Machine Learning Engineer

August 2026 AI construction roundup: estimating agents, takeoff automation, and a busy month for robotics. The AI and construction-tech news that actually matters to contractors this month: Procore's Digital Coworkers, Trimble's AI takeoff for MEP, Higharc's $95M raise, Monumental's bricklaying robots, and new AI procurement rules to watch. DeadFront Team Construction AI Research Welcome to its monthly roundup of AI and construction-technology news worth a contractor's attention. Its filter is simple: does this change anything for the people estimating jobs, chasing bids, or running the work? Deadfront skip the press-release fluff and the "AI will revolutionize everything" takes, verify every item against a primary or reputable source, and give you a plain read on what it means. Here's what landed over the past five weeks or so. On July 23, Procore announced three Digital Coworker packages and expanded its AI agent library to 20 pre-built agents. The Starter pack bundles five ready-to-use agents - Submittal Review, RFI, Contract Review, Deep Search, and Daily Log - while Pro opens up the full library across bidding, safety, quality, scheduling, and risk, and Enterprise adds Agent Studio for building your own. Procore also previewed a capability called Skills that lets a company teach its AI its own standards and processes using plain-language prompts or existing documents, so agents apply your way of working rather than a generic default. What it means for you: The interesting piece here isn't the agent count - it's Skills. The knock on general-purpose construction AI has always been that it doesn't know your division-of-work conventions, your preferred RFI language, or the clauses that always bite you. A "teach it your standards" layer is Procore acknowledging that. If you're already on the platform, the Contract Review and RFI agents are the low-risk place to start; treat their output as a first-pass reviewer, not a signer. Trimble rolled out AI-powered takeoff and estimating features inside its Accubid Anywhere MEP platform. Four pieces stand out: auto-scale setup that identifies sheet names and scale across a plan set, symbol recognition that counts devices like receptacles and switches, conduit auto-routing that computes linear footage including vertical rises and drops, and an embedded AI assistant for natural-language questions about historical pricing. Trimble cites up to 60% time savings on takeoff-specific tasks, better than 50% on manual symbol counting, and says 4,000-plus contractors are already using AI in its MEP estimating tools, with over 3 million symbols detected to date. What it means for you: For electrical and mechanical shops, symbol counting and conduit runs are exactly where estimator hours drain away and where fatigue-driven miscounts creep in. Numbers like "60% faster" are best treated as a ceiling on specific tasks, not your whole bid cycle - but even a partial gain on device counts frees your senior estimators to spend more time on scope interpretation and risk, which is where bids are actually won or lost. If you're an Accubid shop, this is worth piloting on a live job you've already estimated the old way, so you can check the AI's counts against a known-good number. Residential design-to-construction platform Higharc closed a $95M Series C on July 3, led by Insight Partners with Wellington Management joining, bringing total funding past $170M. Alongside the raise, Higharc announced a partnership with building-products distributor US LBM to generate material takeoffs directly from builder floorplans - turning 2D plans into data-driven 3D models and pushing quantities toward the supply chain. What it means for you: This one is squarely a homebuilder story, but the direction of travel matters for everyone. When a distributor and a design platform wire takeoffs straight into materials ordering, the estimate stops being a standalone document and becomes the front end of procurement. Production builders and their trades should expect more "quantities as a service" pitches; the question to ask any vendor is how the model handles plan revisions, because a takeoff that's fast but stale on addendum three is worse than no takeoff at all. Arrakis, founded only in January 2026, came out of stealth on July 22 with $38M in total funding, including a $30M Series A led by Blossom Capital at a reported $140M post-money valuation. The company builds AI agents that plug into operational workflows for industrial firms across aerospace, energy, logistics, manufacturing, and construction, pitching itself at the "last mile" problem - the gap where most AI pilots stall out before they reach real production use. What it means for you: Arrakis isn't a construction-specific tool, and it's early. Note it as a signal, not a buy: serious money is now betting that the payoff from AI is in industrial and field operations, not office chatbots. For contractors, the useful takeaway is the framing - the hard part isn't a flashy demo, it's the last mile of getting an agent to work reliably inside your actual process. That's the bar to hold any vendor to. Amsterdam's Monumental closed a $32M Series B on July 15, led by Khosla Ventures, to scale its autonomous bricklaying robots across Europe and the UK and enter the US market (more from Construction Briefing). The robots are wheeled autonomous vehicles with computer vision and small onboard cranes that lay brick and mortar on active sites. To date the company says its machines have delivered brickwork for roughly 100 homes plus a school, a hotel, commercial units, and stretches of Amsterdam's canal walls. What it means for you: The pitch is explicitly aimed at the masonry labor shortage, not at replacing crews wholesale - these robots handle repetitive straight-run brickwork while skilled masons handle detail and judgment. It's still early and geographically limited, so US masonry contractors don't need to react today. But it's worth watching whether a "robot on the wall" changes how GCs schedule and price masonry scope over the next couple of bid cycles, particularly on large, repetitive residential and institutional work. Two compliance items are worth putting on your radar this month, especially if you do public work or operate across borders. California's AI procurement order. Governor Newsom's Executive Order N-5-26 pushes AI oversight into state contracting, directing agencies to build AI safeguards into procurement and expecting vendors to show governance around bias, safety, transparency, and risk management, with implementing guidance rolling out