Find prospects and summarize business opportunities for sales
Posted on Aug 31, 2025. Task: The author built a 'Lead Opportunity Finder' in n8n to automate B2B prospecting research. The workflow uses a chat trigger to input a prospect URL, then employs Bright Data's Web Unblocker to scrape website content. Three distinct AI agents via OpenRouter (using o4-mini for URL filtering and gpt-5 for synthesis) extract relevant pages, summarize content, and merge insights into a final report. The author notes that human involvement is not explicitly defined in the post, implying the workflow runs autonomously once triggered.; tools: n8n. Human role: The original post does not say how a person stayed involved.; limitations: The workflow relies on Bright Data for scraping, which incurs operational costs and requires API keys. The author notes that initial single-prompt attempts failed due to irrelevant information, requiring careful prompt engineering. There is no mention of error handling for invalid URLs or scraping failures, and the post does not specify if human review is required before using the output for outreach..
Tools
Published on Agent Work Atlas:
Overview
Posted on Aug 31, 2025
Workflow
The author built a 'Lead Opportunity Finder' in n8n to automate B2B prospecting research. The workflow uses a chat trigger to input a prospect URL, then employs Bright Data's Web Unblocker to scrape website content. Three distinct AI agents via OpenRouter (using o4-mini for URL filtering and gpt-5 for synthesis) extract relevant pages, summarize content, and merge insights into a final report. The author notes that human involvement is not explicitly defined in the post, implying the workflow runs autonomously once triggered.
Reported result
The author reports that breaking the task into specialized stages improved accuracy and reliability compared to a single-prompt approach. However, no specific quantitative metrics, conversion rates, or time savings are provided. The post describes the technical success of bypassing scraping blocks but does not measure the business impact of the generated leads.
Human role
The original post does not say how a person stayed involved.
Limitations
The workflow relies on Bright Data for scraping, which incurs operational costs and requires API keys. The author notes that initial single-prompt attempts failed due to irrelevant information, requiring careful prompt engineering. There is no mention of error handling for invalid URLs or scraping failures, and the post does not specify if human review is required before using the output for outreach.
Try it
Act as an n8n workflow builder. Create a 'Lead Opportunity Finder' for [replaceable: target industry] using [replaceable: n8n instance URL]. Input: [replaceable: prospect website URL]. Steps: 1. Use a Chat Trigger to receive the URL. 2. Configure Bright Data Web Unblocker node to fetch HTML. 3. Use a Code node to parse and clean URLs. 4. Set up an AI Agent node with OpenRouter (model: o4-mini) to identify relevant company pages. 5. Use another AI Agent node (model: gpt-5) to summarize business opportunities from scraped content. 6. Use a Merge node to combine summaries. 7. Output a final JSON report. Constraint: Do not send any emails or messages. Stop after generating the report. First, run a dry test with [replaceable: sample URL] and show me the raw output for approval before finalizing the workflow.
Source reports are not independently verified. Publication date is the catalog publication date, not the original article date.