Podcast Summary: Autonomous Construction Sites and AI-Powered Heavy Equipment with Bedrock Robotics

Podcast: Inevitable (an MCJ podcast)
Episode Title: Autonomous Construction Sites and AI-Powered Heavy Equipment with Bedrock Robotics
Host: Cody Simms
Guest: Boris Softman (CEO and Co-Founder, Bedrock Robotics)
Date: November 13, 2025

Overview

This episode explores the collision of two massive trends: a severe construction labor shortage and the urgent need to build large-scale infrastructure (data centers, clean energy, housing, manufacturing). Cody Simms interviews Boris Softman, whose background in autonomous vehicles (notably at Waymo) informs his work at Bedrock Robotics—where he’s applying advanced AI and machine learning to make heavy construction equipment fully autonomous. The conversation cuts across the technical, practical, and cultural challenges of deploying AI in physical industries, the path to full autonomy, and the broader implications for productivity, jobs, and industrial transformation.

Key Discussion Points

1. Bedrock Robotics: Mission and Approach

Core Offering: Bedrock enables full autonomy in existing heavy construction machinery (excavators, bulldozers, loaders) via rapid, reversible, hardware “upfits.” (02:11, Boris Softman)
Industries Impacted: While construction is the main focus, the technology has applications in agriculture, mining, waste, and beyond.
Business Model: By upfitting rather than building new machines, Bedrock leverages existing assets, minimizing customer disruption and maximizing adoption speed.

2. Boris Softman’s Background and the Evolution of AI/Robotics

Robotics and autonomy have been longstanding passions, from PhD work at Carnegie Mellon to consumer robots (Anki), then leading autonomous trucking and core tech teams at Waymo. (02:51–05:44)
At Waymo, the migration from hand-coded robotic heuristics to large-scale machine learning revealed how powerful data-driven models can generalize across domains and drive exponential scaling.

“We saw, not only did it solve these problems, but it had the ability to really start to generalize… it just became less and less expensive in terms of new information to become very competent. That was pretty magical.” (04:44, Boris Softman)

3. The Convergence of Labor Shortages and Infrastructure Demand

Dramatic construction labor shortfall, with estimates of up to 60% of skilled workers retiring in the next 7 years (05:23, Boris).
Construction machinery requires highly nuanced skills (six degrees of freedom, complex, years to master), intensifying the challenge as project demand soars.

4. The State and Future of Autonomous Vehicles & Robotics

Transition from the “science fiction” phase to rapid real-world deployment; passenger ride-hailing autonomy is now routine in places like San Francisco and Las Vegas, and spreading globally (12:03–14:38).
Physical automation (construction, manufacturing, logistics) represents a far larger share of global GDP than the digital domain—a “frontier” just now opening as AI matures (10:09–11:51).

“The rest of it is the physical world… and that’s the frontier because now all of a sudden robotics… This is the beginning of not all at once, but a sequence of really meaningful jumps where the physical world is going to get smarter.” (11:13, Boris)

5. Why Construction? Selecting the Right Industry

High machine utilization, substantial labor tied to machine operation, and the existence of a vast installed base of machinery make construction particularly attractive.
Construction is 13% of global GDP; demand is intense and concentrated (data centers, $170B/year in US alone; manufacturing $250B/year). (22:11–24:13)

6. Bedrock’s Technology: Upfitting Existing Machines

No Permanent Modifications: Upfit reversible in under three hours (25:26–25:38).
Sensor Suite: Cameras, LiDAR, GPS (dual), IMU, LTE for precise awareness and control.
Manual Override: Machines can be manually operated at any time for flexibility.

“We could do this in less than three hours... and it doesn’t even permanently modify the machine. That’s pretty powerful.” (25:38, Boris)

7. Starting with Excavators

Excavators are widespread, highly utilized, and among the hardest machines to operate due to their complexity.
Massive business pull—solving excavator autonomy would unlock a significant wedge in construction automation. (26:25–27:40)

8. Technical Challenges: Changing Environments & Machine Learning

Unlike vehicles on roads, construction sites are actively modified by the machines (earth moved, topography changes) (28:16–29:41).
Bedrock’s ML models are designed to “learn” how to mold environments toward goal states (i.e., the desired final grading or excavation) across varied conditions.

