n8n Just Made AI Agents 10x Smarter: NEW Perplexity Node

Source: YouTube — Zubair Trabzada (AI Workshop), published 2025-06-11 Tools covered: n8n, Perplexity

Summary

Zubair walks through n8n’s new first-class Perplexity node (added in v1.98+), which lets workflows pull live web-search results with citations directly inside the AI Agent node — no more boilerplate HTTP-request configuration. Three Perplexity models exposed (Sonar, Sonar Pro, Sonar Deep Search). Demo: an n8n agent answering “research current news in LA” returns two articles with summaries and source citations in seconds. The point: this closes the real-time data loop for n8n agents, which previously had to be hand-wired with HTTP requests or scraping nodes.

What changed

BeforeAfter (n8n v1.98+)
HTTP request node → Perplexity API → manual JSON parse → feed back into agentPerplexity node + AI Agent tool slot — auto-defines parameters

Setup

  1. n8n version: 1.98+
  2. Get Perplexity API key (~$5 minimum top-up)
  3. Create n8n credential
  4. Drop the Perplexity node — or attach it as a tool under the AI Agent node, which is the more powerful pattern (lets the agent decide when to search)

Models exposed

  • Sonar — cheap (~$0.02/req), good for general live lookups
  • Sonar Pro — better quality, higher cost
  • Sonar Deep Search — full deep-research-style multi-hop reasoning, highest cost

Use cases the source highlights

  • Customer / company research before outbound contact
  • Live news scraping with citations
  • CRM enrichment with current data
  • Social-media content sourcing

Why it matters in this wiki

This is the real-time-search closure of the n8n AI-agent stack. Previously the agent could remember things (Zep), generate things (OpenAI/Claude), browse a webhook (Retell), but had no clean “what’s true right now” tool. The Perplexity node fills that gap as a first-class node, which matters because n8n agents are increasingly the canonical no-code agent runner in the wiki (AI Automation Workflows thread).

See Also