How Cavanagh and Palantir Are Building Construction’s OS for the 21st Century
Editor’s Note: This blog post is co-authored by Joe Patrois, AI Lead, Thomas Cavanagh Construction, and Jesse Velay-Vitow, Deployment Strategist, Palantir, reflecting insights from both sides of this transformative partnership.
Introduction
Construction is one of the world’s oldest industries and a prerequisite for nearly all others, especially manufacturing, infrastructure, energy, and defense. Yet, due in part to its physical and decentralized nature, the adoption of new technologies has been slow and fragmented. Many companies still operate on paper, relying on spreadsheets and siloed systems that make efficient operations a daily challenge.
With recent breakthroughs in large language models, a new opportunity has emerged to reimagine the old ways of working entirely, not simply digitize them. The challenge is no longer how to automate paperwork but how to rebuild the operating system of construction itself.
These shifts pose a fundamental question for the industry: what would a construction company look like if it were redesigned from first principles, with artificial intelligence interwoven across all foundational operations?
Thomas Cavanagh Construction and Palantir have partnered to answer this question. In this blog post, we share our ongoing journey building a modern operating system for construction, including the discoveries, the design principles, and the blueprint for how an entire industry can evolve from fragmented systems to connected intelligence.
The State of Cavanagh Pre-Palantir
Cavanagh’s experience mirrors the broader reality of the construction industry. Decades of growth, layered systems, and incremental fixes had produced an organization that worked hard but had begun to outgrow the tools designed for an earlier stage of the business. The family-owned business built on direct communication and trust had grown into a complex network of processes and tools that no longer scaled with the business. Each new form, spreadsheet, or software system solved an immediate need, yet collectively they created friction rather than flow.
Processes were born out of necessity rather than design. Information existed everywhere but moved slowly, filtered through manual entry and reconciliation. People carried the burden of connection. Foremen coded costs by hand. Payroll waited on the field. HR waited on operations. IT waited on HR. Each step made sense in isolation but together formed a bottleneck that constrained the entire organization.
Frustration became embedded in the workflow itself. The field felt overwhelmed by administrative expectations without the tools or visibility to meet them. Office teams grew frustrated when data was incomplete or late. Trucks crossing scales could not be optimized. At the expense of morale, time, and money.
These were not failures of people — in fact, the strong company culture and committed teams are what power the organization’s success — but symptoms of systems designed around control instead of clarity. Software began to shape behavior in unproductive ways, reinforcing blame between departments instead of enabling shared understanding. Accountability became fragmented, and success required human endurance rather than design.
Furthermore, a deeper issue ran beneath the surface: an over-reliance on data that did not serve decision-making. Much of what was being captured existed only because other systems demanded it, while the information that truly mattered — the state of the work, the readiness of a crew, the movement of material — was not easily accessible.
The challenge became not to collect more data or add more systems, but to reconnect the existing pieces into a single, coherent whole — one where data serves reality, not the other way around.
Recognizing the limits of its legacy tools wasn’t a crisis, but a straightforward realization. After 70 years of steady growth, Cavanagh’s needs had evolved, and the next step was clear. As it always has, the organization embraced its trusted “all in” approach and began the work of raising the bar for the industry.
Initial Engagement — From Discovery to Design
Our discovery effort began with a clear ambition to build a single operating system capable of running an entire construction company from quarrying rock to homebuilding.
The partnership opened with a Palantir Foundry bootcamp focused on fleet and job costing. These early exercises revealed a simple truth: every workflow depends on high-quality data captured at the source. Accurate data existed in pockets, but not consistently across the company. Following the bootcamp, Cavanagh ran an internal deep dive, mapping how work was actually performed rather than how it was documented. Teams examined what information was truly needed to make decisions and what could be eliminated entirely.
At the end of this period, Cavanagh formally engaged Palantir in a multi-year partnership running through 2030. The build would begin in earnest in January 2025. The first phase focused on building the foundation by capturing accurate operational data so every downstream workflow could run cleanly.
Our Method
Palantir Forward Deployed Engineers worked side-by-side with Cavanagh subject-matter experts. Together they decomposed operations, hiring, fleet, and back-office processes step by step. The north star was simple: the field is the core of the business, and everything else should make field work faster, clearer, and easier. If operators could not work because they were blocked by a cost code, it meant the system was asking the wrong person for the wrong information. Discovery work focused on moving responsibility to the right place and removing steps that did not serve execution.
