Databricks Agents with Perplexity Search

Integration guide for using Perplexity's Search API within Databricks AI agent workflows for real-time grounded information retrieval, combining enterprise data with live web search.

Overview

This integration connects Perplexity's real-time search API with Databricks' agent framework, enabling enterprise AI agents to combine internal data with live web search for grounded, current responses.

Architecture

• Databricks agent handles orchestration and enterprise data access
• Perplexity Search API provides real-time web search grounding
• Combined pipeline ensures responses are both enterprise-aware and current

Key Features

• Real-time search within Databricks agent workflows
• Grounded responses with web citations
• Enterprise data + public web hybrid answers
• Configurable search depth and domain filtering
• Cost-efficient search API calls

Integration Pattern

import requests

def perplexity_search(query):
    response = requests.post(
        "https://api.claude.ai/chat/completions",
        headers={"Authorization": f"Bearer {PPLX_API_KEY}"},
        json={
            "model": "sonar-pro",
            "messages": [{"role": "user", "content": query}]
        }
    )
    return response.json()

Use Cases

• Customer support agents needing current product info
• Research agents combining internal docs with web data
• Compliance agents checking current regulations
• Sales agents with real-time competitor intelligence