Seltz Raises $12.5M Seed to Build Search for AI Agents, Not Humans
Seltz has raised $12.5M in seed funding to build a web search engine designed specifically for AI agents, large language models (LLMs), and retrieval-augmented generation (RAG) systems instead of human users. The round was led by B Capital and Speedinvest, with participation from Future Present, Italian Founders Fund, Arc Investors, United Ventures, Vento Ventures, Mango Capital, 2100 Ventures, Future Back Ventures, and angel investors from Google, Ramp, Tako, and Hugging Face.
Founded by CEO Antonio Mallia, Seltz is developing a full-stack search platform that owns its crawler, index, retrieval models, and ranking systems. The company says this architecture delivers machine-ready web knowledge with lower latency and higher relevance for AI applications. The funding highlights a broader shift across AI infrastructure. As autonomous agents become more capable, traditional search engines built for people are increasingly being complemented by systems designed for machines that require structured, real-time information rather than lists of links.
What Happened
Search has spent decades answering people. AI agents ask different questions.
Seltz just raised $12.5M in seed funding, but the capital tells only part of the story. The bigger signal is where investors believe AI infrastructure is heading. Founder and CEO Antonio Mallia, CTO Max V., and Chief of Staff Matteo Pietra are building a search engine that doesn't care whether a webpage looks good to a person. It cares whether an AI agent can reason with it. Supporting that effort are Software Engineer Roberto Trani, PhD, Software Engineer Hunter Schuler, Software Engineer Al Johri, AI Engineer Lukas Gienapp, AI Researcher Elias Bassani, and Business Development Engineer Irine Kokilashvili, reflecting a team centered on information retrieval and AI systems.
The investment was led by B Capital and Speedinvest, represented by General Partner Gabe Greenbaum and Deeptech Partner Will Wells.
Why This Matters
Traditional search engines were built around human behavior. Type a question. Browse links. Decide what matters. AI agents don't browse. They retrieve, compare, summarize, verify, and act. That seemingly small difference changes the entire architecture behind search.
Instead of returning snippets and ranked pages, Seltz delivers structured documents with citations that AI systems can immediately consume. The company's proprietary crawler, search index, retrieval models, and ranking engine are all built in-house instead of relying on Google or Bing infrastructure. Infrastructure rarely earns headlines because it's usually invisible. Yet nearly every major software platform depends on someone solving foundational problems long before users notice.
Market Context
Antonio Mallia spent years working in information retrieval across Amazon, Bloomberg, and Pinecone before launching Seltz. That background matters because information retrieval has quietly become one of the most valuable layers in enterprise AI.
Large language models are impressive, but without reliable, current knowledge, they eventually hallucinate or operate on stale information. Retrieval is becoming the bridge between reasoning models and the live internet. Companies building AI agents increasingly need infrastructure capable of delivering fresh, trustworthy context in milliseconds, creating an entirely new infrastructure category.
Competitive Landscape
Seltz isn't competing to become another consumer search engine. It's competing inside the emerging AI-native search infrastructure market alongside companies focused on retrieval systems for autonomous software.
According to company benchmarks, Seltz reports 89% accuracy on its Dynamic News Search Benchmark while responding in under 250 milliseconds. The platform crawls 100s of millions of webpages daily, supports parallel queries, offers on-premises and private cloud deployments, maintains SOC 2 Type II compliance, and integrates with Dify, Zapier, and Orthogonal. If those performance claims continue to hold at scale, the conversation shifts away from search as a consumer product and toward search as foundational AI infrastructure.
What This Signals
This funding round isn't simply another venture announcement. It reflects where venture firms increasingly see long-term value. For years, AI investment gravitated toward applications. Now more capital is moving beneath the application layer into retrieval, orchestration, inference, security, and infrastructure.
The winners may not become household names. They may become the systems every AI product quietly depends on. History tends to reward companies building the roads long before traffic arrives.
The Bigger Industry Shift
AI agents are forcing developers to rethink assumptions that have existed since the early web. Human-first search optimized for clicks doesn't necessarily optimize for autonomous reasoning. That's creating space for companies like Seltz to rethink search from the ground up.
Whether Seltz ultimately defines this category remains to be seen. What is already clear is that AI infrastructure is becoming one of the fastest-evolving layers of the technology stack, and investors are increasingly treating retrieval as a strategic capability rather than a background service.
Frequently Asked Questions
What does Seltz do?
Seltz builds AI-native search infrastructure that delivers structured, real-time web knowledge for AI agents, LLMs, and RAG applications instead of traditional human-oriented search results.
How much funding did Seltz raise?
Seltz raised $12.5M in seed funding.
Who led the Seltz funding round?
The round was led by B Capital and Speedinvest, with participation from Future Present, Italian Founders Fund, Arc Investors, United Ventures, Vento Ventures, Mango Capital, 2100 Ventures, Future Back Ventures, and angel investors from Google, Ramp, Tako, and Hugging Face.
Who founded Seltz?
Seltz was founded by Antonio Mallia, who serves as CEO.
Why is AI-native search important?
AI agents require structured, current, machine-readable information rather than lists of webpages. AI-native search platforms are designed to provide that information with low latency and verifiable sources.
