Construction projects like Hudson Yards, are massive, multi-billion $, decade-long initiatives, employing hundreds of companies and thousands of people. Delays due to factors like site conditions, materials availability, supplier churn, labor and equipment issues, and new regulatory requirements create delays, design changes and scope pivots – all very expensive, resulting in significant cost and schedule overruns. The Hudson Yards project in New York City was impacted by a shortage of steel and other building materials due to market demand, tariffs and production issues. This led to billions of $ of cost overruns and delays in project completion. Other examples include the Big Dig project in Boston, the Burj Khalifa in Dubai and the Bay Bridge Extension project in San Francisco.
Semiconductors and lean manufacturing philosophies factor in supply chains, process && regulatory constraints, and manufacturing cost at every step of the design phase. DFM (Design for Manufacturing) and DFR (Design for Reliability) are embedded into the entire design activity, typically through a gate review process which breaks up the design activity into phases. Moving from one phase to the next requires a gate review which ensures that relevant issues in the volume manufacturing phase are considered while the design is still malleable, and can adjust to manufacturing constraints.
Civil engineering projects traditionally do not follow this discipline – partly because each project is unique and has its own specific requirements. Unlike semiconductor and other types of manufacturing which focuses on producing millions of identical units, construction is essentially a one-off project, with a final outcome that ideally meets the original cost and schedule requirements. Companies engaging in manufacturing also rely on prior experience and knowledge as they iterate towards the next generation design. The use of AI techniques to capture past knowledge (for example, aerodynamic design on a sports car) to launch new models is an exploding trend. Construction projects are also starting to leverage this trend, during the bidding, project planning, architecting, engineering design and construction phases.
Krane – Bringing Predictability to Construction Supply Chains
Founded in 2022 by CEO Eshan Jayamane, the company focuses on using AI tools to evaluate suppliers, supply schedule risks and procurement management. Mr. Jayamane comes from a family of construction contractors in Sri Lanka. Having pursued graduate education at Berkeley, followed by extensive stints in large construction projects (industrial, energy infrastructure, data centers, hospitals), he realized that procurement and supply chains spanning thousands of items and suppliers needed intelligent solutions for supply chain management during the design and construction stage.
Krane’s solutions bring visibility via a unified, real-time platform to manage submittals, lead times, deliveries, and purchase orders that are updated in real time. This helps project managers to factor in critical supply chain issues early in the design phase, and manage delays as construction proceeds. The company recently raised a $9M funding round.
Krane’s provides its customers and their suppliers with AI-based agents which integrate with leading construction platforms and accounting systems for seamless project collaboration and efficiency. The AI agents train by absorbing data from past construction projects and using it to analyze data from hundreds of sub-contractors and suppliers. This is integrated into a unified dashboard for the Program Management Office (PMO) and prime contractor to manage. There are six such agents:
The agents help Krane’s customers through a 3-step process:
Construction procurement strategy: customer uploads drawings and specifications. Milo, Arlo and Lana create submittal and procurement logs that satisfy the schedule, and prioritize critical path items
Real time tracking of construction procurement: Theo, Chase and Lana query and receive updates from subcontractor and suppliers
Control Material Deliveries, Invoice Reconciliation: Rio and Theo track delivery schedules, confirm material arrivals, and consolidate receipts to streamline invoice reconciliation and payment processing.
These agents transform traditionally disconnected systems into an easy to view dashboard that exhibits various real-time aspects of the construction process.
According to Mr. Jayamane, “Historically, construction has treated every project like a one-off effort, but that mindset is starting to change. Large contractors and developers are now thinking much more like enterprises, consolidating procurement and looking for systems that give them visibility across the entire supply chain. Our platform captures live data about how materials are moving through the construction supply chain based on a portfolio of $17B worth of projects. We give the industry data that has never been available at scale and AI-powered workflows that manage the chaos of supply chains.”
UCSF Health Services
One of the construction projects that Krane is working on is the Helen Diller Hospital Project in San Francisco, which started in April 2024. It is a 900,000-GSF, 15-story healthcare facility, with new inpatient towers, expanded surgical and imaging services, seismic retrofits and phased renovations on a constrained urban campus. The initiative is driven by the University of California at San Francisco (UCSF) Health Services. UCSF is part of the State of California and UCSF Health Services is one of its divisions, with a major presence in healthcare delivery in California. It is a non-profit, with revenues of ~$9B, 36,000 employees and ~3.5M patients served in 2025. Additionally, it receives significant philanthropic support. Construction on the Helen Diller project is expected to be complete in 2029. Post pre-occupancy testing, it plans to accept patients in 2030.