“You’re actually shifting the topography… you are running these patterns across thousands and thousands of hours of data… and eventually those patterns generalize.” (28:21–29:41)

9. Why Full Autonomy, Not Teleoperation?

Teleoperation is a tool but doesn’t deliver the transformational value (labor-free, 24/7 ops, full digitization) of end-to-end autonomy.
Building for teleop diverts years of effort from true autonomy and leaves the human in the loop (29:41–30:54).

10. Control, Connectivity, and Reliability

All real-time, safety-critical autonomy functions run locally on the machine; cloud or remote interfaces are used for fleet coordination and less time-sensitive tasks (32:09–34:36).
Leveraging the automotive supply chain's advances in compute and sensor cost/reliability.

11. Go-to-Market: Customers, Use Cases, and Initial Deployments

Customers: General contractors and subcontractors who own/operate fleets and manage large-scale earthwork (34:36–36:48).
Beachhead Sectors: Large, repetitive, high-utilization projects—factories, data centers, warehouses, municipal plants, massive civil works.
Status: Autonomy testing since late last year; aiming for “driverless” first deployments next year (37:20–38:11).

“The fact that we can go from zero to first deployments in a few years, that's a sign of how powerful some of these evolutions and learnings… are.” (37:25, Boris)

12. Funding and Team

Raised: $80M in Seed + Series A (notable investors: Eclipse, 8VC, former Waymo CEO John Krafczyk, Nvidia’s venture arm, Two Sigma Ventures)
Strategy: Blend top robotics/AI technical talent with domain expertise and strong investor/support network. (39:05)

13. Cultural Adoption and Respecting Industry Workflows

Key Insight: Tech must fit seamlessly into existing construction workflows rather than force dramatic operational changes.
Respect for Experience: Deep collaboration with experienced operators; Bedrock’s systems designed to operate safely around people and machines, with minimal additional constraints.

“The mistake that AI robotics companies can make is coming into a space and saying, 'Well, I'm going to reinvent everything…' That's an easy way to be ignored… We try to be incredibly respectful of the way people operate today.” (41:34, 42:09, Boris)

14. The Near-Term Value and Long-Term Vision

Immediate Painkiller: Addressing immediate labor shortages and safety issues is the “painkiller”; full site, 24/7, digitized work is the “vitamin” for the future (44:31).
Platform Play: Over time, Bedrock aims to generalize to multiple machines, sectors, and possibly redefine construction equipment form factors.
Core Product: The “Bedrock Operator”—a system that gets better over time and gains new capabilities, transcending single-machine or skill limitations. (47:09)

15. The Broader Context: A New Industrial Revolution

The transition parallels prior industrial revolutions—initially disruptive, but ultimately expansionary and wealth-generating.
Physical AI systems are the next big wave, with the potential to radically increase productivity beyond what digital-only AI achieved.

“This is the merger of the digital world and the physical productivity that is in front of us…” (48:04, Boris)

Notable Quotes & Memorable Moments

On AI’s Generalization Power:

“The system would extend super well to other cities… and you started getting more and more of a subsidy on the data side, where it just became less and less expensive in terms of new information to become very competent. And that was pretty magical.” (04:44, Boris)
On Construction’s Bottlenecks:

“You have this pretty fascinating split of supply and demand and a value proposition that, as we started talking to customers, goes far beyond labor… compress schedules by working 24 hours… Predictability is very hard.” (22:11, Boris)
On Operating in Traditional Industries:

“The mistake… is coming into a space and saying, 'I'm going to reinvent everything.' … That’s an easy way to be ignored because people don’t have time to change your entire workflow.” (41:34, Boris)
On Form Factors & Human Constraints:

“…most machines were designed to optimize the constraints of what one person can physically do and… In the very, very long term. I think there’s an opportunity to actually rethink what this looks like.” (46:08, Boris)
On the Promise of Physical AI:

“The next big wave is going to be the physical manifestations of AI… I think it’s going to be waves of these progressions over the next few decades, but I’m incredibly excited. This is the merger of the digital world and the physical productivity that is in front of us.” (48:04, Boris)

