Bricks And Bytes

“We never use the word AI anywhere. You’re saying we’re an AI company, not me.”

Mo Akbari-Hochberg built HoloBuilder and sold it to FARO. Now he is back with BuildCrew, a preconstruction platform backed by Khosla Ventures that uses digital workers to help contractors with estimating, trade scoping, bid levelling and preconstruction RFIs.

In this episode, Mo explains why he believes 99% of construction AI teams do not properly understand the technology they are building with, why access to the same AI models does not produce the same results and what contractors should look for when buying AI software.

We also discuss how BuildCrew can produce a rough order of magnitude estimate in 4 to 5 hours, compare 10 electrical bids in around 20 minutes and identify gaps across construction drawings and project documents.

🎥 Link in comments.

#BricksAndBytes #ConstructionAI #ConstructionTechnology #Preconstruction #ConstructionEstimating

Our Sponsors:

BreadCrumb- 50,000+ projects globally. All running safer, faster, with Breadcrumb. - breadcrumb.co

Aphex is the multiplayer planning platform where construction teams plan together, stay aligned, and deliver projects faster – check out aphex.co

Archdesk - “The #1 Construction Management Software for Growing Companies - Manage your projects from Tender to Handover” check archdesk.com

Chapters

00:00 Intro

01:00 How Mo Built a Startup: Founder Story

03:27 How to Build a High-Performing Team

05:30 AI in Construction: The Complete Breakdown

08:13 What is an AI Digital Intern?

11:16 BuildCrew AI vs Traditional Construction Tech

14:09 UI/UX & Reliability in B2B SaaS

16:33 Biggest Pain Points in Pre-Construction

19:29 High-End Residential Construction Challenges

22:23 Scaling Beyond Luxury Homes to Commercial

24:44 Customer Feedback & Product-Market Fit

28:54 How BuildCrew AI Saves Time on Site

31:02 Scaling Your Core Startup Team

34:19 Startup Hiring Strategy: Attracting Top Talent

35:58 Biggest Founder Mistakes & Lessons Learned

37:10 How to Build a Great Tech Product

38:10 How to Raise Venture Capital (VC Strategy)

39:29 AI Productivity Benchmarks & Industry Insights

43:38 Practical Use Cases of AI in Construction

47:26 Biggest Misconceptions About AI

52:17 Adapting to Rapid Changes in AI & Tech

54:52 Will AI Replace Estimators? (Future of ConTech)

This week's briefing covers three things that landed in the same week and all point at the same question. Where is the work going next.

In the last five weeks more than two billion dollars has been poured into construction machines that drive themselves. TerraFirma raised 115 million. Travis Kalanick, the man who founded Uber, raised 1.7 billion for Atoms, which includes an autonomous mining division. Komatsu signed a partnership to put autonomous bulldozers and excavators on American jobsites. And Bedrock Robotics, who regular listeners will remember from our Texas site visit, announced fully autonomous excavators working on live customer sites in Nevada and Texas. But this is not one market. There are five distinct business models in play, and they do not all sit on your side of the table. One of them is bidding against you for the same earthworks contracts.

Then the reason all this money is chasing machines in the dirt. Travis Kalanick put it simply. Manufacturing changes atoms, real estate stores them, transport moves them. Uber was the network layer for moving things. His new company is building the same thing for entire industries. Construction sites may be where AI learns to understand the physical world, and every hour an autonomous machine works it is banking data that no lab can buy off a shelf.

Then the two numbers that matter for your order book. American housebuilders dropped 12.4% in July to their weakest level since November 2022. The strange part is that permits went up 5% to 1.44 million. Builders are paying for permission and refusing to break ground. That is them telling you in the clearest language they speak that they do not trust demand at today's prices. And the work that has been filling the gap, data centres, hit a new kind of wall this week. Not money. Not power. Votes. Fort Worth voted unanimously to pause new data centre construction. Louisville voted 24 to 1 to pause approvals for six months. New York already has a statewide freeze in place. Permission has become the scarce input, and it is being decided city by city.

Drop your answers to this week's questions in the comments of this week's LinkedIn post.

We spoke to Kevin Peterson this week. He runs the tech at Bedrock Robotics.

He told us their control room "looks like StarCraft, actually."

You click on an excavator on a screen and give it a job. Then it just digs. Nobody sitting in it.

This went live on real customer sites on Monday.

The same day, a company called Gravis Robotics got $200 million from SoftBank.

We asked Kevin about the timing. Nobody seems to know.

Full chat with Kevin and Alain is out now on Bricks Bucks and Bytes.

