This Actually Surprised Me
The video demonstrates Claude or a similar AI model's ability to generate a real estate listing flyer by pulling live data, text, and images directly from a Zillow listing URL. The presenter uses a $5.3 million San Diego home as a test case. The result includes accurate address, price, property details, and actual listing photos.
Summary
The presenter opens by claiming they will show viewers capabilities of an AI model that most people are unaware of. To demonstrate, they use a real, active Zillow listing for a $5.3 million modern coastal home in San Diego as a test case.
The experiment involves giving the AI a prompt to create a real estate listing flyer, providing only the Zillow page URL and a brief description of the property type. The AI is instructed to figure out the remaining details on its own.
The result is a polished flyer that accurately reflects the live listing: it includes the correct address, the correct price, and verified property details scraped directly from the Zillow page. Most impressively to the presenter, the AI also pulled in the actual listing photographs from the webpage and incorporated them into the flyer design.
The presenter highlights this as one of the most surprising outcomes of their testing, emphasizing that the AI autonomously retrieved and integrated both textual and visual content from a third-party website. They conclude by noting the practical implication for real estate agents, suggesting that creating physical listing flyers — traditionally a manual task — has been dramatically simplified by this capability.
Key Insights
- The AI model was able to autonomously pull accurate property details — including address and price — directly from a live Zillow listing URL without the user manually inputting that data.
- The presenter claims the AI also retrieved and embedded the actual listing photographs from the Zillow page into the generated flyer, which he described as the most surprising capability he discovered during testing.
- The presenter used a $5.3 million San Diego modern coastal home as a real-world test case, not a fabricated example, to validate the AI's output against a verifiable source.
- The user's prompt was deliberately minimal — providing only the URL and a general instruction to 'figure everything else out' — suggesting the AI demonstrated significant autonomous decision-making in design and content selection.
- The presenter frames this capability as a direct workflow replacement for real estate agents who produce physical listing flyers, implying the task that previously required manual effort can now be completed in a single AI prompt.
Topics
Transcript
[0:00] If you're watching this and you think you know everything this model can do, I'm going to show you some stuff you had no idea this model was capable of. I thought this would be an interesting idea to take an actual listing of a real house that's for sale right now from Zillow. This is a house that's $5.3 million in San Diego. I wanted to see if it would pull the information from the actual Zillow page to make the flyer. So, I gave it this prompt. Create a real estate listing flyer for this modern coastal home in San Diego. And then I linked it straight up to that Zillow page. I told it to basically…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Matt Wolfe
AI News: GPT-5.6 and the new Super App are a Massive Leap!
OpenAI released GPT-5.6, a major leap forward in AI capabilities, alongside the new unified ChatGPT work app integrating code, browsing, and agent features. The week also saw significant model releases from xAI (Grock 4.5), Meta (Llama Spark 1.1), and research from Anthropic on AI reasoning patterns, establishing a new competitive landscape in AI development.
I Built A Monetizable Business With AI
The creator built a monetizable finance dashboard in one day using Hyper Agent from Airtable, which deploys a team of AI agents that continuously monitor news, track funding rounds and IPOs, analyze market sentiment, and generate daily briefings without manual intervention. The system demonstrates how specialized AI agents can work together autonomously to create a functional business product.
The ONLY AI Benchmark You Need!
A developer created "Buccy Bench," a humorous yet functional AI benchmark that tasks different language models with drawing Gary Busey as SVG code rather than images. The benchmark tracks model evolution over time while measuring performance metrics like cost, tokens, and execution speed.
GLM-5.2 - The Open Model That's As Good As Opus!
A comprehensive review of GLM-5.2, an open-weight Chinese AI model with a 1 million token context window, demonstrating its capabilities for coding, document analysis, and agentic workflows at significantly lower costs than frontier models like Claude Opus and GPT-4.5. The speaker tests various use cases including website building, Chrome extensions, game development, and data organization, concluding it's valuable for long, code-heavy, token-expensive tasks despite not universally outperforming closed-source alternatives.
Don't Fall For This AI Trap
The speaker emphasizes that power users distinguish themselves by knowing what NOT to automate with AI, rather than automating everything. They argue that AI works best for clear, straightforward tasks but struggles with nuanced, artistic work requiring consistency—using their failed YouTube thumbnail automation as an example.