Timestamps for Key Segments

[02:11] What is Bedrock Robotics? (Boris Softman)
[04:44] The magic of ML generalization from Waymo to construction
[12:03] The reality and near-future of autonomous vehicles in cities
[15:51] The scale of industrial robotics energy needs
[22:11] Why construction is the ideal entry point for robotic autonomy
[25:26] Technical overview: upfitting, not retrofitting
[26:25] Choice of starting with excavators
[28:16] Navigating ever-shifting construction environments with ML
[29:41] Why not start with tele-op or “level 2” solutions
[32:09] System architecture: on-machine autonomy, cloud only for non-time-critical ops
[34:36] Early customers and use cases
[37:20] Deployment status, speed vs. Waymo’s timeline
[39:05] Funding, team, and capitalization
[41:34] Adapting to (and respecting) construction industry workflows
[44:31] Near-term “painkiller,” long-term “vitamin” value
[46:08] Long-term opportunity to rethink machine form factors

Conclusion

Bedrock Robotics is bringing advanced autonomy to some of the world’s most challenging, high-impact environments—not by replacing proven machines, but by layering in AI-driven intelligence and control. The technology addresses urgent labor shortages, safety, and productivity bottlenecks today, while laying the groundwork for a radical reimagining of how things are built tomorrow. Boris Softman's journey from academic robotics to consumer products to the vanguard of industrial autonomy illustrates the power of cross-pollination between digital and physical innovation.

For more info, visit: Bedrock Robotics | MCJ Collective

Podcast Summary: Autonomous Construction Sites and AI-Powered Heavy Equipment with Bedrock Robotics

Overview

Key Discussion Points

1. Bedrock Robotics: Mission and Approach

Core Offering: Bedrock enables full autonomy in existing heavy construction machinery (excavators, bulldozers, loaders) via rapid, reversible, hardware “upfits.” (02:11, Boris Softman)
Industries Impacted: While construction is the main focus, the technology has applications in agriculture, mining, waste, and beyond.
Business Model: By upfitting rather than building new machines, Bedrock leverages existing assets, minimizing customer disruption and maximizing adoption speed.

2. Boris Softman’s Background and the Evolution of AI/Robotics

Robotics and autonomy have been longstanding passions, from PhD work at Carnegie Mellon to consumer robots (Anki), then leading autonomous trucking and core tech teams at Waymo. (02:51–05:44)
At Waymo, the migration from hand-coded robotic heuristics to large-scale machine learning revealed how powerful data-driven models can generalize across domains and drive exponential scaling.

“We saw, not only did it solve these problems, but it had the ability to really start to generalize… it just became less and less expensive in terms of new information to become very competent. That was pretty magical.” (04:44, Boris Softman)

3. The Convergence of Labor Shortages and Infrastructure Demand

Dramatic construction labor shortfall, with estimates of up to 60% of skilled workers retiring in the next 7 years (05:23, Boris).
Construction machinery requires highly nuanced skills (six degrees of freedom, complex, years to master), intensifying the challenge as project demand soars.

4. The State and Future of Autonomous Vehicles & Robotics

Transition from the “science fiction” phase to rapid real-world deployment; passenger ride-hailing autonomy is now routine in places like San Francisco and Las Vegas, and spreading globally (12:03–14:38).
Physical automation (construction, manufacturing, logistics) represents a far larger share of global GDP than the digital domain—a “frontier” just now opening as AI matures (10:09–11:51).

“The rest of it is the physical world… and that’s the frontier because now all of a sudden robotics… This is the beginning of not all at once, but a sequence of really meaningful jumps where the physical world is going to get smarter.” (11:13, Boris)

5. Why Construction? Selecting the Right Industry

High machine utilization, substantial labor tied to machine operation, and the existence of a vast installed base of machinery make construction particularly attractive.
Construction is 13% of global GDP; demand is intense and concentrated (data centers, $170B/year in US alone; manufacturing $250B/year). (22:11–24:13)

6. Bedrock’s Technology: Upfitting Existing Machines

No Permanent Modifications: Upfit reversible in under three hours (25:26–25:38).
Sensor Suite: Cameras, LiDAR, GPS (dual), IMU, LTE for precise awareness and control.
Manual Override: Machines can be manually operated at any time for flexibility.

