AI Web Research Workflow: SerpAPI, Claude, Google Docs

Turn a research question into a cited report—automatically

This n8n AI Web Research workflow takes a research query via webhook, uses Anthropic Claude to plan targeted searches, pulls web results with SerpAPI, synthesizes a structured, URL-cited report, then saves it to Google Docs—while notifying Slack and logging run details to Postgres.

What this workflow does

• Receives your query: Listens for a POST webhook containing a research question.
• Plans search sub-questions: Uses Anthropic Claude to generate 3–5 focused sub-questions and splits them into individual search tasks.
• Fetches web results: Uses SerpAPI to query Google results for each sub-question and aggregates the organic results.
• Synthesizes a cited report: Uses Anthropic Claude to write a structured research report with inline URL citations.
• Scores confidence: Uses Anthropic Claude to score overall confidence and flag low-confidence claims.
• Saves and notifies: Creates a Google Doc for the report, then posts a high-confidence or human-review message to Slack based on a confidence threshold.
• Logs everything: Stores run metadata (query, doc URL, confidence, and low-confidence count) in Postgres.

Use cases

• Generate rapid, cited background research for product or market decisions.
• Produce web-based briefings for SaaS operators and founders with traceable sources.
• Automate research-to-document workflows for analysts and ops teams (Doc + Slack + database log).
• Speed up investigative tasks by breaking questions into targeted sub-queries.

Technical details

• AI: Anthropic Claude for planning, synthesis, and confidence scoring
• Web search: SerpAPI HTTP requests
• Webhook trigger to accept research queries
• Google Docs to create the report document
• Slack notifications to a configured channel
• Postgres to log query, doc URL, confidence, and low-confidence count
• Workflow nodes: if, set, code, slack, webhook, postgres