Features
Industries
Who we serve
Resources
Sign in
Get Started - $500/month
No commitment. No implementation fee. Unlimited users. Cancel anytime
Greg Smith brings a Moneyball mindset to what has traditionally been a biologically driven field — and the results are staggering. Lower costs. Zero vegetation-related outages on managed systems in over a decade. A 10-year cost trajectory that's declining, not escalating, even in a high-inflation environment. And a forward-looking model that projects costs 60 times better than the previous decade.
Greg has more than 20 years of experience across manufacturing, supply chain, and electric utility operations. Over the past decade, he's led everything from fleet and supply chain to transmission and vegetation programs under FAC 003. For the last five years, he's been all-in on vegetation management, and he's developed an approach that challenges nearly every assumption the industry has operated on for decades.
I've known Greg for close to a decade. He's one of the sharpest operational minds in the utility space, and this conversation — covering AI, LIDAR, condition-based planning, field data collection, and the future of veg management — is one of the most important episodes we've done for anyone who cares about grid reliability, ratepayer costs, and the future of utility operations.
I've always believed there are three foundational aspects of any utility: asset inspection, vegetation management, and storm response. Greg agrees, and he emphasizes that all three are deeply interconnected. How well you manage vegetation directly impacts how many outages you have during storms, which directly impacts reliability metrics, which directly impacts what ratepayers pay.
The numbers tell the story. Utilities have invested approximately $1.3 trillion since 2014 to improve the grid, with another trillion projected between now and 2029. Yet reliability metrics across the country have barely budged. Storm damage alone costs Americans roughly $150 billion a year. Not all of that is attributable to vegetation, but a substantial portion is — and the industry knows it.
The question Greg is answering isn't whether veg management matters. It's whether the way the industry has been doing it for decades is the right approach. His answer: it's not.
The traditional approach to vegetation management is cycle-based: pick a time interval — say, every four years — and address vegetation issues across an entire right-of-way corridor during that cycle. It sounds logical. It's organized. It's predictable.
It's also enormously wasteful.
Greg uses a simple analogy. Imagine only going into your backyard once every four years to deal with vegetation. You'd have foundation issues, sidewalk damage, deck problems, overgrown flower beds — and the cost to restore everything would be massive. Your spouse would probably leave you.
The same principle applies to utility rights-of-way. Over 108 miles of transmission line, vegetation profiles aren't consistent. Some areas have dense forests with fast-growing species. Others have sparse, low-growth landscapes. But in a fixed-cycle approach, you're sending the same crew with the same equipment across the entire corridor, regardless of what's actually there.
As Greg puts it: sometimes you're loaded for bear to shoot a squirrel. Other times you're loaded for squirrel and you've got a bear.
The result is chronic overspending in some areas and underspending in others — and the underspending creates future overspending, because the problems you didn't address grow larger, denser, and more expensive to manage. Time drives size and density. Size and density drive cost. The longer you wait, the more expensive it gets.
Greg's approach flips the model. Instead of treating an entire system uniformly on a fixed cycle, he uses real-time data to identify where the actual threats are and targets those specific areas with appropriately sized responses.
Here's a concrete example from his program. Last year, his team applied herbicide across 262 miles of right-of-way. On average, it was about 10 miles per day, using 260 gallons total. Spread evenly, that doesn't sound like a big problem.
But the data told a different story. 96 of those 260 gallons were concentrated in just 12 miles. Two other stretches accounted for 20 and 40 gallons in similar distances. The rest of the 262 miles required very little treatment.
Without data, you'd never know that. You'd treat the entire system the same way and miss the concentrations entirely — or you'd discover them after the fact and have to send crews back, doubling your cost.
With data, Greg restructured the following year's plan. The 35-36 miles with elevated indicators got an elevated response. The remaining miles got appropriately light treatment at very low cost. The crew does more targeted work, the utility spends less money, and the vegetation threats are actually addressed.
This is where Greg's approach gets genuinely revolutionary. The industry standard cost for veg management is $5,000 to $8,000 per mile according to KPMG data. Greg's program is achieving costs as low as $6 per acre in low-density areas and targeting $50 per acre in dense forest environments. The difference is staggering.
The key insight is what Greg calls managing "below the grass." LIDAR — the industry's go-to technology for vegetation assessment — doesn't see a tree until it is a tree. Brush under five inches or below 12 feet doesn't register. But that brush is the future forest that will eventually cost $5,000-$8,000 per mile to manage.
