Procore Analytics on Databricks: Zero-Copy Open Sharing for AI-Ready Construction Data

Summary

Zero-copy sharing: Open Sharing distributes project data into Databricks with Unity Catalog governance.
Data Explorer: self-service dictionary makes shared Procore data discoverable with schemas and lineage.
Human-in-the-loop AI: reasoning and governance enable teams to analyze projects, safety and efficiency.

Procore Analytics on Databricks: Zero-Copy Open Sharing for AI-Ready Construction Data

Talk by Gustav Choto, AI Solution Specialist, Procore; Jeremy Chasen, Sr. Manager, Strategic Partnerships, Procore
AI gives unreliable answers when company data is fragmented across disconnected systems. Procore Analytics addresses that for construction by sharing project data into Databricks through zero-copy Open Sharing, formerly Delta Sharing, governed by Unity Catalog, so owners and builders get analytics-ready data without building or maintaining custom ETL pipelines. This video covers the five data challenges behind the construction data paradox and how a live self-service data dictionary makes shared data discoverable.
Procore's team covers the construction "data paradox": a survey of more than 1,200 decision makers found 46% of project time goes to the downstream impacts of information gaps. They outline five challenges (limited visibility, fragmented data, technical debt, manual cleanup, lost AI advantage) and show how the cloud connector shares data to Databricks, Power BI or Tableau without moving it, how Data Explorer acts as a live data dictionary, and an intelligence, reasoning and governance architecture with human-in-the-loop AI. Vantage Data Centers, a joint Procore and Databricks customer, shows the results, and the video closes with RAG and vision AI for safety.

Chapters

FAQs

What is the construction data paradox?

A survey of over 1,200 decision makers found 46% of project time goes to downstream impacts of information gaps. Construction data is fragmented across disconnected systems, making it hard to reason about projects.

How does Procore share data to Databricks?

Procore Analytics uses zero-copy Open Sharing (formerly Delta Sharing) to share project data into Databricks without moving it, governed by Unity Catalog, so owners and builders get analytics-ready data without custom ETL pipelines.

What is Data Explorer?

A live self-service data dictionary that makes shared Procore data discoverable, showing schemas, lineage and governance so users understand what data is available and how to use it.

How does AI help in construction?

Intelligence, reasoning and governance architecture with human-in-the-loop AI lets teams use RAG and vision AI for safety analysis, equipment tracking and decision-making on projects.

Full transcript

`00:00` Introduction

Hello everyone and welcome. This is the first time I've given a speech with people wearing headphones. This is actually pretty cool. I almost feel like I should be wearing one too. Thanks for joining us for today's session, revolutionizing construction intelligence. So over the next 20 minutes, we're going to explore how construction organizations, owners, builders, specialty contractors, and other stakeholders in this industry are transforming fragmented project data into a foundation for analytics, AI, and better decision making.
We're going to discuss the challenges that industries are facing today, how leading organizations are solving them, and why trusted data has become the foundation for the next generation of construction technology and the built world. Before we start, a quick note from our legal team. One second. oh let's see ah there it is okay oh sure that sounds yeah oh okay interesting awesome all right so i don't know if there are any speed readers out in the audience right now but this is a safe harbor statement from our company.
So before we get started, this is just a quick note from our legal team. Please don't make any purchasing decisions based on any forward-looking statements we discussed today. With that, out of the way, let's dive in. Let's see why this clicker, oh, there it is. Hi, everyone. I'm Gustav Chodo. I'm an AI solution specialist at Procore. Actually going to lose the headphones, feels a little weird. I spent the last 20 years of my career focusing on the built environment.
I started on the construction side working with some of the biggest general contractors in this industry. And actually about I think 15 or 12 years ago I actually worked on the Moscone Center expansion when I was working with WebCore builders. And I actually have spent about the last 10 years working on the tech side. So I made a transition into construction technology, which has really given me an opportunity to see firsthand how technology is transforming the built environment.
First as a builder, and now I'm helping organizations, owners, builders, specialty contractors, leverage technology, data, and AI to improve outcomes. One second, I can hear this. So I'm excited to be here today because our conversations that we're having aren't really just about AI. It's about data. and the massive amount that we're dealing with every single day in our personal lives and as companies. And more specifically, how companies, our clients, are leveraging this massive amount of data to make better decisions, have better outcomes, and ultimately seed the foundations for agentic AI.
So let's start with Procore and why this challenge matters. At Procore, our mission is simple, to improve the lives of everyone in construction, and we do this by supporting companies that build the world. And here you see a few of our clients. And our platform connects people, workflows, and data across the entire project lifecycle, from pre-construction all the way to closeout. And today, more than 17,000 customers across 150 countries use Procore to manage their projects.

