Web Search Tools — Real-Time Knowledge Retrieval
What are the Web Search Tools?
Large Language Models are limited by their training cutoff dates. To make your AI agents aware of recent events, lookup pricing, or retrieve live information, Langoedge provides Web Search Tools.
These tools can be attached to any graph node, allowing the AI to construct search queries, invoke professional search APIs, and use the retrieved search results as reference context for answers.
Langoedge exposes two primary search tools:
search_web(Tavily search): Optimized for search engine retrieval, returning clean, condensed content snippets specifically formatted for LLM ingestion.exa_search(Exa search): A neural search engine that uses semantic link mappings and crawls to surface relevant research links and clean text snippets.
1. Tavily Search (search_web)
Tavily is a search engine built specifically for AI agents and LLMs. It aggregates, filters, and summarizes search results across multiple engines to optimize token consumption.
Tool Parameters
| Parameter | Type | Required | Description |
|---|---|---|---|
query |
string |
Yes | The search term or natural language question to lookup. |
Technical Execution
When the node executes, the agent formulates a search query (e.g. "Langoedge release notes 2026"). The search_web tool calls the Tavily API, matches relevant pages, extracts text snippets, and returns them in a structured text block:
{
"results": "Snippet 1: Langoedge is a platform for building AI agents... Snippet 2: Version 2.0 features low-latency voice gateways..."
}
2. Exa Search (exa_search)
Exa uses a neural database that indexes the web as a network of semantic vector associations, allowing agents to search for pages using natural language descriptions rather than just keywords.
Tool Parameters
| Parameter | Type | Required | Description |
|---|---|---|---|
query |
string |
Yes | The semantic search query. |
Best Practices for Search Integration
Query Optimization
Instruct your LLM node to write clear, target-oriented search queries. Avoid conversational filler like *"Can you find..."* — use exact keywords and parameters.
Result Summarization
Web searches yield long text snippets. Always place a summarization node after search operations to extract relevant points before feeding them back into the main state history.