Exa co-founder Will Bryk predicts that AI systems will conduct more searches than humans by 2026, a shift that will require a new kind of search infrastructure. This claim underscores the need for an embedding-based, agent-first search and data marketplace to prevent misinformation and enable better decision-making.
TL;DR
- Exa's Will Bryk forecasts AI searches will exceed human searches in 2026, driven by the proliferation of AI agents.
- Exa positions itself as the retrieval layer for the agentic web, offering a search API optimized for AI systems.
- The company's data marketplace aims to provide a unified query surface for both public and private data sources.
What happened
In a recent 17-minute talk, Will Bryk, co-founder of Exa, argued that the volume of searches issued by AI systems will surpass human search volume by 2026. He further projected that within a few years, AI systems will search roughly a thousand times more than humans. Exa, founded in 2021 as Metaphor and now serving over 5,000 companies and 400,000 developers, positions itself as the retrieval layer for the agentic web. The company sells a search API rather than a consumer destination, aiming to provide a queryable database of the world's information.
Bryk's talk covered Exa's mission, history, product, and roadmap. He argued that the web's vastness and the lack of a complete information infrastructure lead to poor individual and collective decisions. He tied this information gap explicitly to the 2028 US presidential election, emphasizing the need for near-perfect information. Exa's history includes a pivot from a consumer search engine to an API business after inbound requests from developers. The company's current capabilities include a 200-millisecond search endpoint, token extraction for downstream LLM cost savings, structured output, and domain and time-window filtering.
Why it matters
The shift in search volume from humans to AI systems necessitates a new kind of search infrastructure. Bryk argued that Google's optimization for human queries is misaligned with what AI agents need, which is closer to a database lookup. Exa's search API is designed to meet these needs, providing a more efficient and accurate search experience for AI systems. This shift has significant implications for developers, startups, and investors in the AI space, as it highlights the growing importance of AI-first search solutions.
Exa's data marketplace extends its search capabilities beyond the public web, allowing private data providers to monetize access through the same query surface. This creates a new economy for agents, enabling them to access a broader range of information. The company's roadmap includes achieving 'perfect search,' where any information query returns near-perfect results. This exponential progress in search technology has the potential to revolutionize how AI systems interact with and retrieve information from the web.
Key facts
- Exa predicts AI searches will exceed human searches by 2026, according to Will Bryk's talk.
- The company serves over 5,000 companies and 400,000 developers, per Exa.
- Exa's search API offers a 200-millisecond search endpoint, according to the company.
- The company's data marketplace allows private data providers to monetize access, per Exa.
- Exa's roadmap includes achieving 'perfect search' with near-perfect results for any query, according to Will Bryk.
- The company's search volume thesis is based on the proliferation of AI agents, which will ground each interaction in retrieval, according to Bryk.
- Exa's customer mix spans frontier labs through YC startups, according to Will Bryk.
- The company's search quality for agents is better than Google's, according to Bryk.
Context
Exa's predictions and developments come at a time when AI agents are becoming increasingly prevalent. These agents require efficient and accurate search capabilities to function effectively. The shift from human to AI search volume highlights the need for new search infrastructures that can meet the unique demands of AI systems. Exa's positioning as the retrieval layer for the agentic web underscores the growing importance of AI-first search solutions in the broader AI landscape.
The proliferation of AI agents and the need for efficient search capabilities are driving innovation in the AI industry. Companies like Exa are at the forefront of this shift, developing solutions that cater to the unique needs of AI systems. As AI agents become more integrated into various applications and services, the demand for advanced search technologies will continue to grow. This trend presents significant opportunities for developers, startups, and investors in the AI space.