Greg shared an example from a recent site visit near College Station, Texas. Six-month-old trees were already 12 feet tall — growing from dormant root structures that produce fast-growing species. If left on a traditional four-year cycle, those become 25-foot trees requiring expensive removal crews. But right now, a targeted basal application can handle them for $70 per acre.
The math is simple: $70 per acre now versus $5,000-$8,000 per mile later. And the window to act is narrow. Every month you wait, the cost escalates.
Greg views LIDAR as a great starting point and an important long-term verification tool, but not the silver bullet the industry sometimes treats it as. LIDAR is expensive per mile, per structure, per span. It captures a snapshot in time — typically once every four to six years for most utilities, with California being the exception at annual flights. And it misses everything below its detection threshold.
The real value of LIDAR, in Greg's framework, is as the initial prioritization layer. LIDAR identifies your biggest threats. Then you go to those threats with appropriately sized crews and field data collection tools to manage the transition — tracking size and density changes in real time, not waiting four years for the next LIDAR flight to tell you what happened.
Over time, as you collect more field data between LIDAR cycles, the LIDAR itself becomes less critical for operational decisions. But the combination of periodic LIDAR snapshots with continuous field data creates an increasingly intelligent dataset — which is exactly where AI comes in.
Greg is candid that he's not deep into AI implementation yet — he describes himself as being "at the precipice" of that journey. But he sees clearly where AI fits, because he's been doing the work manually that AI would automate.
AI's superpower in veg management is its ability to process massive datasets and identify patterns that humans simply can't find manually. Imagine 13,000 or 20,000 miles of right-of-way generating thousands of field photos and data points daily. A human team can't review all of that in real time. AI can look at last year's photo next to this year's photo and immediately flag: is it better or worse? Instead of reviewing 10,000 images, the office team reviews the 6 or 300 that actually matter.
More importantly, AI learns. Each consecutive data collection — LIDAR, field observations, herbicide application rates, photo comparisons — feeds the model. Over time, AI moves from reactive analysis to predictive intelligence. Greg believes that with the right data, AI could eventually predict vegetation threats with the same accuracy that CenterPoint Energy is achieving with fuse failure predictions — 80-85% accuracy in pilot programs.
The critical prerequisite is data quality. Garbage in, garbage out. Without reliable field data collected consistently over time, AI has nothing to learn from. That's why the field data collection piece — making it easy for crews in the field to capture accurate information while they work — is foundational to everything else.
This is a topic Greg and I are both passionate about, and it's one of the biggest implementation challenges in the industry. The people who wield chainsaws don't want to be tech guys. They chose to work outdoors in wild spaces precisely because they don't want to sit at a desk. They don't want to feel like they're on a leash. And they certainly don't want software that was designed for the office user and bolted onto their workflow as an afterthought.
Greg's principle is simple: the tools have to make the field worker feel like they're winning. If the data collection adds burden without visible benefit to the person collecting it, adoption will fail. But if you can show a crew foreman that reporting this information means they did a good job — and that it's recognized and celebrated — they respond.
The practical considerations matter enormously. The tool has to work offline, because these crews are in areas with no cell coverage. It has to work like the apps they already use on their phones — thumb-driven, intuitive, minimal training required. And it has to integrate into the work, not be additional work layered on top.
Greg shared something that validated this approach: contract crews who work with his utility's data collection tools are now comparing that experience to their work elsewhere in the industry — and they prefer it. They actively want to come back. That's the sign that the implementation is working.
The broader philosophy: knowledge workers are in the office, revenue generators are in the field. Software should be designed to make the field worker's job easier, not to make the office worker's job easier at the field worker's expense.
Here's where Greg's results become impossible to ignore. He has cost models going back to 2018 and forward-looking projections through 2031. His costs are declining year over year. He's consistently beating his forecasted performance measures — doing more work each year than projected, which means actual costs are coming in lower than forecast.
By 2029, he expects a significant cost decrease. His 2031 numbers are lower than any year in the preceding decade. And when he projects those 2031 numbers forward with worst-case assumptions, the 2031-2041 window looks 60 times better than the previous 10 years.
