Web Content as Context

User Problem

Users often need to generate applications or components (charts, tables, etc.) based on external reference information found on the web, such as up-to-date documentation or specific articles. The LLM, trained on historical data, cannot natively access the current, full content of a specific URL or the latest information found through a web search. This leads to outputs that can be outdated, inaccurate, or fail to follow the instructions contained on a specific web page.

What is it?

Web Context enables the Plotly Studio LLM to fetch and integrate real-time or specific web content into its context. This is achieved by allowing users to provide one or more URLs (links to websites, documentation pages, etc.). The system will retrieve the text content from the specified source(s) and use it to "ground" the LLM's response.

What does it allow users to do?

This feature allows users to ensure the LLM generates artifacts based on the most relevant and current information available on the public web. Specifically, users can:

• Specify a Content Source: Provide specific URLs (e.g., a link to a REST API documentation page or a public data report) to ensure the LLM's generated code, charts, or tables are based on that precise source content.

• Generate Accurate Code and Designs: Direct the LLM to follow live design standards or technical specifications hosted on a website, reducing the need for manual corrections and speeding up the development of accurate, well-contextualized applications.

• Maintain Context Scope: Attach URL context to an entire project or to an individual component being generated, allowing for global or surgical context infusion into the LLM's task.