#bricksbucksandbytes #bricksandbytes #aec #constructiontech #ai

Our Sponsors:

BreadCrumb- 50,000+ projects globally. All running safer, faster, with Breadcrumb. - breadcrumb.co

Aphex is the multiplayer planning platform where construction teams plan together, stay aligned, and deliver projects faster – check out aphex.co

Archdesk - “The #1 Construction Management Software for Growing Companies - Manage your projects from Tender to Handover” check archdesk.com

Chapters00:00 Intro

00:55 Construction Autonomy's Biggest Week04:25 Bedrock Robotics Goes Fully Autonomous + Gravis $200M SoftBank Round06:54 Why Autonomous Machines Took 25 Years to Arrive09:52 How Waymo Technology Made Jobsite Autonomy Possible12:50 What Does Fully Autonomous Actually Mean?16:06 Who Is Investing in Construction Robotics (SoftBank, CapitalG, NVIDIA)18:59 Autonomous Excavation vs Excavators: The Real Market22:01 Kevin Peterson Interview: No Driver, No Operator, No Oversight24:48 Bedrock vs Gravis: The Autonomy Race Heats Up28:01 Self-Driving Cars vs Construction: Which Is Harder?31:42 How Autonomous Excavators Handle Safety on Site33:39 Training an Excavator Like an LLM: Physical AI Explained36:44 Orchestrating Machine Fleets on Construction Sites39:10 Why Construction Sites Break Self-Driving Logic44:33 World Models, Procore DroneDeploy Deal & AI Consolidation51:16 Will AI Replace Construction Consultants?

"Don't build. It's that simple."

Strange advice from a man who built his own Procore.In 2011, Novo Construction evaluated Procore and walked away.

They built Sentinel instead, their own system of record. Fifteen years later, Colin Stoner (Chief Innovation Officer, 22 years at the company) told us why that call made them AI ready before most of the industry had a data strategy.

The catch in his advice: build only if your team has total buy in. Without it, don't start.

✓ Every AI build at Novo starts with one question: "What do you hate about your job?"

✓ RFIs drafted by AI trained on more than a decade of their own writing

✓ Claude Code in use every day. His word: "religiously"

✓ Headcount untouched.

His goal, verbatim: "same headcount, more revenue, more projects"

#bricksandbytes #bricksbytes #aec #constructiontech #ai

Chapters

00:00 Intro01:00 Introduction to Colin Stoner and Novo Construction

04:02 The Role of AI in Construction

07:01 Technology Evolution in Construction

10:07 Challenges in Construction Technology Adoption

12:57 Building Technology Internally vs. Buying

16:00 The Sentinel System: Overview and Benefits

19:07 AI Integration and Future Prospects

22:06 Advice for Companies Considering Internal Development

27:11 Streamlining Certificate of Insurance Tracking

29:18 Revolutionizing RFI Processes with AI

33:03 Enhancing Project Management with AI Tools

35:20 Navigating External Tech Solutions

39:52 Challenges of Tech Sales in Construction

42:57 Preparing for AI Integration in Construction

46:24 The Role of AI in Workforce Dynamics

47:34 Maintaining Human Touch in Construction CommunicationOur Sponsors:

BreadCrumb- 50,000+ projects globally. All running safer, faster, with Breadcrumb. - breadcrumb.co

Aphex is the multiplayer planning platform where construction teams plan together, stay aligned, and deliver projects faster – check out aphex.co

Archdesk - “The #1 Construction Management Software for Growing Companies - Manage your projects from Tender to Handover” check archdesk.com

This week's briefing covers three things that landed in the same fortnight and point in the same direction.

First, a closed room in central London. The CTO of DPR Construction, a 14 billion dollar US contractor, sat around a table with tier one contractors, engineers, investors and founders. The consensus was uncomfortable. The biggest clients on the planet are telling their supply chains to price on outcomes instead of hours, and to expect that price to fall every year. Nobody in the room, client side included, had a working answer for what replaces the hours model. The near term AI pressure lands in the back office first, not on site. The people best placed to rethink how this industry works are the exact ones pinned to the most lucrative projects, with no time to look up.

Then the numbers. The Associated Builders and Contractors survey puts the average US contractor backlog at eight months flat, the lowest since January, down across every region, every sector and every company size. The 12% of contractors holding data centre work are sitting on 11.4 months of runway. The other 88% are at 7.5. That gap went from two and a half months in June to 3.9 months in July. Fluor, one of the biggest engineering and construction firms on earth, saw its shares jump 17.4% in a single day for saying the better money sits one layer below the boom, in power, on contracts where someone else carries the risk.

Then open weight AI. Meta and Nvidia both released open weight models this week. The cost of running these models has fallen from 30 dollars per million tokens in 2023 to fractions of a cent today. Gartner projects costs fall another 90% by 2030. The firms selling AI tools built on top of someone else's model are about to get squeezed out. The construction firms that win are the ones that take their own project data, the one asset the big labs cannot get anywhere else, and train on top of it. Not your weights, not your product.