Three key questions drove the process:
- How is scheduling actually done today, and what does the crew need to start work on time?
- How are job costs truly calculated in practice, and where are the blind spots?
- Who is responsible for each piece of equipment at any moment, and how is that responsibility recorded?
The answers to these questions didn’t come from strategy decks or executive briefings. They emerged through whiteboard sessions, field observations, and the collective effort of questioning every assumption about how construction should operate. Each realization became a building block of what would become our core operating system.
By asking these questions from a field-first perspective, incentives were realigned. Fleet and field teams began working from the same picture when assigning trucks. Project managers and finance shared a single view of how equipment time and costs should land. The goal shifted from collecting data for systems to capturing the facts of work as it happens.
Five Discoveries that Shaped our Solution
1. Tame Complexity with Unified, Live Data
As organizations grow, coordination becomes exponentially harder. What once relied on direct communication becomes mired in layers of interaction. Ten people create 45 coordination lines; 100 create nearly 5,000. In daily operations, this means dispatchers spend more time aligning with each other than with the field, while onboarding requires multiple departments to complete dependent tasks.
Instead of layering more systems, we must unite them in a single, shared source of truth. Everyone — from the field to finance — should work from the same live data to create natural coordination and enable scale without friction.
2. Automate and Anchor Accountability to Mirror Reality
Every asset and action must tie back to a responsible person. Traditional systems separated accountability from action by assigning equipment to jobs or asking drivers to code costs. Operators should be responsible for the assets they control; foremen should be responsible for how those assets are used.
The system should automatically record responsibility through daily workflows (e.g., when assets are checked out, inspected, reassigned, or returned), creating a living map of accountability. This eliminates confusion and ensures that responsibility mirrors real-world operations.
3. Design for Natural Workflows and Purposeful Friction
Most systems make the right thing difficult and the wrong thing easy. The team recognized this backward logic across daily operations and realized it was necessary to flip the script. The right action needs to be effortless, the wrong one inconvenient. Purposeful friction should be introduced only when a decision carries risk, cost, or accountability.
This means a foreman cannot close a shift without confirming quantities, but a driver is never stopped at a scale for missing a cost code. Workflows should request only the information truly needed, and if something can be determined automatically, the system does not ask again. This design eliminates wasted effort and aligns behavior with intent.
4. Field-Driven Data, Not Data-Driven Field
In Cavanagh’s previous systems, like most construction firms, field staff were responsible for feeding financial and project systems that gave little back to them. The people building were also doing data entry.
We needed to reverse this relationship. When a foreman plans a shift, dispatches trucks, or confirms completed work, they should simply be performing their job, and the system captures the data automatically. Project teams and accountants receive real-time information without additional effort from the field. Operations drive production, and systems exist to serve that reality.
5. Challenge Legacy Processes and Prioritize First Principles
Processes often persist because “that’s how it has always been done.” Through field observations and process mapping, the team found countless examples: reports no one read, approvals that added no value, data captured only because another system demanded it.
We made a commitment to challenge legacy processes and prioritize first principles. Ask what the work is, what prevents it from being done efficiently, and what information truly needs to be passed along. The willingness to question and rebuild, rather than patch and adjust, became one of the defining cultural shifts of the initiative.
With these principles in hand, the team had a blueprint. Now we needed to build it.
The Build — Total Operations Management
Four domains were identified as the foundational objects of Cavanagh’s ontology: Contracts, Labor, Equipment, and Materials. These serve as the backbone of the system, ensuring that every piece of information about work, cost, and production ties back to a single, consistent structure.
This became Total Operations Management (TOM) — a unified operating system where every major function of Cavanagh runs on a single platform, Foundry.
With this shift, Cavanagh has entered a new operating era. What began as discovery has evolved into a live, production-ready system that now runs a growing share of daily operations. Dispatch, Trucking, Site 360, and Cavanagh Connect all flow through TOM, connecting people, equipment, and work in real time.
