Skip to main content

Web search

Overview

CodeVista integrates Google Search capabilities to supplement the knowledge of its AI models and provide additional sources of information for you to cross-check and validate the generated responses. The Web Search feature is available in two modes: "Search After" and "Search Before".

Feature

Search Before

Search After

Purpose

Augment CodeVista's knowledge with the latest information from web searches to generate a more comprehensive and up-to-date response.

Cross-check and validate the key ideas in CodeVista's response against the latest web search results.

Trigger

Use the magic command /web_search in the prompt input box.

Click or tap the magnifying glass icon after CodeVista generates a response.

Process

1. Analyze the prompt for key ideas.
2. Perform Google Web searches related to those ideas.
3. Read through top search results.
4. Synthesize an answer based on the gathered information, citing sources.

1. Summarize the key ideas from CodeVista's response
2. Generate queries based on those ideas.
3. Perform Google Web searches for those queries.
4. Display the top 3 search results for each query.

Output

A response synthesized from the information gathered from web searches, with citations.

The top 3 web search results for each query, allowing users to cross-check CodeVista's response.

Known Limitations

Long processing time before the first streaming response. Some queries may return no search results.

Some queries may return no search results.

Search Before

Start typing the command /web_search

image.png

After the command, type the prompt you want to search for.
Example: "/web_search latest 2025 AI trends"

image.png

Review search results.

image.png

Search After

Click the magnifying glass icon.

image.png

Review the search results.

image.png

By leveraging the Web Search feature in CodeVista, you can enhance the knowledge and accuracy of the AI-generated responses by incorporating the latest information from the web. This feature can be particularly useful for topics or technologies that are rapidly evolving or for which CodeVista's knowledge base may be outdated.