NLWeb aims to turn web properties into AI-powered apps by enabling natural language interactions with their content, potentially changing how people access information beyond conventional browsers and search results.

Microsoft describes NLWeb as a modern leap similar to how HTML opened publishing, but with AI agents that query content directly instead of users navigating links.

The project is led by R.V. Guha, a Microsoft executive known for work on web infrastructure standards like RSS, RDF, and Schema.org, highlighting a focus on structured data and interoperability.

Its fate remains uncertain—adoption could be limited or lead to broad standardization—making Build 2026 a potential inflection point for progress.

NLWeb endpoints function as Model Context Protocol servers, an open standard from Anthropic that connects AI systems to external data sources, integrating websites into the AI agent ecosystem.

NLWeb is an open, Microsoft-backed protocol unveiled at Build 2025 that lets websites answer natural language questions directly, without a separate search engine intermediary.

Unlike a typical chatbot, NLWeb embeds AI querying capability directly into each website endpoint via MCP, enabling seamless AI-driven queries over published content.