For decades, construction relied on fragmented tools and disconnected databases that required constant reconciliation. TOM replaces this patchwork with a single, living architecture. Every object — truck, activity, crew, or shift — exists once in one ontology shared across every workflow. The same data powers dispatch, timecards, rentals, procurement, and payroll. The result is not another layer of software but a unified operating core that reflects reality as it happens. The legacy ERP is being steadily removed from operations. By the end of the year, it will remain only as a financial ledger while TOM becomes the system that runs the business.
Because TOM is built on a clean ontology, development now moves with remarkable speed. Workflows that once took months to integrate can now be built in weeks. Each new tool plugs directly into the same backbone without disrupting others. The speed of development has pushed the organization to its limits on change management rather than technology. The challenge now is helping people adapt to how quickly the system evolves, not waiting for the tools to catch up.
Cavanagh’s success comes from a single, uncompromising decision: to go all in. Rather than building pilots or partial integrations, the company committed to one platform and one source of truth. This removed duplication and forced clarity in design and decision-making. It aligned leadership, development, and field teams around the shared objective of building a modern operating system for construction rather than another collection of disconnected applications.
A New Operating Model for Construction
Technology alone does not change an organization. What TOM has done is return every role to its proper focus. Dispatchers dispatch. Foremen build. Accountants account. Data now moves automatically between them without extra effort or translation. Field staff see what they need to execute, while office teams see the financial reality that work creates. The burden of administration has been replaced with clarity of responsibility. Operations now happen in real time rather than after the fact.
The next phase is already underway. As the ERP is fully removed from daily operations, the company will operate entirely within TOM, a live environment where contracts, resources, and costs are continuously connected. This foundation enables the next leap: reasoning directly on structured and unstructured data. Large language models will not simply summarize dashboards but will interact with operational truth and take autonomous action when appropriate, allowing teams to offload manual tasks with confidence.
Cavanagh is building something rare in construction: a fully integrated operating system that connects every person, process, and decision in real time. The result is a business that moves at the speed of its own work, guided by live data instead of delayed reports. What began as an experiment in Foundry and Palantir AIP has become the foundation for how Cavanagh builds.
The same ontology and workflows that power Cavanagh can now be scaled across organizations of any size. Because TOM connects every domain through a shared structure, it enables replication without re-engineering. This creates a true operating model for construction — a framework capable of scaling horizontally across projects and vertically across divisions, suppliers, and partners.
The transformation underway at Cavanagh is more than a technological upgrade; it is a reimagining of what construction operations can be. By turning data into a natural byproduct of the work itself, Cavanagh is breaking free from the long-standing constraints of fragmented systems and delayed visibility. The TOM model replaces control with clarity, allowing efficiency to emerge naturally and empowering teams to focus on building rather than bureaucracy.
The core ontology powering Cavanagh’s dispatch, trucking, and site management is adaptable to any contractor or supplier facing similar operational challenges. By connecting work, resources, and costs in real time, organizations can move away from manual cost coding and reconciliation toward activity-based costing, where financial truth is captured as work happens. This approach yields immediate insight while also driving adoption by putting field needs first — making the record of the work indistinguishable from the work itself and allowing technology to fade into the background.
With a unified digital foundation, Cavanagh has achieved something most construction companies never reach: the ability to scale without friction. The same ontology that connects today’s projects can support ten times the workload without adding complexity. Growth no longer multiplies communication lines or administrative burden; it amplifies capacity and coordination.
Ultimately, this signals a future where technology amplifies craftsmanship rather than replacing it, where data serves reality, and where building becomes a clear, collaborative, and continuous process. The lessons from Cavanagh offer a blueprint for the industry, proving that with the right systems, construction can move beyond its legacy limitations and realize its full potential.
The partnership between Thomas Cavanagh Construction and Palantir continues to expand, setting a new precedent for how industry and technology can co-create real operational transformation. Together, we are building the modern operating system for construction — an intelligent, unified foundation that will define the next generation of infrastructure, housing, and civil works.
“We’ve gone all in so much so that every other software must justify its existence. And so far they haven’t been able to. 97% of our employees use Foundry every day. Foundry is our operating system.”
Revolutionizing Construction was originally published in Palantir Blog on Medium, where people are continuing the conversation by highlighting and responding to this story.