For engineer and innovator Matteo Zamparini, one the most significant productivity bottlenecks in the modern world isn’t talent scarcity or lack of ambition, it’s time wasted on repetitive tasks.
From tracing parcel maps as a data science intern to co-founding a startup that identified renewable energy opportunities for commercial property owners, his career has followed a consistent theme: build technology to automate grunt work so that human work can be used for more strategic tasks.
As the founding engineer at PropRise, a Y Combinator-backed company, he builds AI agents that work around the clock to track huge amounts of disparate real estate data and generate insights that uncover potential investment opportunities.
It all revolves around his vision of a world where AI doesn’t just assist people, but joins them into the workforce to execute long running tasks.
The Real Cost of Manual Data Analysis
In a software-assisted world, lack of digital tools is no longer the main problem. Instead, as
As a student intern, he spent months using tools like digital maps, permitting sites, statistical computing software and real estate listings websites to collect and eventually analyze data that could be used for making investment and development decisions. “I remember thinking: ‘Satellite imaging built with Lidar is cool,’” he recounts. “‘But is spending weeks on end on maps and websites trying to find the right data points really the best use of human intellect?’”
This was one of the insights that sparked the idea for Spaceflare, a startup he founded with fellow UC Irvine students after graduating in 2021. The model he built used public records and satellite imagery to identify properties that could be a good fit for solar panel installation, enabling clean energy companies and property owners to quickly identify potential opportunities for energy generation and revenue.
A process that would likely have taken weeks of manual analysis could now be done in minutes with Spaceflare.
Building AI Agents that Act and Reason—Within Reason
Today, Zamparini is the developer behind the AI agents that power
PropRise is not just another machine learning tool that automates data collection; its agents operate independently, structuring the data they find and continually evaluating the credibility and accuracy of a wide range of sources. They pursue objectives set by human operators, but can also escalate decisions back to them to ensure guardrails and oversight over significant decisions.
“We developed agentic systems that can reason, act, and even seek human feedback or approval mid-task,” he explains. “This blend of autonomy and human oversight is needed to build reliable AI workflows.”
He sees this human-in-the-loop approach as a crucial component to a future in which AI agents can take over all manual work without risking loss of control over strategic decisions.
Reliability Over Novelty: Architecting Agentic Systems for Real-World Use
In Zamparini’s eyes, the emergence of open standards like Model Context Protocol (MCP), AGNTCY, HumanLayer or the language BAML is the single most important trend in AI. To him, they are the foundational building blocks for the next generation of AI systems, and they already power his architecture models.
When software outputs feed into financial models or strategic decisions, the results they produce need to be verifiable. For the data they analyze to be comprehensive, it has to be easily accessible.
Emerging frameworks like HumanLayer, BAML, MCP, and AGNTCY are creating a baseline that can make this type of agentic ecosystem possible.
HumanLayer, for example, delivers a protocol that adds robust human‑in‑the‑loop oversight to autonomous AI agents. It intercepts high‑stakes actions and routes them to humans—through Slack, email, SMS, and other channels—for approvals, feedback, or escalation, then feeds the response back so agents can continue working safely.
MCP, is an open standard that standardizes how AI models connect to data inputs, allowing large language models (LLMs) to plug directly into diverse databases and APIs to create a more comprehensive pool of information and agent-to-agent collaboration.
AGNTCY is an open ecosystem for agent-to-agent interoperability that takes the concept one step further by defining a shared identity layer and interaction model for AI agents, enabling tools made by different developers to share information securely and predictably across platforms. AGNTCY’s goal is to create an “Internet of Agents,” similar to the shared data protocols in the 1990s that drove global adoption of the internet.
Together, these frameworks enable AI agents that are not only powerful, but deployable, testable, and maintainable.
Building a Template for the Future of Work
Shared standards and languages are key to developing AI which can reliably automate mundane work. And they apply far beyond commercial real estate; at their best, they enable AI agents and agentic systems to transform how professionals and experts work across industries.
“With AI assistance, we can essentially free people’s time,” Zamparini concludes. When we can truly trust automated work and recommendations to be comprehensive and accurate, these systems can do more than just improve existing workflows, they can make space for new industries and innovation.
For Zamparini, the future of AI is not a better digital assistant, it’s a system that lets humans build new things faster, advances current industries, and even spawns new ones. Rather than a productivity upgrade, it’s redirecting human intellect and creativity.
AI agents remove the noise, clearing the clutter so builders can move faster, think deeper, and dream bigger.