This is a submission for the AI Agents Challenge powered by n8n and Bright Data and my first n8n workflow.
What I Built
Searching for deals manually can be time-consuming and frustrating. Most of the time, this costs us as we often miss out on the best prices and limited-time offers. This AI agent solves this problem by automating the deal-hunting process and providing a single, intelligent interface for finding the best discounts.
How It Works
This AI agent, built with n8n and powered by Bright Data, is designed for efficient web scraping. The agent then uses the Bright Data n8n Verified Node as a tool to perform real-time web scrapes for the requested product based on the user's chat message. The output (the scraped data) is then presented to the user in a clear, friendly, and conversational summary.
Demo
Agent link: http://localhost:5678/workflow/2hIkozBxX4QEcxyZ
Demo Video: https://youtu.be/TGljGiiHpRo
n8n Workflow
https://gist.github.com/Bennykillua/107287dc7a31dda58c6b5a7c0bf6ab88
Technical Implementation
- Our trigger is a chat.
- The agent is configured as a highly enthusiastic “Deal Hunter” AI. Its system prompt ensures it only responds when the user requests a deal, maintaining an energetic and competitive tone. This keeps the agent focused on just finding and suggesting the best possible product deals.
- For the model, I make use of OpenAI’s GPT-4.1-mini model. This model strikes a balance between cost efficiency and high-quality natural language understanding.
- The memory, I made use of a simple memory buffer that stores the last 10 exchanges. Hence, it remembers what items the user is interested in during the session and avoids repeating questions unnecessarily.
Bright Data Verified Node
The node (brightDataTool), which I used as a tool for my AI agent, is set to filter a marketplace dataset. It dynamically receives parameters from the AI agent (Dataset, Field_Name, Field_Value). These parameters are auto-generated based on the user’s request (e.g., “laptop under $500” → dataset = electronics, field = price, value = 500).
The node returns a filtered snapshot of matching deals from Bright Data’s real-time datasets.
Journey: Process, Challenges, and Learnings
I started learning about n8n just 10 days ago, though I have heard of the tool. I learned through YouTube tutorials, driven by my desire to complete this task. My background is in data analysis and business process automation, utilizing Microsoft Power Platform, which has provided me with a strong foundation. However, working with an AI-powered agent in n8n was a very new experience for me, and to be honest, it feels quite different compared to the more structured workflows in Power Platform.
Exploring n8n has been exciting because it combines low-code workflow automation with AI-driven logic and third-party data integrations, making it very powerful and fun to build with. I’m still learning the tool, but I already see how flexible it is for creating intelligent automations.
Challenges
Since this was my first deep dive into AI agents, I had to adapt to how they handle prompts, memory, and external tool usage, which was a learning curve. There are features I would have liked to implement but couldn't cause of the deadline:
- Sending emails automatically after finding deals.
- Returning richer data through the Bright Data node (more descriptive results and structured insights).
These were limited by both the tight deadline and my current skill level with n8n.
What I Learned
Building with n8n is very approachable even for someone coming from a Power Platform background. AI agents introduce a new layer of flexibility and intelligence that traditional automation tools lack. Lastly, the Bright Data node integration demonstrated to me how external data can enrich AI workflows, bridging the gap between automation and actionable, real-world insights.
To be honest, this has been a great first step into AI-driven automation with n8n, and I’m looking forward to building more advanced workflows. The community has also been pretty helpful.
Top comments (0)