`03:50` Procore: connecting construction projects, people and data

More than 3 million projects have been built on our platform, and it's growing every day. And to continue supporting this industry, Procore has invested more than $1 billion in research and development since 2021. And for owners, developers, builders, bringing everyone on a unified platform creates opportunities for better performance, better visibility, and ultimately just better business outcomes. And that's why we exist.
But as project portfolios grow, so does the challenge of managing these enormous volumes of data sets that are generated across the asset lifecycle. And the reality is that it's not getting smaller, it's only growing. So the reality facing our industry today is construction complexity is growing faster than productivity. Humans just cannot keep up with the massive amount of data. Projects are larger, buildings are smarter, stakeholders are more interconnected.
And the volume of data continues to increase every single day. Yet many of the organizations still struggle to get the right information to the right people at the right time. For owners, which we know we have a lot here at the conference, that often means reduced delayed on returns, reduced visibility, and just missed opportunities. For builders, which we also have here at the conference, it creates additional risk, schedule pressure, and margin erosion.
And the lack, the challenge isn't the lack of data. We know the data exists. It's out there. It's in all our sallowed systems. The challenge is really turning that data into intelligence. To better understand this challenge, Procur surveyed more than 1,200 construction decision makers across all our verticals. And one thing we found that was remarkably consistent. 46% of project time is spent dealing with the downstream impacts of information gaps.

`06:21` The construction data paradox: 46% of time lost to information gaps

18% is spent searching for information. And 28% is spent on rework or correcting avoidable issues. Just think about that for a moment. Organizations have more data than ever before. yet our teams are continuing to spend enormous amount of time searching for information instead of acting on it. The industry isn't suffering from a data shortage. It's struggling to generate insights from the data it already has. And at Procore, this is what we're calling the data paradox.
Having data doesn't automatically mean you have the answers. Across large capital programs and complex project portfolios, we're consistently seeing five major challenges. The first one is limited visibility. This means that our clients who have fragmented data have a lack of visibility across their asset lifecycle. The second one is fragmented data. It's trapped inside of disconnected system. You have data in SharePoint, data in your systems of record and email, in all the systems that you may be using today.

`07:27` Five data challenges: visibility, fragmentation, technical debt

Third is the technical debt. It's created by the custom integrations and the data pipelines that we're having to manage. And then the fourth is the manual cleanup and the data preparation effort that's slowing down the reporting and the decision-making. And then finally, it's a loss of the AI advantage. Because if your data isn't structured, you can't be trusted. And AI simply can't operate efficiently over unstructured data.
And this is really where the conversation shifts from collecting data to actually creating intelligence. And at Procore, to address these challenges, we develop Procore Analytics. Procore Analytics helps organizations transform construction data into analytics business-ready intelligence. So instead of building and maintaining complex integrations, teams can securely share construction data directly into the tools and platforms they already use, whether that's Databricks, a data warehouse, a lake house, or a BI platform.

`08:36` Procore Analytics: share data without custom integrations

The result is comprehensive visibility across projects and portfolios. And because data is refreshed regularly, organizations can combine construction data with operational and financial systems to make faster, more informed decisions. Let's look at how this works in practice. The process starts with project data being captured inside of Procore. using cloud connector and open sharing formerly known as delta sharing that information can be securely shared with your broader analytics ecosystem importantly this happens without maintaining custom integrations or physically moving data across environments so once connected your teams can use the data wherever they generate value whether it's in power bi and databricks Tableau, or inside a custom internal application.

`09:23` Cloud connector and zero-copy Open Sharing with Databricks

This creates a unified foundation for analytics across the enterprise and enables a true single source of truth. One challenge we hear frequently is that even when teams gain access to data, understanding the data remains difficult. And that's why we built Data Explorer. Data Explorer serves as a live self-service data dictionary. Teams can browse schemas, validate data types, understand relationships, and explore data structures before building reports or writing a SQL.

`10:22` Data Explorer: a live self-service data dictionary

Behind the scenes, we're simplifying thousands of relational tables into analytics-ready entities optimized for reporting and analysis. The result is faster onboarding, faster reporting, and faster time to value. And that's the key there. Because for owners, time is money. So once the trusted data is available, organizations can start moving into AI-powered workflows and beyond just the analytics. And to support this, we've built a construction-specific architecture based on three pillars.

`11:22` Intelligence, reasoning and governance for construction AI

Intelligence, reasoning, governance. The intelligence layer combines construction knowledge with real-time project data. The reasoning layer provides context, memory, and workflow execution capabilities. And governance ensures enterprise security permissions and human oversight remain built into every interaction. So you never have to worry about security. And together, these layers are transforming your data into a trusted operational data environment.
And we see this transformation in three distinct phases. assistance, guidance, automation, and connected. Because when we have clean, structured, domain-specific construction data indexed in a knowledge graph, it unlocks an entirely new possibility. Teams can search deeply across all of their platforms, locate and summarize information across thousands of project documents. And this becomes extremely important for our owner clients who need expertise at every turn, but are limited in their headcounts.
So our AI solution provides an expert in your back pocket that leverages your data. Your responses are always grounded in your information. And that's how we mitigate hallucination and combat what generic AI is doing currently. They can leverage AI assistance to support reporting, coordination, project execution. They can automate repetitive workflows through intelligent agents that operate with human oversight because we follow a human in the loop framework.
Because the goal isn't to replace people. The goal is helping people spend less time gathering information and more time making better decisions. Let's look at this in the real world. Vantage Data Centers is a global hyperscale provider and a joint customer of Databricks and Procore. And unlike traditional real estate organizations, Vantage measures success using metrics such as cost per megapot deployed. And because they build highly industrialized campuses around the world, waiting days for information to move through legacy ETL pipelines simply wasn't acceptable.

