Insyte

Inspiration

When we talked with CBRE reps and looked at how property managers and clients currently review building data, we noticed the information is all there, but it’s scattered, hard to compare, and often communicated differently depending on who is looking at it. That leads to slow decision-making and misalignment.

So we wanted to create one shared dashboard where both sides see the same insights, at the right level of detail, and in language that actually makes sense to them. Our goal was to make property data quicker to understand, easier to trust, and more actionable for everyone involved.

What it does

Our platform is a dual-view dashboard powered by agentic AI that automatically summarizes key insights across property data, including maintenance timelines, pricing trends, and physical building details. It tailors information to each audience so both clients and property managers get what matters most to them. Clients can interact with a conversational chatbot to ask questions and explore properties through an intuitive map that highlights investment-relevant details. Property managers, on the other hand, get AI-driven data visualizations and a map view connected to deeper operational metrics, allowing them to prioritize issues, understand performance trends, and make informed decisions faster. The result is a single system where both sides share clarity, stay aligned, and move forward with confidence.

How we built it

We built our Next JS Platform using a DynamoDB database populated with curated commercial property data sourced from Smarty’s Address Verification API for the Dallas area. From there, we expanded the dataset with additional fields that property managers typically track, based on interviews and research into their workflows. We processed and uploaded this data into DynamoDB through AWS Lambda functions, allowing our dashboard to dynamically query and display property information across our interactive Leaflet maps and Flask. For our agentic AI layer, we used Anthropic’s Claude models to interpret documents like contracts and maintenance logs, summarize key insights, and generate tailored responses for each user’s view in the dashboard. Additionally, we have an AI chatbot that can answer questions the client has about real estate trends. Together, this created a seamless pipeline from raw property data to meaningful, conversational insights.

Challenges we ran into

The biggest challenge was the learning curve. We were working with technologies we hadn’t used before, including LangGraph, agentic AI patterns, Leaflet, and Next.js, so we had to build and learn at the same time. On top of that, we ran into strict time and budget limits, especially since many commercial property datasets and AI model tiers are expensive, which forced us to curate and generate parts of our dataset manually. Integrating DynamoDB, AWS Lambda, and the dashboard UI while keeping everything stable under those constraints required constant iteration, but it pushed us to be resourceful and intentional with every feature we shipped.

Accomplishments that we're proud of

We are proud that we were able to build a fully working dual dashboard using technologies that were completely new to us. None of us had experience in real estate or tools like LangGraph, Leaflet, DynamoDB, or agentic AI workflows, but we learned quickly and turned those skills into something functional in a short amount of time. We also developed a stronger understanding of both the client and property manager perspectives, and how AI can actually support clearer communication between them. Seeing everything come together, from the data pipeline to the interactive map to the AI insights, was a huge milestone for our team.

What we learned

We learned how to work with an entirely new tech stack, from setting up DynamoDB and AWS Lambda pipelines to building interactive maps in Leaflet and implementing agentic AI with LangGraph and Claude. We also gained a deeper understanding of how property managers and clients actually use data and how different their priorities can be. Most importantly, we learned how to design AI features that don’t just generate information, but communicate it clearly and in context. This project pushed us to think not only about building a tool, but building one that is genuinely usable and valuable.

What's next for Insyte

Next, we want to scale Insyte beyond a single region and support larger, real-time datasets. We also plan to expand our agentic AI features to handle more complex documents and multi-step workflows, such as comparing leases across multiple properties or predicting long-term maintenance needs. Another major goal is to introduce collaboration tools, allowing multiple property managers and clients to communicate, share insights, and make decisions together directly inside the platform. In the future, we envision Insyte becoming a shared workspace where real estate teams can interact with their data, their buildings, and each other, all in one place.