What is NLWeb?
Microsoft just announced a new tool, NLWeb, at MSBuild 2025. NLWeb (Natural Language Web) makes adding a natural language interface, such as chatbots and agents, easy for any website. With just a few lines of code, users can ask questions or make requests in everyday language—like “Show me vegan recipes for dinner” or “What’s your return policy?”—and get smart, relevant answers from the site’s own data.
NLWeb brings conversational interfaces directly to your website. With powerful AI models, NLWeb connects your website’s existing data, such as product lists, articles, or events. It uses common web data formats, so you don’t need to rebuild your site from scratch. NLWeb acts as a bridge between your content and large language models, making your site’s information accessible both to humans and to AI agents.
Every NLWeb instance is also a Model Context Protocol (MCP) server, allowing websites to make their content discoverable and accessible to agents and other participants in the MCP ecosystem if they choose. Ultimately, we believe NLWeb can play a similar role to HTML in the emerging agentic web.
Key Features of NLWeb
NLWeb helps you add chat-like AI features to any website, making it easier for visitors to find what they need. Here are its main features explained simply, with examples:
Open and Free
You don’t have to pay to use NLWeb. Anyone can download it, use it, or even help make it better.
Example: A small business owner can add NLWeb to their site without worrying about extra costs.
Natural Language Interface
Visitors can type questions or requests in everyday language, just like texting a friend or using a virtual assistant.
Example: On a travel site, a user could type, “Show me family-friendly hotels in Paris under $200,” and get a direct answer.
Flexible AI Model Support
You get to choose which AI model powers the chat, whether it’s a popular one or something you’ve built yourself.
Example: A tech company could use their own AI model for privacy, while a blogger might use a well-known public AI model for convenience.
Works With Existing Data
NLWeb uses the information your website already has, so you don’t need to redo your content or data.
Example: An online bookstore can use its current list of books and categories, letting users ask, “Do you have any mystery novels by Agatha Christie?”
Content Sharing Protocol
You can make your website’s information available to AI agents or other smart tools, if you want.
Example: A job board site can let AI agents search for jobs on behalf of users, like, “Find me remote software engineering jobs.”
Customizable and Platform-Agnostic
NLWeb works on any operating system and with different databases or AI models, so you can set it up however you like.
Example: A company can run NLWeb on their own servers with their preferred database, or a blogger can use it on a simple website hosting service.
Cost-Efficient
Since NLWeb uses your website’s existing data, you don’t need expensive technology or big servers to make it work.
Example: A local restaurant can add NLWeb to help customers ask about menu items or hours without paying for a fancy chatbot service.
Easy to Set Up
You can get started by adding just a few lines of code to your website.
Example: A nonprofit can add NLWeb in an afternoon, letting visitors ask, “How can I volunteer?” or “What are your upcoming events?”
Better User Experience
People find what they need faster and more easily, making them more likely to return to your site.
Example: An e-commerce store lets shoppers type, “Show me red dresses in size medium,” and shows matching products right away.
Ready for the Future
NLWeb is built for a new kind of internet where AI agents help people get things done online.
Example: In the future, someone could ask their digital assistant to book a hotel room, and the assistant could interact directly with hotel websites using NLWeb.
Data Control and Privacy
You decide what information is shared and how it’s used, so your data stays in your hands.
Example: A healthcare website can make sure only safe, approved information is available through the chat interface.
Helpful Documentation and Tools
NLWeb comes with guides and tools to help you get started and add more features as you go.
Example: A developer can follow step-by-step instructions to connect NLWeb to their site’s product catalog.
How Does NLWeb Work?
NLWeb is a technology that lets websites add a conversational, AI-powered interface with just a few lines of code. NLWeb leverages semi-structured formats like Schema.org, RSS and other data that websites already publish, combining them with LLM-powered tools to create natural language interfaces usable by both humans and AI agents.
As an open project, NLWeb is technology agnostic. It supports all major operating systems and allows developers to choose the components that best suit their needs, supporting all major models and vector databases.
Here’s how it works, step by step:
-
Uses Your Existing Data: NLWeb connects to the data your website already publishes, like product lists, articles, reviews, or events. It leverages common web formats such as Schema.org and RSS, which are already used by millions of websites to organize and share information.