Drop your answers to this week's questions in the comments of the LinkedIn post.

"The US is dangerously close to the brink of being Japan 1989."

That is not a hot take from a doom account. That is Patric on this week's Bricks, Bucks & Bytes, laying out why asset overvaluation in private markets and infrastructure has him worried about stagflation.

On this episode:

✅ Why data center contractors are sitting on 11.4 months of backlog while everyone else has 7.5

✅ The AI productivity story corporate America still cannot prove with real numbers

✅ Why the construction RFP process is fundamentally broken, and who is actually responsible for it

✅ What Meta's new partnership with North America's Building Trades Unions means for the data center buildout

Dustin DeVan put it plainly: "You can't now build a project that much cheaper."

Where do you land? Are we in a two speed economy that eventually corrects, or is this just what infrastructure spending looks like now? Tell us why we're wrong.

#bricksandbytes #bricksbytes #bricksbucksandbytes #aec #construction #constructiontech #ai #vc

Our Sponsors:

BreadCrumb- 50,000+ projects globally. All running safer, faster, with Breadcrumb. - breadcrumb.co

Aphex is the multiplayer planning platform where construction teams plan together, stay aligned, and deliver projects faster – check out aphex.co

Archdesk - “The #1 Construction Management Software for Growing Companies - Manage your projects from Tender to Handover” check archdesk.com

Chapters

00:00 Intro

01:08 Welcome Back and Personal Updates

03:43 Exploring Proto Town and Bedrock Robotics

06:41 Travel Frustrations and Unexpected Adventures

09:30 The Concept of Proto Town in Europe

12:12 AI Price Collapse and Economic Implications

14:58 Data Center Boom and Its Effects

17:31 Market Dislocations and Construction Industry Insights

19:51 Interest Rates and Economic Perspectives

22:50 Potential Risks and Future Predictions

25:39 Final Thoughts on Economic Trends

30:38 Understanding Time vs. Cost Savings

32:07 The Role of Technology in Construction

33:36 Owners and Technology Awareness

34:41 The Importance of Pricing Optionality

36:14 The Shift in AI Pricing and Accessibility

38:42 Navigating AI Models and Corporate Needs

43:50 Leadership and Product Understanding

45:54 Governance and AI Deployment Challenges

53:34 Rethinking RFP Processes for Better Outcomes

56:25 Meta's Role in Construction Training Initiatives 

"Everybody thinks that autonomous heavy equipment is commonplace, whereas in fact the adoption is basically zero."

Frederik Filz-Reiterdank said that to us from the operator's seat of an autonomous roller, on a live chip fab site in upstate New York, with his dozers and dump trucks working away behind him.

His company, Crewline AI, raised a $7.1M seed earlier this year led by Initialized Capital and Nebular to automate the machine operators hate driving.

What we got into:

✅ Why the roller is the wedge for construction autonomy, and why the excavator falls last

✅ A retrofit kit that turns a standard drum roller into a self-driving one in under an hour

✅ Selling a "virtual roller operator" instead of renting out a robot

✅ The four month insurance grind nobody warns robotics founders about

✅ Why 95% of candidates are disqualified the moment they hear about the forward-deployed lifestyle

Full episode now live on YouTube and Spotify.

#bricksandbytes #bricksbytes #bricksbucksandbytes #aec #construction #constructiontech #ai #vc

Our Sponsors:

BreadCrumb- 50,000+ projects globally. All running safer, faster, with Breadcrumb. - breadcrumb.co

Aphex is the multiplayer planning platform where construction teams plan together, stay aligned, and deliver projects faster – check out aphex.co

Archdesk - “The #1 Construction Management Software for Growing Companies - Manage your projects from Tender to Handover” check archdesk.com

Chapters

00:00 Teaser

01:00 Introduction and Context of the Conversation

03:43 Autonomy in Construction: Current State and Misconceptions

09:21 Choosing Rollers: The Perfect Wedge for Automation

11:47 Understanding Pain Points in Construction

14:55 Business Models: Robotics as a Service vs. Outcome-Based Pricing

18:49 Deployment Strategy for Autonomous Rollers

21:00 Insurance and Liability Challenges

22:09 Building Trust with Customers

24:45 Overcoming Skepticism in Construction

27:49 Future of Automation in Construction

32:10 First Steps for Contractors

A 4x levered AI fund handed its entire public book to Citadel. The S&P 500 crossed $70 trillion days later.

And somewhere in the middle of all that, Patric put a date on the next one:

"We're 15 months away from that Palantir moment, 100%."

Physical automation. His argument: Asia already won the hardware game, so the money is with whoever actually deploys the machines. Which, quietly, is a contractor business.