In a high-inflation environment where every other cost in the utility world is escalating, Greg's veg management costs are going down. And he's not squeezing his vendors to get there. The margins for contract crews remain the same — they're just treating more miles per year at lower density and size structures, with fewer people, fewer safety incidents, and fewer reliability issues.
The most striking statistic: in the 10 years Greg has been operating at his current utility, they have never had a vegetation-related outage on any system they've managed. Not one.
Greg made a provocative observation near the end of our conversation. If you Google his approach right now — use the best AI search available — the internet would tell you it doesn't work. Industry leaders have told him to his face that what he's doing can't succeed.
The fundamental issue is that traditional cycle-based programs are self-reinforcing. You attack vegetation aggressively, let it recover for four years, then attack it again. Over time, the most aggressive species — the ones with the most seeds, the strongest root systems — come back even stronger. The industry is essentially cultivating monoculture environments of the most problematic species through its own management practices, then spending more money to fight the problem it created.
Greg's approach breaks that cycle. Instead of attacking and retreating, he manages the transition from problematic vegetation to stable, low-maintenance ground cover — meadows, grasslands, biodiverse environments that support wildlife and require minimal ongoing intervention. It costs money upfront, but the trajectory is unmistakable: each year costs less than the last, until eventually the system largely maintains itself biologically.
Greg Smith is doing something that matters — not just for his utility, but for every ratepayer in the country. He's demonstrating that vegetation management doesn't have to be a perpetually escalating cost center. With the right data, the right field tools, the right analytical approach, and a willingness to challenge industry orthodoxy, costs can decline while reliability improves.
The pieces are all there: LIDAR for initial prioritization, field data collection tools that work for the people actually doing the work, condition-based planning that targets resources where they're needed, and AI that can process the resulting data at a scale no human team can match.
The industry just has to be willing to look below the grass.
Greg, thank you for your partnership, your intellectual honesty, and your relentless drive to find a better way. The ratepayers of America are better off because of the work you're doing, whether they know it or not.
Listen to the full episode on From Boots to Boardroom.
From Boots to Boardroom is presented by KYRO AI — Digitize work and maximize profits.
Greg Smith brings a Moneyball mindset to what has traditionally been a biologically driven field — and the results are staggering. Lower costs. Zero vegetation-related outages on managed systems in over a decade. A 10-year cost trajectory that's declining, not escalating, even in a high-inflation environment. And a forward-looking model that projects costs 60 times better than the previous decade.
Greg has more than 20 years of experience across manufacturing, supply chain, and electric utility operations. Over the past decade, he's led everything from fleet and supply chain to transmission and vegetation programs under FAC 003. For the last five years, he's been all-in on vegetation management, and he's developed an approach that challenges nearly every assumption the industry has operated on for decades.
I've known Greg for close to a decade. He's one of the sharpest operational minds in the utility space, and this conversation — covering AI, LIDAR, condition-based planning, field data collection, and the future of veg management — is one of the most important episodes we've done for anyone who cares about grid reliability, ratepayer costs, and the future of utility operations.
I've always believed there are three foundational aspects of any utility: asset inspection, vegetation management, and storm response. Greg agrees, and he emphasizes that all three are deeply interconnected. How well you manage vegetation directly impacts how many outages you have during storms, which directly impacts reliability metrics, which directly impacts what ratepayers pay.
The numbers tell the story. Utilities have invested approximately $1.3 trillion since 2014 to improve the grid, with another trillion projected between now and 2029. Yet reliability metrics across the country have barely budged. Storm damage alone costs Americans roughly $150 billion a year. Not all of that is attributable to vegetation, but a substantial portion is — and the industry knows it.
The question Greg is answering isn't whether veg management matters. It's whether the way the industry has been doing it for decades is the right approach. His answer: it's not.
The traditional approach to vegetation management is cycle-based: pick a time interval — say, every four years — and address vegetation issues across an entire right-of-way corridor during that cycle. It sounds logical. It's organized. It's predictable.
It's also enormously wasteful.
Greg uses a simple analogy. Imagine only going into your backyard once every four years to deal with vegetation. You'd have foundation issues, sidewalk damage, deck problems, overgrown flower beds — and the cost to restore everything would be massive. Your spouse would probably leave you.
The same principle applies to utility rights-of-way. Over 108 miles of transmission line, vegetation profiles aren't consistent. Some areas have dense forests with fast-growing species. Others have sparse, low-growth landscapes. But in a fixed-cycle approach, you're sending the same crew with the same equipment across the entire corridor, regardless of what's actually there.
