Microsoft has introduced Web IQ, the latest addition to its IQ suite, unveiled at Build 2026. The IQ suite, which includes Work IQ, Fabric IQ, and Foundry IQ, builds on Microsoft’s former “Graphs” data architecture, using the same data as a sparse vector store to provide grounding data for LLM-based applications. By treating data as embedding vectors and integrating with Model Context Protocol (MCP) servers, Microsoft aims to tie LLM output to data, reducing hallucination risk and improving accuracy. Web IQ is a modernization of the retired Bing Search APIs, functioning as an agent-focused web search tool that uses the Bing search index to supply up-to-date general information. It extends the IQ suite’s purpose of grounding AI applications with Microsoft-hosted data to include broader web data, useful for scenarios like ecommerce product comparisons or weather-sensitive stock management via Bing’s weather feeds. Microsoft has tuned its search vector index and technology stack for agents, which operate differently from human searchers, requiring persistent queries, refined reasoning, and quick responses with large data volumes, unlike typical one-word or two-word human requests.