Microsoft Unveils Web IQ AI Grounding Platform Built on Bing Search

Microsoft has announced Web IQ at Build 2026, a new suite of AI-native grounding APIs designed to help AI applications and autonomous agents access up-to-date information from the web when generating responses.

The company says Web IQ allows AI systems to ground answers using web pages, news articles, images, and videos. Unlike traditional search services that primarily return lists of links, Web IQ delivers relevant passages and structured evidence objects that AI models can use directly.

According to Microsoft, this approach reduces the amount of unnecessary context sent to AI models, lowering token consumption while improving the quality and accuracy of grounded responses.

Built specifically for AI agents

Web IQ is powered by Bing’s global search index and was designed from the ground up for AI-native workloads. Microsoft says the platform has been redesigned across every major layer of the retrieval stack, including indexing, retrieval, ranking, passage selection, and orchestration.

Instead of forcing developers to build complex retrieval pipelines on top of conventional search engines, Web IQ aims to provide AI systems with concise evidence that can be consumed more efficiently by large language models.

Microsoft says this allows AI applications to retrieve the information they need without processing large volumes of irrelevant content.

Powered by Harrier embeddings and DiskANN technology

Under the hood, Web IQ uses Microsoft’s Harrier embedding models alongside technology derived from DiskANN, the company’s large-scale nearest-neighbor search system.

DiskANN has been a core component of Microsoft’s large-scale search and retrieval infrastructure for years, helping deliver fast similarity search across massive datasets.

By combining advanced embeddings with optimized retrieval systems, Microsoft says Web IQ can provide high-quality results while maintaining low latency.

Focus on speed and efficiency

In the company’s internal testing, Web IQ reportedly achieved sub-165 millisecond p95 latency across five data center regions. Microsoft claims this makes the platform nearly 2.5 times faster than the next-best alternative under comparable configurations.

The company also says Web IQ delivers grounded results using significantly fewer tokens than competing services, reducing inference costs for developers building AI applications.

Microsoft did not identify which competing products were used for the comparison.

Built with publisher controls in mind

According to Microsoft, the service supports robots exclusion protocols, publisher controls, and content access preferences. This means website owners can continue managing how their content is accessed and indexed through the same mechanisms already used by Bing.

The company says maintaining compatibility with existing publisher controls remains a key part of the platform.

Availability

Web IQ is not generally available at this time. Microsoft has opened registration for developers and organizations interested in early access. Existing customers using Grounding with Bing Search will continue to have access to that service.

Microsoft says Grounding with Bing Search will remain available as an entry point for organizations that prefer a more traditional search-based grounding experience.

Part of Microsoft’s broader AI push

The Web IQ announcement was one of several major AI announcements made during Build 2026.

Microsoft also introduced new MAI AI models, unveiled Microsoft Scout, an autonomous work assistant for Microsoft 365, and announced the new Rayfin platform as part of its expanding AI ecosystem.

With Web IQ, Microsoft is positioning Bing’s search infrastructure as a foundation for the next generation of AI agents, offering developers a way to access current web information with lower latency, fewer tokens, and more structured evidence than traditional search APIs.