Seltz Secures $12.5 Million to Build the Web Search Layer for AI Agents

🔄 Update — 25 June 2026: The Rise of Agentic AI, Computer Use, and Physical AI

The ecosystem of autonomous AI agents is expanding rapidly with Google introducing “computer use” capabilities in Gemini 3.5 Flash, allowing agents to navigate screens directly. At the same time, Nvidia is pushing agentic AI into drug discovery and physical systems, while MIT researchers have unveiled new methods to dramatically improve agent speed and energy efficiency.

What’s new?

Computer Use Capabilities: Google’s Gemini 3.5 Flash introduces native computer use, allowing agents to interact with computer interfaces just like human users.
Nvidia & Physical AI: Nvidia CEO Jensen Huang declares agentic AI as the current reality and points to physical AI as the next growth wave, expanding applications into areas like drug discovery.
Hardware & Software Efficiency: MIT researchers have developed new architectures to tackle the latency and energy consumption bottlenecks inherent in multi-agent workflows.

Why this adds to the article

These advancements highlight the critical need for a low-latency, machine-readable data infrastructure like Seltz’s Web Knowledge API. As agents take on complex, real-world tasks—from drug discovery to operating desktop computers—eliminating search latency and data overhead becomes paramount.

Summary

The startup Seltz has announced a $12.5 million Seed funding round to redesign web search from the ground up for artificial intelligence systems. Co-led by Speedinvest and B Capital, Seltz is building a Web Knowledge API designed specifically for machine consumption, delivering structured real-time data to autonomous agents and LLMs with ultra-low latency. The company aims to eliminate the inefficiencies of traditional, human-oriented search engines and significantly boost the performance of autonomous workflows.

What happened?

Funding Round: Seltz raised $12.5 million in a Seed round co-led by Speedinvest and B Capital, with participation from Italian Founders Fund, United Ventures, 2100 Ventures, Future Back Ventures, and arc Investors.
Founders & Team: Founded in 2025, the startup is led by CEO Antonio Mallia (formerly at Amazon, Pinecone, and Bloomberg) and Elias Bassani. Both co-founders possess deep expertise in information retrieval.
Product Launches: Seltz unveiled version 1 of its API at VivaTech 2026 and launched the Dynamic News Search Benchmark (DNSB), a public tool to compare retrieval quality and latency across different AI search providers.
Use of Funds: The capital will be used to expand Seltz’s crawling and indexing infrastructure, scale its research and engineering teams, and accelerate go-to-market efforts.

Why it matters

Traditional search engines like Google or Bing are designed for humans to navigate links, advertisements, and complex layouts. For AI agents, parsing these HTML-heavy pages introduces high latency and fills the context window with irrelevant information. Seltz resolves this with a Web Knowledge API built in Rust that delivers clean, structured, and metadata-enriched data. With response times under 200 milliseconds, it removes a major bottleneck for multi-step agentic workflows.

Evidence

Funding Details: Coverage from FinSMEs and Startup Business confirmed the funding amount and the lead venture capital firms.
Technical Launch: Documentation regarding API v1 and the Dynamic News Search Benchmark (DNSB) at VivaTech 2026 highlights the focus on latency and machine-readable data.
Founder Background: The executive track records of Antonio Mallia and Elias Bassani validate the company’s technical capabilities in search and database systems.

Analysis

The investment in Seltz points to a broader trend in AI infrastructure: the rise of the “Agentic Web Standard.” As AI agents become more autonomous, they require a completely different web interface than human users. Text scraping and HTML parsing are temporary, fragile, and costly workarounds. A structured, AI-native search layer is a critical prerequisite for agents to make reliable, real-time decisions. The backing of top-tier European and global VCs indicates that the market for AI developer tools is accelerating rapidly.

Practical Takeaways

For Developers: Seltz’s API v1 offers a viable alternative to traditional search APIs, helping to reduce token consumption and latency in Retrieval-Augmented Generation (RAG) applications.
For Tech Architects: The Dynamic News Search Benchmark (DNSB) should be utilized to benchmark and compare the performance of different search retrieval options for agentic frameworks.
For Enterprises: The shift toward AI-native data retrieval underscores the need to build agentic workflows that minimize data overhead and prioritize structured inputs.

Open Questions

Can Seltz scale its web index to be competitive in size and quality with the crawlers of major technology giants?
How will Seltz handle paywalled and copyrighted content as autonomous agents consume information directly through the API?
What pricing and licensing models will emerge for large-scale, automated queries by agent fleets?