This is a submission for the Bright Data AI Web Access Hackathon
What I Built
We built Realyze — an AI-powered real estate chatbot that helps users find the perfect home just by describing what they want. Whether you're looking for a "3-bedroom apartment in Lagos with a sea view" or a "budget-friendly studio near Yaba," Realyze goes beyond static listings to fetch real-time property data from the web.
It combines conversational AI with web scraping to deliver up-to-date property recommendations, eliminating the frustration of outdated or irrelevant listings. Think of it as your personal real estate agent, available 24/7.
Realyze uses:
🧠 CrewAI to coordinate multi-agent collaboration
🕸️ Crawl4AI for scraping logic orchestration
🚪 Bright Data Web Unlocker API to access real estate websites in real time
💬 MongoDB for persistent chatbot memory
🌐 Django (backend) and React (frontend)
Demo
GitHub Repository: https://github.com/T4910/purpletalk-interface-ai/
Video demo: https://youtu.be/GgbcCl3ACGs
View preview in the comment section
How I Used Bright Data's Infrastructure
Bright Data’s Web Unlocker API was critical to the success of Realyze. Real estate websites often have anti-scraping mechanisms or CAPTCHAs that block bots. Bright Data seamlessly bypassed these barriers, ensuring that we could:
Scrape multiple real estate websites in real time
Maintain reliability without being blocked
Scale our scraping agents without worrying about proxies, headers, or captchas
This infrastructure allowed us to focus more on building the intelligence and UX of Realyze, rather than the nitty-gritty of anti-bot evasion.
Performance Improvements
Using real-time web data access gave our chatbot a serious edge. Traditional approaches that rely on static datasets or scheduled scrapers often serve outdated listings. But Realyze:
Delivers fresh, real-time property listings
Adapts instantly to market changes and new listings
Improves user trust and satisfaction by showing only available properties
Without Bright Data, this level of responsiveness wouldn’t have been possible.
If you liked what we built, check out Bright Data’s GitHub and show them some love! 🌟
Team Credits:
oluwaseyifunmi_oshinfowok
codej
Top comments (9)
Link to deployed URL: ec2-3-86-209-219.compute-1.amazona...
Update:
Here's the new (and stable-ish) link to the project: realyze.floo.com.ng/
Love to get feedback on what we can improve.
Love how you solved the outdated listing problem with real-time data. Are you planning to expand to cover other cities or countries next?
Hey @dotallio — thank you for the kind words (you have no idea how much it means to us)!
Our current solution actually does support multiple countries already 🌍
I’d have loved for you to try it out — but unfortunately, our OpenAI credits just ran out 😅
We’re working on getting everything back up and running soon!
Just checked the application itself and the feel is really nice not to talk about the functionality and performance which are good qualities of a software.
The ability to do this much with little is truly amazing. Nice work y'all
Weldone guys
This is wonderful to see
Nice one bro. It's really innovative 💯
Superb👏
This is so good
Well done!!
Well-done