Propy CEO on AI real estate agents

Key Concepts

• AI Agent (Avery): An autonomous software entity capable of performing tasks like making phone calls and interacting with third parties.
• RAG (Retrieval-Augmented Generation): A framework that allows AI models to access external, up-to-date data to improve the accuracy and relevance of their responses.
• MCP (Model Context Protocol): A standard for connecting AI assistants to systems, databases, and tools, enabling the agent to retrieve specific information in real-time.
• Regulatory Disclosure: The ethical and legal requirement to identify an AI agent as non-human during interactions.

---

1. The Role and Functionality of "Avery"

The video introduces "Avery," an AI agent developed by Proper Title. Avery is designed to handle administrative tasks, specifically interacting with lenders to request mortgage payouts. Unlike text-based chatbots, Avery utilizes voice technology to conduct real-time, conversational interactions.

2. Operational Methodology: RAG and MCP

The core technical challenge addressed is the agent's ability to handle the unpredictable nature of lender inquiries. Because every lender has unique requirements and asks different questions, the agent requires a robust architecture to function autonomously:

• RAG Integration: Used to provide the agent with the necessary context and documentation to answer specific, variable questions on the spot.
• MCP Tooling: Acts as the bridge that allows Avery to interface with internal databases and external systems, ensuring the agent has the "context" required to complete complex workflows without human intervention.

3. Strategic Disclosure and Ethics

A significant portion of the discussion centers on the ethics of AI transparency.

• Current Strategy: The company has intentionally chosen not to disclose that Avery is an AI during initial interactions. The rationale is to ensure the agent successfully gathers the required information without the potential bias or friction that might occur if the human on the other end knew they were speaking to a machine.
• Future Outlook: The speaker acknowledges that this landscape is shifting. They anticipate that future regulations will mandate immediate disclosure, requiring agents to explicitly state their identity (e.g., "I'm an AI agent Avery and I'm with Proper Title").

4. Real-World Application

The primary use case highlighted is the automation of mortgage payout requests. By deploying a voice-enabled AI, the company aims to streamline communication with lenders, replacing manual phone calls with an automated process that can navigate the specific, varied demands of different financial institutions.

5. Synthesis and Conclusion

The transition from manual administrative tasks to AI-driven automation is currently in a "testing" phase where efficacy is prioritized over transparency. By leveraging advanced frameworks like RAG and MCP, the company has successfully created an agent capable of handling complex, real-time voice interactions. However, the long-term viability of this approach is contingent upon upcoming regulatory frameworks that will likely force a shift toward mandatory AI disclosure. The success of such agents relies heavily on their ability to access context-specific data dynamically to resolve the unique requirements of third-party stakeholders.