This is a submission for the AI Agents Challenge powered by n8n and Bright Data

What I Built

Picture this: You're a developer in Bangalore (or any tech hub), and you're constantly hearing about "that amazing meetup I missed last week" or "did you know there was a React conference yesterday?" Sound familiar? Yeah, me too.

That's when I decided to build my own AI-powered event discovery agent that I lovingly call the "Event Butler" - because who doesn't want a digital butler that's better at finding events than you are at finding matching socks?

The Problem: Manually checking Eventbrite every morning is about as fun as debugging CSS in Internet Explorer. Plus, with the sheer volume of events posted daily, the good stuff often gets buried under "Learn Blockchain in 2 Hours" seminars.

The Solution: An intelligent agent that:

• 🕐 Wakes up at 8 AM every day (more reliable than me, honestly)
• 🕷️ Scrapes Eventbrite for fresh tech events in Bangalore
• 🧠 Uses Google Gemini AI to curate and create engaging newsletters
• ✨ Delivers beautifully formatted Morning Brew-style emails to my inbox
• 💡 Adds personality and wit to make event discovery actually enjoyable

Think of it as having a super-enthusiastic friend who's really good at finding cool events, never sleeps, and has impeccable taste in tech meetups.

Demo

Since I'm rocking the n8n cloud free tier (hey, we all start somewhere! 💸), I can't make my workflow public. But I've got you covered with a full video walkthrough and all the code you need to build your own event-finding sidekick.

📹 Full Demo Video:

In the video, you'll see the entire workflow in action - from the initial scrape to that satisfying moment when a perfectly formatted newsletter lands in my inbox. It's oddly therapeutic, like watching a really well-organized spreadsheet come to life.

📧 Sample Newsletter Output:

What lands in my inbox every morning - beautifully formatted and surprisingly witty

n8n Workflow

🔗 Complete Workflow JSON: https://gist.github.com/Varshithvhegde/a40820f639034bac181e93146a9a21a1

The gist includes:

• Complete workflow configuration
• AI agent system instructions
• Email template structure
• Custom parsing logic for event data

Pro tip: The workflow JSON is like a recipe - you can totally customize the "seasoning" (prompts) to match your city, interests, or sense of humor!

Technical Implementation

The AI Agent: My Digital Event Connoisseur

Model Choice: Google Gemini (because it's chatty, smart, and doesn't judge my terrible event discovery skills)

System Instructions: I gave my AI agent a personality somewhere between a tech-savvy friend and a Morning Brew editor. The prompt instructs it to:

• Deduplicate similar events (because "JavaScript Workshop #47" doesn't need to appear 5 times)
• Categorize by timing: Today, Tomorrow, This Week, Upcoming
• Add engaging commentary and subtle humor
• Format everything in beautiful HTML template that doesn't look like it was designed in 1995
• Detailed System Prompt is here : https://notepage.vercel.app/n8n-brightdata-challenge-system-prompt

Memory: The agent remembers event patterns within each session to avoid redundancy, though it starts fresh each day (like my motivation to go to the gym).

Tools Used:

• Data Processing: Custom JavaScript nodes for parsing messy HTML into structured gold
• Email Generation: AI agent with carefully crafted prompts
• Delivery: Gmail integration (because who doesn't love getting good news in their inbox?)

Bright Data Verified Node: The MVP of This Operation

Here's where Bright Data really shines (pun intended ✨). Eventbrite is not exactly thrilled about bots scraping their content - they're quicker to block IPs than a Twitter user blocking trolls.

How I used Bright Data:

• Reliable Access: No more "403 Forbidden" errors ruining my morning routine
• Residential Proxies: Appears as legitimate traffic (because it is!)
• Rate Limit Handling: Built-in throttling that plays nice with Eventbrite's servers
• Global Infrastructure: Ensures consistent access even during peak traffic

The Bright Data node basically turned my scraper from a sketchy back-alley operation into a legitimate, well-dressed data collection service. It's like upgrading from sneaking into events to having VIP passes.

Configuration Highlights:

• Target URL: Eventbrite's Bangalore tech events page
• Headers: Proper user-agent and browser fingerprinting
• Retry Logic: Because sometimes the internet hiccups
• Data Extraction: Clean HTML ready for parsing

Journey

The "This Should Be Easy" Phase 🤔

Started with the classic developer mindset: "How hard could it be to scrape some events?"

Narrator: It was harder than expected.

Initial Challenges:

• Eventbrite's anti-bot measures (they're surprisingly sophisticated)
• Inconsistent HTML structure (apparently not everyone follows semantic HTML practices)
• Data quality issues (turns out event descriptions can be... creative)
• Email formatting nightmares (CSS in emails is still terrible in 2025)

The "Why Won't This Work?!" Phase 😤

Spent way too many hours debugging CSS selectors that worked perfectly in the browser but failed spectacularly in the scraper. Also learned that Eventbrite returns different HTML structures depending on whether you're a logged-in user, mobile visitor, or bot.

Key Breakthrough: Bright Data's verified node solved 90% of my scraping headaches. Suddenly, consistent data! It was like finding out your code works on the first try - magical but slightly suspicious.

The "AI Makes Everything Better" Phase 🤖

Initially, I was just planning to send raw event data. But then I thought, "What if the AI could make this actually interesting to read?"

Game Changer: Crafting the right prompt for Gemini. Getting the AI to:

• Understand event context and importance
• Add personality without being annoying
• Format HTML consistently
• Handle edge cases gracefully

The prompt engineering took longer than the scraping logic, but it was totally worth it. Now my newsletters read like they're written by a witty tech journalist who actually cares about my calendar.

The "It Actually Works!" Phase 🎉

Three months in, and I've:

• Discovered 47 events I would have missed
• Saved approximately 2 hours per week on event hunting

Unexpected Bonus: The AI's commentary often highlights events I wouldn't have considered. Turns out machine learning isn't just good at recognizing cats in photos - it's also pretty decent at understanding what makes events worth attending.

What I Learned

1. Good prompts are worth their weight in gold - Seriously, spend time on this
2. Reliable data sources matter more than perfect parsing - Bright Data was a game-changer
3. Automation should enhance, not replace, human judgment - The AI suggests, I decide
4. Small conveniences compound into big productivity gains - 10 minutes saved daily = 60+ hours yearly
5. Building for yourself often creates the best products - I actually use this every day

Future Enhancements (Because We're Never Done)

• Multi-city support (for when I travel or move)
• Smart filtering based on past attendance and preferences
• Calendar integration (one-click event adding)
• Community features (share interesting finds with friends)
• Analytics dashboard (because data is beautiful)

---

The Bottom Line: If you're tired of playing event discovery roulette, build yourself an AI assistant. Your future networking self will thank you, and you'll never have to hear "Oh, you missed that amazing talk?" ever again.

Now if only I could build an AI agent to actually make me show up to the events... 🤷‍♂️

---

P.S. - If you build your own version, I'd love to see what events your AI discovers! Drop a comment with your city or any cool modifications you make. Let's make FOMO a thing of the past, one automated newsletter at a time.