Dustin doesn't buy the humanoid hype. Martin wants to know who gets hurt when the music stops.

New Bricks, Bucks & Bytes out now.

#bricksandbytes #bricksbytes #bricksbucksandbytes #aec #construction #constructiontech #vc #ai

Our Sponsors:

BreadCrumb- 50,000+ projects globally. All running safer, faster, with Breadcrumb. - breadcrumb.co

Aphex is the multiplayer planning platform where construction teams plan together, stay aligned, and deliver projects faster – check out aphex.co

Archdesk - “The #1 Construction Management Software for Growing Companies - Manage your projects from Tender to Handover” check archdesk.com

Chapters

00:00 Teaser

01:00 The AI fund that had to sell everything

04:04 Who is Leopold Aschenbrenner?

06:54 The S&P 500 hits $70 trillion

10:04 Without data centers, GDP is barely growing

13:04 Can regulation actually stop a crash?

15:53 Why the market might never crash again

24:34 Nobody is funding the physical economy

25:53 Asia won robot hardware. The Palantir moment is next

31:04 Tax the robots, retire at 55

34:39 Retraining workers for the robot economy

We asked Scoreboard AI's Charles Ma what the single most expensive mistake contractors make at buyout is. His answer:"Outsourcing their understanding of the project to their trade partners."Charles was employee #1 at Versatile, one of construction tech's most well-funded bets (over $100M raised). Now he's building Scoreboard AI to fix a much quieter problem: scope gaps that get buried in project documents and don't surface until the crew is standing there waiting for a fix.In this episode we get into:- What Versatile got right, and where the product fell short of the ROI it promised- Why Charles thinks "AI wrappers are dead" in preconstruction- How hard bid, GMP, and design-build buyout processes each create their own scope gap risk- Why Scoreboard has deliberately stayed away from flashy fundraise announcementsFull episode is live on Bricks and Bytes Youtube Channel#bricksandbytes #bricksbytes #aec #construction #constructiontech #ai #vc

Chapters

00:00 - Teaser

01:00 - Quickfire Round

02:32 - Inside Versatile: the early days and the Chase Center project

09:27 - Is construction tech overhyped or underhyped right now

11:10 - Why Scoreboard has avoided the flashy fundraise playbook

16:43 - The perverse incentives of VC-backed construction tech

20:21 - Do "AI wrappers" win the current land grab

26:02 - Why Charles believes AI wrappers are dead

28:25 - What a scope gap actually is, and why it happens at buyout

32:55 - Hard bid vs. GMP vs. design-build: where the gaps creep in

36:50 - Measuring the real cost of scope gaps

41:51 - How Scoreboard AI actually works

43:21 - Show your work: building trust in AI outputs

48:59 - Pricing philosophy: running Scoreboard like a Costco

51:26 - Expansion beyond the US

52:41 - What's working in go-to-market right now

54:08 - Will AI ever run preconstruction end-to-endOur Sponsors:

BreadCrumb- 50,000+ projects globally. All running safer, faster, with Breadcrumb. - breadcrumb.co

Aphex is the multiplayer planning platform where construction teams plan together, stay aligned, and deliver projects faster – check out aphex.co

Archdesk - “The #1 Construction Management Software for Growing Companies - Manage your projects from Tender to Handover” check archdesk.com

This week's briefing covers three things that are already changing the shape of construction, one of them from a dirt road in Texas at 4am.

Bedrock Robotics was founded by the team behind Waymo's driverless technology. They do not build new excavators. They take the ones contractors already own, fit them with sensors and a computer, and the machine digs on its own. This week we filmed a full documentary with them at Proto Town, a 12 acre test site near Austin where some of the most advanced hard tech companies on earth go to build things they cannot build anywhere else. The operator who has spent his career inside these machines watched one run without him and his answer was relief. The people closest to autonomous equipment are not the ones campaigning against it. The contractor running them on live sites across Texas today says productivity is up, revenue is up, and his crews are happier.

Then BlackRock. On Tuesday Meta and BlackRock announced a $14 billion data centre campus in El Paso. BlackRock owns 80%. Meta leases its own building back, operates it, and manages the construction. Roughly 90 cents in every dollar is borrowed. Wall Street used to lend money to these builds. It is now the one that owns them. That changes who your actual client is on the biggest programmes in the market.

And then the numbers behind the whole cycle. Over the past fortnight the four biggest data centre customers on earth all opened their books. Moody's tallied what six technology companies will spend on this build this year. $785 billion, heading for a trillion next year. Their word for it was unprecedented. When a ratings agency uses that word, it means they have never had to grade anything like it.

The question is no longer whether the work can be built. It is who is paying for it, and whether they can keep on paying.

Drop your answer in the comments of this week's LinkedIn post.