As Greg puts it: sometimes you're loaded for bear to shoot a squirrel. Other times you're loaded for squirrel and you've got a bear.
The result is chronic overspending in some areas and underspending in others — and the underspending creates future overspending, because the problems you didn't address grow larger, denser, and more expensive to manage. Time drives size and density. Size and density drive cost. The longer you wait, the more expensive it gets.
Greg's approach flips the model. Instead of treating an entire system uniformly on a fixed cycle, he uses real-time data to identify where the actual threats are and targets those specific areas with appropriately sized responses.
Here's a concrete example from his program. Last year, his team applied herbicide across 262 miles of right-of-way. On average, it was about 10 miles per day, using 260 gallons total. Spread evenly, that doesn't sound like a big problem.
But the data told a different story. 96 of those 260 gallons were concentrated in just 12 miles. Two other stretches accounted for 20 and 40 gallons in similar distances. The rest of the 262 miles required very little treatment.
Without data, you'd never know that. You'd treat the entire system the same way and miss the concentrations entirely — or you'd discover them after the fact and have to send crews back, doubling your cost.
With data, Greg restructured the following year's plan. The 35-36 miles with elevated indicators got an elevated response. The remaining miles got appropriately light treatment at very low cost. The crew does more targeted work, the utility spends less money, and the vegetation threats are actually addressed.
This is where Greg's approach gets genuinely revolutionary. The industry standard cost for veg management is $5,000 to $8,000 per mile according to KPMG data. Greg's program is achieving costs as low as $6 per acre in low-density areas and targeting $50 per acre in dense forest environments. The difference is staggering.
The key insight is what Greg calls managing "below the grass." LIDAR — the industry's go-to technology for vegetation assessment — doesn't see a tree until it is a tree. Brush under five inches or below 12 feet doesn't register. But that brush is the future forest that will eventually cost $5,000-$8,000 per mile to manage.
Greg shared an example from a recent site visit near College Station, Texas. Six-month-old trees were already 12 feet tall — growing from dormant root structures that produce fast-growing species. If left on a traditional four-year cycle, those become 25-foot trees requiring expensive removal crews. But right now, a targeted basal application can handle them for $70 per acre.
The math is simple: $70 per acre now versus $5,000-$8,000 per mile later. And the window to act is narrow. Every month you wait, the cost escalates.
Greg views LIDAR as a great starting point and an important long-term verification tool, but not the silver bullet the industry sometimes treats it as. LIDAR is expensive per mile, per structure, per span. It captures a snapshot in time — typically once every four to six years for most utilities, with California being the exception at annual flights. And it misses everything below its detection threshold.
The real value of LIDAR, in Greg's framework, is as the initial prioritization layer. LIDAR identifies your biggest threats. Then you go to those threats with appropriately sized crews and field data collection tools to manage the transition — tracking size and density changes in real time, not waiting four years for the next LIDAR flight to tell you what happened.
Over time, as you collect more field data between LIDAR cycles, the LIDAR itself becomes less critical for operational decisions. But the combination of periodic LIDAR snapshots with continuous field data creates an increasingly intelligent dataset — which is exactly where AI comes in.
Greg is candid that he's not deep into AI implementation yet — he describes himself as being "at the precipice" of that journey. But he sees clearly where AI fits, because he's been doing the work manually that AI would automate.
AI's superpower in veg management is its ability to process massive datasets and identify patterns that humans simply can't find manually. Imagine 13,000 or 20,000 miles of right-of-way generating thousands of field photos and data points daily. A human team can't review all of that in real time. AI can look at last year's photo next to this year's photo and immediately flag: is it better or worse? Instead of reviewing 10,000 images, the office team reviews the 6 or 300 that actually matter.
More importantly, AI learns. Each consecutive data collection — LIDAR, field observations, herbicide application rates, photo comparisons — feeds the model. Over time, AI moves from reactive analysis to predictive intelligence. Greg believes that with the right data, AI could eventually predict vegetation threats with the same accuracy that CenterPoint Energy is achieving with fuse failure predictions — 80-85% accuracy in pilot programs.
The critical prerequisite is data quality. Garbage in, garbage out. Without reliable field data collected consistently over time, AI has nothing to learn from. That's why the field data collection piece — making it easy for crews in the field to capture accurate information while they work — is foundational to everything else.