`13:55` Vantage Data Centers: a single source of truth on Databricks

So by connecting Procore Analytics directly with Databricks through zero copy open sharing, they established a centralized source of truth across projects and portfolios. And as their director of architecture describes it, they now have real data with integrity to support every single decision they make at every level of the business. And that visibility is allowing them to benchmark performance, evaluate investments, and deploy capital with greater confidence.

`14:45` Key takeaways for a trusted data foundation

So as we wrap up, we'd like to leave you with four key takeaways. The first one is construction complexity continues to increase, making scalable decision-making more important than ever. Second, analytics-ready data sets dramatically reduce the effort required to generate business value from construction data. Third, unified visibility transforms project information into a predictive business asset rather than a historical record.
And fourth, successful AI initiatives require a trusted data foundation. Clean, connected, domain-specific data is what makes advanced analytics, AI, and agentic workflows possible. And really that's kind of, that's the message I want to leave with you guys today. Because data is everywhere. Our projects are becoming more and more complex every single day. Even 10 years ago, or 12 years ago, when we expanded the Moscone Center, I remember how fat these drawing sets were. massive specs, contracts, emails, data everywhere.
And the amount of time I was spending to try to coordinate and figure out one decision. Every single day, I remember I had 10 decisions that I needed to make. I could probably get to five of them just because I needed to find the information first, then try to figure out what this piece of information and how it related to something else that I was facing. that I was only able to get to five, so the next day I had 15 things to deal with.
So it starts creating a snowball effect of tasks that I need to do and data that I need to orchestrate across fragmented systems. Today, if we did that project today, I could guarantee you that the sheer amount of documents would probably double, if not triple. I'm seeing the amount of documentation our clients are dealing with today just because of regulations. a lot of the data is actually being generated by AI, and we know AI is verbose, right?

`17:19` RAG, agents and vision AI for safer construction

So what we are bringing to the market now is an ability for clients to ingest their data, index it using retrieval augmented generation, and turn all their fragmented data, regardless of where it lives, into a layer of intelligence that allows them to have an expert in their back pocket. And I know that there's a lot of owners out there in the crowd here at the conference. You see them in the banners, right? Everybody builds.
Everybody needs the built environment, right? From the companies that built these conference halls to the restaurants that you're eating at, to the hotels that you're staying in, to the streets that your Ubers are taking you down. At every turn, you're seeing the built environment. And what we're attempting to do now is connect all the data so that we can build safer buildings. We can have safer construction because there are a lot of people out there that are building and their lives are at risk.
And AI can prevent that by using computer vision. We're bringing to the market a visual language model that allows you to upload videos, upload pictures, upload drawings. drawings and our reasoning model can understand construction because it's been trained and fine tuned on construction data. And this allows our customers, especially our owner customers who have reduced head counts for all their construction teams that are managing that construction, so they can ask the right question, they can predict what may happen, they can say, hey, for the last 10 Starbucks that we've built, tell us the request for information that are pushing our schedule out, that are creating safety issues, that are increasing our cost.
So then on the next 10 projects, we can rectify that and we don't have the same mistakes. We can take videos of things that are happening in the field and pictures and say, hey, tell me if there's anything unsafe. Maybe there's a manhole that's uncovered that a camera can capture and that an AI can analyze, they can alert a safety manager an hour before somebody walks over and falls through it. Right. So it's not just about money.
It's not just about time. It's about the lives that we're also saving. And I would say the time that we save as humans, because the amount of clicks that we're having to deal with in our days, in a way can be considered inhumane. So if we can reduce that amount of, I would say, non-value added activity and bring value right to where you need it in the moment that you need it, we can change this paradigm of just cognitive load and burnout that we're seeing in our industry.
I appreciate all your time, and I'm gonna be here afterwards, so feel free to come and say hi. And feel free to scan our QR code. We have an innovation summit having next, or is it this week? What? It's tomorrow. Completely virtual, so you can, If you've ever been to a virtual conference, please join us tomorrow. And excited to talk with you guys and we'll see you guys here and later today.

Learn more about the Databricks Data and AI platform.

The information provided herein is for general informational purposes only and may not reflect the most current product capabilities or configurations.