In episode 3, Patrick Ferry, Rob Talbot and Wally Malesh compare how they are using voice agents, deciding when paid AI tools earn their keep, and exploring local models. They share practical experiments, including Rob's birthday bot, Wally's price-reduction reports and listing-appointment research.
Real Estate AI Nerds is a recurring conversation about what AI is actually changing in residential real estate. Patrick Ferry, Rob Talbot, Wally Malesh and Jerimiah Taylor compare notes on tools, experiments, workflows, failures, risks and practical plays—without pretending the technology is more mature than it is.
IN THIS EPISODE
Patrick Ferry — Host | Strategic Advisor, AI, Digital Authority & Real Estate Innovation
Rob Talbot — Business Builder & Systems Operator | Brokerage Leadership, Partnerships + Applied AI
Wally Malesh, MBA — Business Advisor & "Business Compounder" | Leadership, P&L, Systems + Scale
CHAPTERS
00:00 Welcome to episode 3
00:17 The ChatGPT voice commercial
03:09 Voice agents that actually do the work
06:32 From read-only to "just do it": earning AI's trust
08:13 When a leader only has ChatGPT on his phone
10:08 Wally's first demo: "Tell me about this property"
11:39 AI as a "very articulate Google search"
13:40 Anthropic's AI jobs report
14:52 From listing books to AI
17:27 The $20, $100 and $200 AI tool stack
18:57 Earn your upgrade and measure the ROI
19:35 Free vs. paid: memory and the second brain
20:37 Open source, open weight and Chinese models
21:57 Why the big brands still do the heavy lifting
23:42 Upload speed and home AI
26:54 Token costs and where compute is heading
27:52 Running local models like Qwen3
28:51 Hermes Agent: give AI its own computer
31:25 Everyday tasks vs. advanced reasoning
33:15 Automate vs. augment
36:19 Local models as an ownership strategy
38:51 Use case: Rob's 30-minute birthday bot
41:57 Use case: Wally's automated price-reduction reports
46:22 Use case: auto-researching listing appointments
47:25 Getting better vs. replacing yourself
48:46 Wrap-up
Want the tools, plays and conversations from Real Estate AI Nerds in your inbox? Join the flagship newsletter:
https://patrick-ferry.kit.com/72a1cd74f8?utm_source=real_estate_ai_nerds&utm_medium=podcast&utm_campaign=launch
This is an informal conversation about the panelists' experiences, opinions and perspectives, shared for education and discussion. Tools, features, pricing and availability change. Verify details before relying on them. Nothing here is legal, tax, privacy, compliance, financial or other professional advice; consult an appropriate professional for your situation.
TECHNICAL NOTES
19:35 — ChatGPT Free can include memory. Available memory features vary by plan, region, platform and settings.
27:18 — The space-computing discussion is speculative. Spacecraft still require engineered heat rejection; cold space does not by itself establish lower operating costs.
37:22 — Open-weight means trained model weights are available under their license. It does not necessarily mean all code or training data are open, or that safeguards are removed.
37:53 — Local models can run on infrastructure you control. Whether information stays offline depends on the full setup, including connected tools, syncing and hosting; local does not automatically mean isolated or secure.
Sources:
https://help.openai.com/en/articles/8590148-memory-in-chatgpt
https://help.openai.com/en/articles/11870455-openai-open-weight-models-gpt-oss
https://www.nasa.gov/smallsat-institute/sst-soa/thermal-control/