This is a topic Greg and I are both passionate about, and it's one of the biggest implementation challenges in the industry. The people who wield chainsaws don't want to be tech guys. They chose to work outdoors in wild spaces precisely because they don't want to sit at a desk. They don't want to feel like they're on a leash. And they certainly don't want software that was designed for the office user and bolted onto their workflow as an afterthought.
Greg's principle is simple: the tools have to make the field worker feel like they're winning. If the data collection adds burden without visible benefit to the person collecting it, adoption will fail. But if you can show a crew foreman that reporting this information means they did a good job — and that it's recognized and celebrated — they respond.
The practical considerations matter enormously. The tool has to work offline, because these crews are in areas with no cell coverage. It has to work like the apps they already use on their phones — thumb-driven, intuitive, minimal training required. And it has to integrate into the work, not be additional work layered on top.
Greg shared something that validated this approach: contract crews who work with his utility's data collection tools are now comparing that experience to their work elsewhere in the industry — and they prefer it. They actively want to come back. That's the sign that the implementation is working.
The broader philosophy: knowledge workers are in the office, revenue generators are in the field. Software should be designed to make the field worker's job easier, not to make the office worker's job easier at the field worker's expense.
Here's where Greg's results become impossible to ignore. He has cost models going back to 2018 and forward-looking projections through 2031. His costs are declining year over year. He's consistently beating his forecasted performance measures — doing more work each year than projected, which means actual costs are coming in lower than forecast.
By 2029, he expects a significant cost decrease. His 2031 numbers are lower than any year in the preceding decade. And when he projects those 2031 numbers forward with worst-case assumptions, the 2031-2041 window looks 60 times better than the previous 10 years.
In a high-inflation environment where every other cost in the utility world is escalating, Greg's veg management costs are going down. And he's not squeezing his vendors to get there. The margins for contract crews remain the same — they're just treating more miles per year at lower density and size structures, with fewer people, fewer safety incidents, and fewer reliability issues.
The most striking statistic: in the 10 years Greg has been operating at his current utility, they have never had a vegetation-related outage on any system they've managed. Not one.
Greg made a provocative observation near the end of our conversation. If you Google his approach right now — use the best AI search available — the internet would tell you it doesn't work. Industry leaders have told him to his face that what he's doing can't succeed.
The fundamental issue is that traditional cycle-based programs are self-reinforcing. You attack vegetation aggressively, let it recover for four years, then attack it again. Over time, the most aggressive species — the ones with the most seeds, the strongest root systems — come back even stronger. The industry is essentially cultivating monoculture environments of the most problematic species through its own management practices, then spending more money to fight the problem it created.
Greg's approach breaks that cycle. Instead of attacking and retreating, he manages the transition from problematic vegetation to stable, low-maintenance ground cover — meadows, grasslands, biodiverse environments that support wildlife and require minimal ongoing intervention. It costs money upfront, but the trajectory is unmistakable: each year costs less than the last, until eventually the system largely maintains itself biologically.
Greg Smith is doing something that matters — not just for his utility, but for every ratepayer in the country. He's demonstrating that vegetation management doesn't have to be a perpetually escalating cost center. With the right data, the right field tools, the right analytical approach, and a willingness to challenge industry orthodoxy, costs can decline while reliability improves.
The pieces are all there: LIDAR for initial prioritization, field data collection tools that work for the people actually doing the work, condition-based planning that targets resources where they're needed, and AI that can process the resulting data at a scale no human team can match.
The industry just has to be willing to look below the grass.
Greg, thank you for your partnership, your intellectual honesty, and your relentless drive to find a better way. The ratepayers of America are better off because of the work you're doing, whether they know it or not.
Listen to the full episode on From Boots to Boardroom.
From Boots to Boardroom is presented by KYRO AI — Digitize work and maximize profits.
Hari Vasudevan, PE, is a serial entrepreneur and engineer focused on AI-driven solutions for utilities, construction, and storm response. As Founder and CEO of KYRO AI, he leads the development of AI-powered software that helps utility, vegetation, and field service teams digitize operations, improve storm response and restoration, and reduce operational risk. He also serves as Vice Chair and Strategic Advisor for the Edison Electric Institute’s Transmission Subject Area Committee and holds bachelor’s and master’s degrees in civil engineering with professional engineering licensure in multiple states.