-
Bridges to AI Models: NLWeb acts as a bridge between your website’s structured data and powerful large language models (LLMs) such as those from OpenAI, Google, Anthropic, or others. When a user asks a question in plain language, NLWeb passes that query to the AI model, along with the relevant data from your site.
-
Conversational Interface: On the front end, NLWeb provides a simple chat box or conversational widget. Users type their questions or requests—like “Find me shoes under $100” or “What’s your shipping policy?”—and the AI responds using your site’s data.
-
Model Context Protocol (MCP): Every NLWeb implementation also acts as an MCP server. This protocol allows your website to make its content discoverable and accessible to AI agents and other tools, if you choose to enable it. This is similar to how websites became accessible to browsers using HTML.
-
Customizable and Flexible: NLWeb is model-agnostic and platform-agnostic. You can choose which AI model and vector database to use, and it works on all major operating systems. It’s lightweight enough to run on anything from cloud servers to laptops.
-
For Both People and AI Agents: NLWeb’s conversational interface is designed for both human users and AI agents. This means your site can be queried by people or by other AI systems, opening up new possibilities for automation, discovery, and interaction.
-
Simple Integration: Developers can add NLWeb to their sites using familiar web technologies (HTML, JavaScript) and connect it to their data sources. There are sample implementations and open-source tools to help you get started quickly.
What Are the Benefits of NLWeb?
-
Easy Integration: Add NLWeb to your site with just a few lines of code, without needing to build a chatbot from scratch or overhaul your site.
-
Conversational Experience: Users interact with your site naturally, asking questions or making requests as if they were chatting with a person.
-
Works with Your Data: NLWeb uses the data you already have, so answers are always up-to-date and relevant.
-
Cost-Effective: NLWeb uses your existing data feeds, making it affordable for businesses of all sizes.
-
Customizable: Choose which AI models to use and control what data is accessible.
-
Future-Proof: Designed for a new era where AI agents can interact with websites and perform tasks for users.
-
Open Source: Anyone can use, modify, or contribute to NLWeb.
Who Should Be Learning NLWeb?
-
Web developers and designers who want to add conversational features to their sites without deep AI expertise.
-
Business owners and publishers looking to enhance user experience and engagement.
-
Product managers and digital strategists planning for the future of AI-driven web experiences.
-
Anyone interested in AI, chatbots, or the next generation of web technology.
How Can Companies Benefit from NLWeb?
-
Better Customer Support: NLWeb can answer common questions instantly, reducing the need for a full-time support team.
-
Personalized Experiences: By understanding user intent, NLWeb can tailor recommendations, content, or products to each visitor.
-
Increased Conversions: Making it easier for users to find what they want can directly boost sales or sign-ups.
-
Data Control and Privacy: Companies keep control over their own data and how it’s used.
-
Faster Implementation: Simple setup means companies can launch conversational features quickly, without major changes to their existing sites.
-
Competitive Edge: Early adopters can stand out by offering smarter, more interactive web experiences.
Is NLWeb Paid?
NLWeb is free and open-source. There is no cost to use the software itself. If you choose to connect NLWeb to commercial AI models or cloud services, those providers may charge their own fees. The core NLWeb framework is free to download, use, and modify.
What Skills Do You Need to Use NLWeb?
-
Basic Web Development: Familiarity with HTML and JavaScript is enough to get started.
-
Understanding of Data Formats: Knowing how your site’s data is structured helps with integration.
-
API Usage: For advanced features, experience with APIs and connecting to AI models or databases is useful.
-
Optional Advanced Skills: If you want to customize which AI models or databases your site uses, some knowledge of AI and backend systems can help, but it’s not required for basic usage.
NLWeb is a simple, flexible, and free way to make your website smarter and more interactive. Whether you’re a developer, a business owner, or just curious about the future of the web, learning NLWeb can help you stay ahead as conversational AI becomes the new standard for online experiences.
How do I get started with NLWeb?
The NLWeb GitHub repo contains everything you need to get started:
- The lightweight code that controls the core service to handle natural language queries, as well as documentation on how this can be extended and customized.
- Connectors to some of the most popular models and vector databases, as well as documentation to add other models of your choice.
- Tools for adding your data in Schema.org, JSONL, RSS and other formats to your chosen vector database.
- A web server frontend for the service and a simple UI that allows users to send queries to the web server.