“We could do this in less than three hours... and it doesn’t even permanently modify the machine. That’s pretty powerful.” (25:38, Boris)

7. Starting with Excavators

Excavators are widespread, highly utilized, and among the hardest machines to operate due to their complexity.
Massive business pull—solving excavator autonomy would unlock a significant wedge in construction automation. (26:25–27:40)

8. Technical Challenges: Changing Environments & Machine Learning

Unlike vehicles on roads, construction sites are actively modified by the machines (earth moved, topography changes) (28:16–29:41).
Bedrock’s ML models are designed to “learn” how to mold environments toward goal states (i.e., the desired final grading or excavation) across varied conditions.

“You’re actually shifting the topography… you are running these patterns across thousands and thousands of hours of data… and eventually those patterns generalize.” (28:21–29:41)

9. Why Full Autonomy, Not Teleoperation?

Teleoperation is a tool but doesn’t deliver the transformational value (labor-free, 24/7 ops, full digitization) of end-to-end autonomy.
Building for teleop diverts years of effort from true autonomy and leaves the human in the loop (29:41–30:54).

10. Control, Connectivity, and Reliability

All real-time, safety-critical autonomy functions run locally on the machine; cloud or remote interfaces are used for fleet coordination and less time-sensitive tasks (32:09–34:36).
Leveraging the automotive supply chain's advances in compute and sensor cost/reliability.

11. Go-to-Market: Customers, Use Cases, and Initial Deployments

Customers: General contractors and subcontractors who own/operate fleets and manage large-scale earthwork (34:36–36:48).
Beachhead Sectors: Large, repetitive, high-utilization projects—factories, data centers, warehouses, municipal plants, massive civil works.
Status: Autonomy testing since late last year; aiming for “driverless” first deployments next year (37:20–38:11).

“The fact that we can go from zero to first deployments in a few years, that's a sign of how powerful some of these evolutions and learnings… are.” (37:25, Boris)

12. Funding and Team

Raised: $80M in Seed + Series A (notable investors: Eclipse, 8VC, former Waymo CEO John Krafczyk, Nvidia’s venture arm, Two Sigma Ventures)
Strategy: Blend top robotics/AI technical talent with domain expertise and strong investor/support network. (39:05)

13. Cultural Adoption and Respecting Industry Workflows

Key Insight: Tech must fit seamlessly into existing construction workflows rather than force dramatic operational changes.
Respect for Experience: Deep collaboration with experienced operators; Bedrock’s systems designed to operate safely around people and machines, with minimal additional constraints.

“The mistake that AI robotics companies can make is coming into a space and saying, 'Well, I'm going to reinvent everything…' That's an easy way to be ignored… We try to be incredibly respectful of the way people operate today.” (41:34, 42:09, Boris)

14. The Near-Term Value and Long-Term Vision

Immediate Painkiller: Addressing immediate labor shortages and safety issues is the “painkiller”; full site, 24/7, digitized work is the “vitamin” for the future (44:31).
Platform Play: Over time, Bedrock aims to generalize to multiple machines, sectors, and possibly redefine construction equipment form factors.
Core Product: The “Bedrock Operator”—a system that gets better over time and gains new capabilities, transcending single-machine or skill limitations. (47:09)

15. The Broader Context: A New Industrial Revolution

The transition parallels prior industrial revolutions—initially disruptive, but ultimately expansionary and wealth-generating.
Physical AI systems are the next big wave, with the potential to radically increase productivity beyond what digital-only AI achieved.