James Pease is Vice President, Health Major Capital, UCSF Real Estate, and manages a multi-billion dollar construction budget. He has been in this role for the past 7 years, prior to which he was with Sutter Health where he pioneered lean techniques in construction management. The Helen Diller project is a $4.5B effort, and will accelerate UCSF’s thrust into regional health services.
According to Mr. Pease, the challenges in construction of a health care facility in a major urban area include:
- Designs must satisfy strict earthquake codes, similar to those for nuclear facilities. It takes ~1.5 years to get such plans approved.
- Construction must follow exactly what is on the approved drawings. Variances from approved designs and Bill of Materials (BOM) cannot occur without review and re-permitting. Deviations due to materials shortages or supplier delays cannot easily be overcome by using alternate channels.
- Key suppliers of construction material need to be audited for quality control and delivery schedule assurance
- On-site construction is difficult due to limited space and parking. Extensive use of modular construction and off-site fabrication is required. Precise coordination across materials, vendors and delivery vehicles to ensure safe, smooth and predictable delivery and on-site assembly is imperative.
UCSF Health’s PMO manages multiple players through use of the Integrated Project Delivery (IPD) method which ensures that the owner, architect, and contractor perform their work under a single multi-party contract. It promotes financial incentives through a profit sharing arrangement which rewards stakeholders if they meet or exceed overall cost and schedule requirements. IPD promotes collaboration during the design-to-construction journey by minimizing waste, eliminating traditional siloed conflicts, and promoting cooperation across the various players. ~300 employees across these different companies work in a collocated office space to promote cross-company collaboration.
Another initiative is to use AI tools like Krane’s (which started working on this project a year ago) to provide management of different documents and schedules, analyze drawings for errors and perform quality checks, supply visibility into the supply chain and BOM delivery status, and coordinate deliveries. Currently, these tasks are done by humans – time consuming, costly, difficult to document, not easily accessible, and prone to error.
Krane’s tools also ensure that in a limited parking and storage environment, delivery logistics are optimally managed. In addition, they also monitor megatrends (political, competitive, economic) to flag risks in supply of critical materials. Mr. Pease believes that in the future, AI tools could learn from past construction experiences to develop an entire design plan that folds in construction constraints, similar to what DFM does today in areas of high volume manufacturing.
Mr. Pease has this to say about the the involvement of Krane for the current project: “Krane provides supply chain visibility, critical on complex projects where different materials are coming from different countries. Traditional models are highly manual, and require checking lead-times on thousands of items through telephone calls. Krane automates this task, and more importantly, they help us predict problems before vendors even tell us they have one. We don’t want to find out a year out that switchgear delivery is two years out. The earlier we know, the more time we have to act.”
Boldt – BUILDING A BOLD FUTURE
Boldt is the lead general contractor (GC) on the UCSF project. Based in Appleton, Wisconsin, the company was started in 1889 by a German immigrant who set up shop to provide carpentry services. Over the next 135 years and 4 generations of family leadership, the company has established itself as an international powerhouse in industrial, education, data center and healthcare construction. Today, it is an employee owned organization with > 1000 employees annual revenues in the $1B range, and offers architecture, design and construction management services.
William Lichtig is the Chief Innovation Officer at Boldt, and has been with the company for 15 years. He describes himself as a “lawyer by training and innovator at heart”. During his lawyer days, one of his clients was Sutter Health, a major healthcare provider in Northern California (a non-profit organization). He worked with Sutter Health and James Pease (now at UCSF Health) to develop IPD and lean construction techniques and interacted with Boldt which performed as a GC on that project.
Mr. Lichtig moved to a second career at Boldt (innovator at heart phase), leading the company’s Integrated Lean Project Delivery® initiative (ILPD), essentially a combination of PLD and lean construction (which focuses on Target Value Design or TVD) , a framework with strict cost and schedule constraints for construction projects at the start of the design phase. TVD was a technique developed by Sutter Health to model the Toyota Production System (TPS) whose objective is to eliminate waste and shorten lead times at every step of the manufacturing process to deliver vehicles to customers at low cost, and high quality. Apart from reducing process times and waste, it also pioneers the concept of JIT (Just in Time) from a supply chain and inventory perspective, and ensuring proactive review and selection of suppliers at every step of the design phase.
TPS was developed and perfected over a 25 year period from 1950-1975, as Japan emerged bankrupt from WW2. It took another 30 years for Sutter Health to embrace lean principles in construction. According to Mr. Lichtig, the reason for this delay are:
- Construction management has traditionally been a low tech endeavor (as opposed to construction machines which are incredibly high tech and have pioneered the use of AI and sensors).
