Tradeshow Scanner
An intelligent lead qualification workflow powered by n8n AI Agent, Bright Data, and GPT-5-mini
(This post is reproduced from on a Hackaton submission by us – you can access the original post here)
Trade shows are goldmines for business development, but manually researching hundreds of exhibitors is time-consuming and inefficient. This workflow automates the entire prospecting pipeline for trade show attendees.
What it does
The workflow intelligently scans and qualifies potential clients from any trade show exhibitor list:
- Automated Data Extraction: Uses Bright Data’s web scraper to fetch the full exhibitor list from the trade show website, then extracts company names, descriptions, booth numbers, and website URLs using HTML parsing
- Smart Batching: Processes companies in configurable batches (currently set to a small number for testing, easily adjustable) to manage API costs during development
- AI-Powered Research & Qualification: Each company goes through an AI Agent that:
- Conducts web research using SerpAPI and targeted website scraping
- Evaluates company size (small/medium/large based on employee count)
- Assesses their public stance on data, AI, and automation initiatives
- Scores each company (low/medium/high) as a potential client based on your specific criteria
- Provides reasoning for each score to help prioritize follow-ups
- Automated Documentation: Results are automatically appended to a Google Sheet with structured data including company details, qualification scores, and AI reasoning
The Intelligence Layer
The AI Agent is configured with detailed context about my company (lyfX.ai). It specifically looks for companies that match our profile and an “ideal” client profile: small-to-medium businesses with a modern, future-oriented approach who value working with specialized, high-caliber teams.
This creates a pre-qualified, prioritized list of booth visits and follow-up targets, transforming what would be hours of manual research into an automated, intelligent filtering system.
Demo
Video
https://www.loom.com/share/978e4d64c11b405aa85c8d72aca8b6a1?sid=0be6cc43-ea17-461a-9f1d-4e563cd8edac
n8n Workflow
https://gist.github.com/andremoreira73/fe972f825a5f41e36c8df5ad7cb1f2bb
Note that the workflow has been sanitized so anyone can adpat it to their own use case. I left the prompts intact, as they may serve as inspiration for others.
Technical Implementation
I designed this workflow with simplicity in mind – keeping the implementation clean while maintaining effectiveness. The HTML parsing and code transformations are minimal, just enough to structure the data and make it digestible for the AI agent. (The JavaScript in the code node was kindly prepared by my good friend Claude 😊)
Model Choice: I am using GPT-5-mini, which I have tested across other setups. What I particularly appreciate for agent workflows is OpenAI’s consistent balance between reasoning capability and reliable structured output formatting. After nearly 2 years with the OpenAI API, there is also the practical advantage of familiarity and development speed.
Memory Architecture: This initial version runs stateless – each company evaluation is independent. For a future version, I am planning to implement memory for more sophisticated research patterns and cross-referencing between companies.
Tool Stack: Kept it focused with just two tools:
- SerpAPI for web search – reliable and fast for internet-wide search
- Bright Data for targeted scraping – handles sites that the agent may want to review
System Instructions: My favorite part! This is where the magic happens – carefully prepared a system prompt with instructions that encode our ideal client profile and evaluation criteria, ensuring consistent and relevant qualification scoring.
The beauty of this approach is its flexibility – swap the trade show URL, adjust the target profile in the system instructions, and you have a reusable lead qualification system for any event.
< Background >
The user provides a list of companies that will be participating in a trade show. The list contain the company's name, description, stand and website.
I will be representing my company (lyfX.ai) at the trade show.
About lyfX.ai: We help organizations integrate Data and AI solutions that address their business challenges and deliver measurable results. By merging deep business acumen, domain knowledge, data science, and modern AI, we deliver strategies that drive growth, enhance efficiency, and increase profitability. We have deep knowhow in chemicals, process engineering, fertilizers and adjacent industries.
We are a small team with very high caliber people: PhDs, one is a full professor at a reputable university in Germany, the other had a long successful career in the chemicals and fertilizer industries. Our team members hold different certifications: e.g. Google Associate Cloud Engineer, Google Professional ML Engineer, among others.
We are proficient in python, LangGraph, n8n, django.
</ Background >
< Objective >
Find out from the list of companies, which ones we should approach as potential new clients.
</ Objective >
< Target client profile >
- small to medium size company
- expresses publicly a high interest in data, AI, automation
- modern and future oriented
- not afraid of working with a small, high caliber team
</ Target client profile >
< Instructions >
For each company in the list, do the following:
1) Find out what they do
2) Is this a small (less than 50 employees), medium (between 10 and 1000 employees) or large company (over 1000 employees)?
3) Based on what you found out, rate how this company scores as a target client profile: low, medium or high?
Be thorough and stick to the facts.
If you don't know something or you don't find something, leave it blank.
</ Instructions >
< Tools >
SerpApi: Google as search engine for internet search
Scraper: Scrape a specific URL as needed
< /Tools >
< Answer format >
JSON format with the elements:
- name
- description
- stand
- website
- score as a target client profile
- reasoning for the score
</ Answer format >Bright Data Verified Node
I used Bright Data at two critical points in the workflow:
- Initial Data Extraction: Bright Data’s web scraper handles the complete exhibitor list extraction from trade show websites, reliably parsing pages (JavaScript-heavy pages, dynamic content, etc).
- As an agent tool for its research: Bright Data enables the AI to perform targeted scraping of individual company websites on-demand. This allows the agent to gather real-time public information about company size, technology focus, and initiatives directly from source.
The integration was easy. Bright Data’s n8n nodes required minimal configuration and handled complex sites that typically require browser automation.
The reliability meant zero manual intervention even when processing hundreds of exhibitors, making it ideal for production use cases where consistency matters.
Journey
This project was an excellent opportunity to explore n8n and Bright Data in depth. While I am invested in LangGraph, I have discovered that n8n excels at rapid prototyping in ways LangGraph does not. My usual workflow development involves numerous Jupyter notebooks for experimenting with graphs and agents, but n8n has opened up a new avenue: I can now create functional prototypes in hours rather than days, then either productionize them in LangGraph or keep them running in n8n directly.
Using Bright Data was a great experience. Previously, I had been using other scraping solutions, but Bright Data’s approach is really elegant and robust. The AI-assisted collector builder particularly impressed me (though this workflow ultimately didn’t require it). What stands out is the platform’s versatility. The tool has definitely earned a permanent place in my tech stack.
Next steps
The combination of n8n’s visual workflow building and Bright Data’s reliable data extraction creates a powerful environment that bridges the gap between “quick experiment” and “production-ready solution.”
This sweet spot is exactly what I have been looking for in my automation projects. I already created a few more business development workflows and I am looking forward to sharing them over time!
EN 
DE 
PT 
ES