million earlier this year.
Even so, Mallia says he’s confident. For him, controlling the entire search stack and building a better product will ultimately be the deciding factor in this fight. It’s a bold bet, but it makes sense when you look at the company’s strategy of not depending on anyone else to deliver results.
A lean and global operation
Seltz is a pretty lightweight operation. Incorporated in the United States and founded last October, the company currently employs just 15 people, with only about half a dozen of them being full-time employees. The team works fully remote and is split between the San Francisco Bay Area and hubs near European universities in Pisa, Italy, and Leipzig, Germany.
Many team members hold PhDs in information retrieval, and Mallia recruited other veterans from Amazon’s AI efforts. Seltz’s advisors and angel investors include executives from Google, Ramp, Cohere, Synthesia, and Databricks, as well as academics from information retrieval labs at NYU and the University of Glasgow. That’s not exactly an ordinary resume.
According to the company, the funding raised will be used to keep developing the search stack, hire more people, and kick off an enterprise-focused sales effort.
Web search will never be the same
The arrival of AI agents as active consumers of information on the web is forcing a complete rethink of how we approach search and data retrieval. For decades, the web was built on a very clear premise: humans browse, humans read, humans click. The entire business model of major search engines, all of SEO logic, the whole content delivery architecture was designed around that premise. And now, all of a sudden, a new type of user shows up and breaks all those rules at once.
AI agents have no patience for slow pages, don’t care about pretty design, and aren’t going to click on an ad. They want the information, they want it now, and they want it in a format they can process directly. This creates a massive clash with the current web infrastructure, which simply wasn’t designed to handle this kind of demand at scale.
What’s happening right now with web search is a lot like what happened with data storage when the volume of information started exploding. New architectures, new players, and new approaches emerged that replaced solutions that once seemed rock-solid and permanent. Artificial intelligence, especially in the form of autonomous agents, is creating similar pressure on how information is indexed, retrieved, and delivered.
And startups like Seltz are betting that whoever solves this problem first will hold a central position in the next generation of web infrastructure. The battle is just getting started, and it’s going to be fascinating to watch who comes out on top. 🚀