- Prior to 1950, the GC handled the entire project once the design phase was complete. Project management was simpler.
- Post WW2, this model broke down, with the Scope of Work (SoW) distributed between multiple GCs and fragmented smaller firms that specialized in certain aspects of construction. Managing this became a problem, with significant cost and schedule overruns.
- Just like TPS at the end of WW2, necessity drove the need for innovation at Sutter Health (a non-profit with no financial buffer to absorb cost and schedule overruns).
AI is the next frontier for optimizing lean construction. Krane agents train on past records and knowledge from Boldt’s legacy projects, and provide an assessment of procurement and suppliers during the design phase. During construction, these agents integrate this knowledge into the workflow. It turns out that one of the companies that Krane CEO Eshan Jayamane worked at prior to starting Krane was Boldt. According to Mr. Lichtig, “Our thesis has always been that you need systems and processes to reintegrate fragmented designers and builders who specialize on certain aspects of a large, complex project. The right information rarely reaches the right person at the right time. Eshan understood this from his time at Boldt, and has focused Krane across the entire lifecycle–procurement, execution, and material management, turning material managers into air traffic controllers who can track everything from fabrication to delivery. Our vision is not just to access data, but to integrate our own knowledge and lessons learned into the Krane platform.”
AI and Risk Management in Large Construction Projects
At it’s heart, all of the lean construction, program management and AI tools described above manage risks – global political conditions, supply shocks,, weather, permits, environmental codes, design errors, etc. Risk management is a core element of a PMO office, and used to identify different buckets of risks at the initial design phase of a project. Traditional approaches were sequential – owner selects the architect and design firm, the design is completed, bids are invited from contractors, after which construction begins. This approach fails to address the risks in the project, leading to costly re-designs, rework and schedule impacts. Newer techniques create a RBS (Risk Breakdown Structure) when the project is in the early design phase (5-10% complete), and constantly integrates risk management solutions in the design and construction phases.
Dr. Chaphalkar is a risk management expert with more than 30 years of experience in civil and infrastructure projects. He heads Pradeep Consulting Services, Inc., which specializes in project controls and risk management. Having spent six years as a Federal Transit Administration Project Management Oversight Consultant, supporting several multi-billion-dollar transportation projects across the country, he started his tenure as a consulting Enterprise Risk Manager for LAX’s $30 billion Capital Improvement Program (CIP) in 2017. The CIP is managed by LAWA (Los Angeles World Airport) and includes SkyLink, LAX’s elevated train that will provide efficient access to terminals, parking, pickup and drop-off curbs, the LAX Rental Car Center, LA Metro’s light rail and municipal bus systems. Upgrades to existing terminals and access roadways, and construction of new terminals are part of the CIP.
Slated to become operational in 2023, the SkyLink ~$2.0B project has been delayed by 3 years (~$1B cost overrun) and is currently undergoing 60 days continuous operational and safety testing. The delay has been due to disputes between LAWA and the builder consortium (LINXS), and a lack of a robust dispute resolution process. There were also disputes between contractors on maintenance and repair systems which delayed track testing.
LAX operates in a challenging environment. One of the busiest airports in the world, it occupies 10X lower area than other comparable airports, and is located near a quiet, residential area of Los Angeles. Any LAWA project has to run a gauntlet of environmental permitting issues, which create risks to airport operations, traffic management, right of way acquisition and laydown areas.
According to Dr. Chaphalkar, other key risks in large public projects are:
- Managing community expectations and public relations are critical given quality of life considerations during the construction and operational phase. Pollution, noise and traffic need to be managed.
- Securing materials, skilled labor and equipment at the right time and location
Risk management has been a highly manual process, that relied on a committee of experts, who conducted a series of workshops between critical stakeholders to uncover and document risks and risk management options. This leads to a RBS that is tracked and modified as the project moves along. Going forward, Dr. Chaphalkar “sees great potential for AI tools to automate project risk registers and provide an accurate forecast of the project costs and schedule. Historical data, for example, from the LAWA project can be used to train these tools for future projects, and create the ability to adjust schedules dynamically as risks are managed.”
Manufacturing industries that expect to make millions of identical units (cars, semiconductor chips, televisions, smart phones, etc.) ensure that design, manufacturing and market drivers are considered at every stage of the journey from conceptual design to high volume manufacturing. Construction project management which focuses on delivering one complex project over a 5-10 year time period is starting to function in a similar way with the advent of lean construction, integrated project delivery and risk management methodologies. The use of AI agents to learn from diverse sources of data to automate various aspects of construction project management is on the rise, with positive impact on schedule, costs and visibility.
