Why Are Public Works Projects Always Over Budget and Never On Time?

Jeremy Suard: 
A construction project is not fun for years. There is disruption to day-to-day life as a citizen when your street is blocked for 10 years. They're super expensive, super long, and they're never on time and never on budget, never. Highways, tunnels, metros, light rail systems, electric car infrastructure, power lines, new electric stations - there's a taxpayer interest that those projects finish on budget. 

Amir Mizroch:
Jeremy Suard, founder and CEO of Exodigo - it sounds like Excalibur, so I love it. Thank you for making the time. I know you're in Miami now. Your company works with transit authorities or construction developers, builders - you are in the construction industry, which is the building industry. How does building construction look from space?

Jeremy Suard:  
You can divide construction into two pieces - private, like buildings or houses, and infrastructure, financed by government. They are completely different worlds with nothing linking them, not even the companies doing the work. It's not the same in how it works, where the budget comes from, what the incentives are, or how complicated the project is. We are currently active in the infrastructure construction market. 

Amir:
So the infrastructure construction market is trains, rails, utilities like water?  

Jeremy Suard:
Yes, utilities and transportation - anything big built in the public right of way, not private areas. It's huge - highways, tunnels, metros, light rail, electric car infrastructure, new power lines, new electric stations for the switch to net zero and the need for electricity everywhere, new gas pipes replacing old ones, replacing water pipes. These projects are financed by the public sector but built on streets, in the public right of way. They're super expensive, super long, never on time or on budget. There are reasons why - politics, bureaucracy and permitting. But a big one is unexpected underground issues. Meaning you discover too late that what's there isn't what you thought, designed or contracted. Like hitting rock when you planned for sand, or finding five power lines you thought were moved. Unexpected issues underground are a major reason projects go over time and budget, and that's our focus. If you prepare a bid for sand then hit rock, the price jumps from say $500 million to $1 billion - who pays the difference? That's what we mean by unexpected.  

Amir: 
Construction always goes over budget and time with inefficiencies. It would be shocking if a big infrastructure project was on time and budget. You mentioned three problem areas, with underground issues being a big one you aim to solve. Can you walk through examples showing why projects have these underground surprises?

Jeremy Suard:
I'll give a bad story then a worse one. In Seattle, a Sound Transit project ended up 220% over budget and took three times longer than planned because of underground utilities. They did a bad survey and had to move dozens of utility lines. Moving a line means the contractor already working has to stop and call companies to get permits to move and relocate the lines, while everything halts. There was a six month delay just for the first round of newly discovered lines, happening over and over. Huge delays and extra costs through no fault of the contractor, who then charges for all their idled workers and equipment. I have an even worse story. 

Amir: 
Let's pause after that first story, which really illustrates the risk of inaccurate utility line maps leading to expensive, dangerous surprises. I want to understand how this happens on big, well-resourced projects. What kind of technology and processes do they currently use to map underground, and why isn't it good enough?

Jeremy Suard:
The technology they have simply isn't good enough yet. They're looking to locate old pipes and see if conditions are sand or rock to plan the dig. The mapping techniques aren't new - subways have been built for ages, always over budget. There are sensor tools and heavy drilling to map utilities combined to form plans. Better tech means less drilling. With poor tech you drill extensively but still get an insufficient map.

Amir:  
Who does this work? Who gathers and assesses underground data to form these utility maps? 

Jeremy Suard:
Engineering firms bring in specialized digging contractors with fast drilling trucks to find things. They use surveyor companies and all available tech tools. For big complex projects it is engineers from large consulting firms, working for project owners, who create the maps. Those maps provide critical data for bidding and contracting. But the maps have an Achilles heel - incomplete underground intel that doesn't emerge until late, forcing expensive corrections. 

Amir: 
So experienced experts use long-established methods and tools for utility mapping, but you're saying those technologies have inherent limitations, even on major projects. How much have those technologies progressed?  

Jeremy Suard:
The technology has steadily but slowly improved over time, with some startups innovating. Consultants can now get about 70-80% accuracy at best with all resources deployed - they won't reach over 80%, based on our data. We've done 60 competitive tests against the status quo, always finding 20% to 100% more lines - it depends on their tech quality and effort.  

Amir: 
And your differentiation is achieving far greater certainty? 

Jeremy Suard:  
We get 99% accuracy, about 20 times better by missing far fewer lines. It makes a huge difference finding 1% versus 20%. On well-resourced government infrastructure projects, we still find 20% more if they are really good. And those incremental lines often cause major timeline and cost overruns.  

Amir:
Let's get into your founding story. How did Exodigo get started to crack this underground mapping problem?

Jeremy Suard:
The idea came when initially looking at medical startups. I wondered why there are so many imaging devices - MRI, CT, ultrasound, x-ray - needing four machines, four doctors, four scans. The root cause was the same as underground - imaging hardware companies optimize sensors for one type, believing that solves everything. But there's no silver bullet. So we take a different approach, combining various scans into one 3D image, aided by AI instead of a doctor. We apply that combo method to create utility maps, with an array of physical sensors plus computer vision scanning records and satellite images over time. All of this disparate data flows to the cloud - 500 gigabytes per acre, completely different types needing harmonization. Our core software IP is a new AI technology, ultra low SNR signal processing, meaning we find meaning in messy data signals. We have the sensor algorithms, computer vision and public record algorithms, and algorithms fusing it all into accurate utility maps.  

Amir:
Could you explain the key algorithms through an analogy? 

Jeremy Suard:
Consider AI analyzing ultrasound images to assess a baby's health. One approach learns how doctors evaluate the images to identify concerns. The other gathers thousands of sample images to train software on normal versus worrisome patterns. We blend both - understanding the science, while amassing training data through each project scanned. The more global data the AI accrues, the smarter it gets. Our computer vision and public record algorithms are more classic AI. Then we have the fusion algorithms reconciling all the data into the best maps.  

Amir:  
How big a leap is this from traditional utility scanning? 

Jeremy Suard:
We collect 10 times the data with far more advanced software. Legacy providers use one or two sensor types with humans interpreting. Our software product removes human variability. Anyone can scan and get consistent results with the system directing field data collection, processing data in the cloud to deliver maps automatically. It's like how anyone can now take good photos without being an expert photographer.  

Amir:
Can you summarize a real-world deployment showing your impact? 

Jeremy Suard:  
We initially piloted with National Grid, finding major lines where they planned excavations after using their typical scanning methods. We changed their plans by providing the more complete underground picture on time. Now we are progressing to broader commercial deployments with National Grid.

Amir: 
Looking ahead, what are the next frontiers you aim to tackle?

Jeremy Suard:
We will expand our scans to include surrounding ground composition, like voids, sinkholes and more. And we are developing self-serve scanning devices democratizing our approach - think of buying an Exopod system on Amazon to map your own backyard. We will open our full capabilities to smaller projects while still serving major infrastructure clients. 

Amir:
Thank you Jeremy for explaining how Exodigo provides a technological leap in underground mapping to avoid dangerous and costly construction surprises. More accurate utility data benefits infrastructure building worldwide and all of our lives.