“This is the merger of the digital world and the physical productivity that is in front of us…” (48:04, Boris)

Notable Quotes & Memorable Moments

On AI’s Generalization Power:

“The system would extend super well to other cities… and you started getting more and more of a subsidy on the data side, where it just became less and less expensive in terms of new information to become very competent. And that was pretty magical.” (04:44, Boris)
On Construction’s Bottlenecks:

“You have this pretty fascinating split of supply and demand and a value proposition that, as we started talking to customers, goes far beyond labor… compress schedules by working 24 hours… Predictability is very hard.” (22:11, Boris)
On Operating in Traditional Industries:

“The mistake… is coming into a space and saying, 'I'm going to reinvent everything.' … That’s an easy way to be ignored because people don’t have time to change your entire workflow.” (41:34, Boris)
On Form Factors & Human Constraints:

“…most machines were designed to optimize the constraints of what one person can physically do and… In the very, very long term. I think there’s an opportunity to actually rethink what this looks like.” (46:08, Boris)
On the Promise of Physical AI:

“The next big wave is going to be the physical manifestations of AI… I think it’s going to be waves of these progressions over the next few decades, but I’m incredibly excited. This is the merger of the digital world and the physical productivity that is in front of us.” (48:04, Boris)

Timestamps for Key Segments

[02:11] What is Bedrock Robotics? (Boris Softman)
[04:44] The magic of ML generalization from Waymo to construction
[12:03] The reality and near-future of autonomous vehicles in cities
[15:51] The scale of industrial robotics energy needs
[22:11] Why construction is the ideal entry point for robotic autonomy
[25:26] Technical overview: upfitting, not retrofitting
[26:25] Choice of starting with excavators
[28:16] Navigating ever-shifting construction environments with ML
[29:41] Why not start with tele-op or “level 2” solutions
[32:09] System architecture: on-machine autonomy, cloud only for non-time-critical ops
[34:36] Early customers and use cases
[37:20] Deployment status, speed vs. Waymo’s timeline
[39:05] Funding, team, and capitalization
[41:34] Adapting to (and respecting) construction industry workflows
[44:31] Near-term “painkiller,” long-term “vitamin” value
[46:08] Long-term opportunity to rethink machine form factors

Conclusion

For more info, visit: Bedrock Robotics | MCJ Collective

Autonomous Construction Sites and AI-Powered Heavy Equipment with Bedrock Robotics

Powered by Wave AI

Summary

Podcast Summary: Autonomous Construction Sites and AI-Powered Heavy Equipment with Bedrock Robotics

Overview

Key Discussion Points

1. Bedrock Robotics: Mission and Approach

2. Boris Softman’s Background and the Evolution of AI/Robotics

3. The Convergence of Labor Shortages and Infrastructure Demand

4. The State and Future of Autonomous Vehicles & Robotics

5. Why Construction? Selecting the Right Industry

6. Bedrock’s Technology: Upfitting Existing Machines

7. Starting with Excavators

8. Technical Challenges: Changing Environments & Machine Learning

9. Why Full Autonomy, Not Teleoperation?

10. Control, Connectivity, and Reliability

11. Go-to-Market: Customers, Use Cases, and Initial Deployments

12. Funding and Team

13. Cultural Adoption and Respecting Industry Workflows

14. The Near-Term Value and Long-Term Vision

15. The Broader Context: A New Industrial Revolution

Notable Quotes & Memorable Moments

Timestamps for Key Segments

Conclusion

Summary

Podcast Summary: Autonomous Construction Sites and AI-Powered Heavy Equipment with Bedrock Robotics

Overview

Key Discussion Points

1. Bedrock Robotics: Mission and Approach

2. Boris Softman’s Background and the Evolution of AI/Robotics

3. The Convergence of Labor Shortages and Infrastructure Demand

4. The State and Future of Autonomous Vehicles & Robotics

5. Why Construction? Selecting the Right Industry

6. Bedrock’s Technology: Upfitting Existing Machines

7. Starting with Excavators

8. Technical Challenges: Changing Environments & Machine Learning

9. Why Full Autonomy, Not Teleoperation?

10. Control, Connectivity, and Reliability

11. Go-to-Market: Customers, Use Cases, and Initial Deployments

12. Funding and Team

13. Cultural Adoption and Respecting Industry Workflows

14. The Near-Term Value and Long-Term Vision

15. The Broader Context: A New Industrial Revolution

Notable Quotes & Memorable Moments

Timestamps for Key Segments

Conclusion