through the year. If you sell to - or bid public work for - the State of California and any AI touches your deliverables, expect certification-style questions to start appearing in solicitations. The EU AI Act clock. August 2, 2026 is a headline milestone in the EU AI Act, when obligations for high-risk systems and certain transparency rules were originally set to apply (Gibson Dunn's rundown explains the moving parts). Worth noting: a proposed "Digital Omnibus" package could defer some high-risk obligations to December 2027, so the exact timeline is in flux. Most US contractors won't be directly regulated here, but if you use software from a vendor with EU exposure, this is the kind of thing that can quietly reshape a product roadmap. What it means for you: You don't need a compliance department over this, but you do need a paper trail. When a vendor tells you their tool "uses AI," a fair follow-up now is: what does it do with its project data, and can you document how the outputs are produced? Those are becoming procurement questions, not just IT ones. The theme across this month's news is consolidation of a trend, not a breakout. The big platforms are packaging AI into named "agents" you can turn on; the estimating specialists are chipping real hours off takeoff; investors are funding the unglamorous last mile of field and industrial operations; and regulators are starting to ask how these tools actually work. None of it is magic, and the ROI still lives in the boring details - accuracy on your drawing sets, how tools handle revisions, and whether the output survives a senior estimator's second look. Its own take, from building specification-analysis tools day in and day out: the contractors getting value aren't the ones chasing every launch. They're the ones picking one painful workflow - takeoff, submittal logs, spec review - piloting a tool against a job they already know the answer to, and measuring before they scale. That's the whole game. If specification review is the workflow eating your preconstruction hours, Deadfront'd like to show you what purpose-built AI does with your actual bid documents - the non-standard clauses, the buried requirements, the risk hiding on page 400. Every item above was verified against a primary or reputable source as of publication. Sources are linked inline. Last updated: August 1, 2026. New public-works electrical and utility bids across WA, ID, OR, and MT - each with an AI mini-brief of due dates, bonds, and gotchas. Free, every Monday, unsubscribe anytime. Ready to catch million-dollar mistakes? See how DeadFront.AI automates specification review and prevents costly errors.

Higharc raises $95 million Series C to expand AI platform for homebuilders and building materials suppliers. June 30, 2026 Higharc has raised $95 million in Series C financing to accelerate the development of its AI-powered homebuilding platform, bringing the company's total funding to more than $170 million. The financing round was led by Insight Partners, with participation from Wellington Management and existing investors including Fifth Wall, Spark Capital, Lux Capital, SE Ventures, Simpson Strong-Tie, PSP Partners, RXR Arden Digital Ventures, Suffolk Technologies, Vertex Ventures, NC Tweener Fund, and MetaProp. Alongside the funding announcement, Higharc revealed a strategic agreement with US LBM, the largest privately owned distributor of lumber and building materials in the United States, extending its AI platform into the building materials supply chain. Higharc develops an AI-native platform that enables homebuilders to manage the entire design-to-construction process. The company's software replaces traditional CAD-based workflows by generating homes as structured 3D spatial data, allowing builders to automate home design, estimating, sales, and construction documentation. "AI isn't just assisting builders. It's reshaping how builders work, cutting time and cost per job while giving buyers a more personalized experience," said Marc Minor, CEO and co-founder of Higharc. "The builders and distributors leading the way use AI to automate their most complex workflows, unlocking teams to focus on product improvement and customer experience." Unlike conventional AI tools that struggle with spatial reasoning, Higharc's platform incorporates building codes, construction standards, and geometric data into a structured digital model, producing build-ready plans, construction documents, real-time cost estimates, and interactive 3D home models. The company said builders using its platform have reduced product development timelines from months or years to weeks or days, shortened community launch schedules by 25% to 50%, and increased project margins by 10% to 15%. "We evaluated a lot of AI tools, but most produce outputs that require so much correction they're unusable," said Kyle Bear, vice president of research and development at Signature Homes. "Higharc is different. It's grounded in how homes are actually built, so the outputs are usable from day one." As part of its expansion strategy, Higharc introduced AI Estimating, a new product designed for building materials distributors. The solution enables distributors to automatically generate accurate material takeoffs from residential floor plans at enterprise scale. The product is powered by Higharc's AutoTranslate AI technology, which converts two-dimensional floor plans into detailed 3D spatial models using proprietary computer vision models combined with construction-specific logic. The system produces material quantity estimates that align directly with purchasable building products. US LBM will serve as the platform's first distribution partner. "With advanced technology accelerating the building materials industry, what sets Higharc apart for US LBM is the ability to go from a static 2D plan to a precise and dynamic 3D data model in one step," said Jonathan Greene, chief digital and technology officer at US LBM. "That unlocks not just fast, accurate quoting but smarter material planning, the ability to sell the whole home and real-time collaboration with our customers." Higharc plans to use the new capital to expand its AI product development efforts and broaden adoption across the residential construction industry, including deeper integration with suppliers and distributors. "We evaluate AI companies across every sector, and Higharc stands out as a clear leader in applying AI to residential construction," said Josh Fredberg, managing director at Insight Partners. "Leading builders are already standardizing on the platform, and we believe its expansion into supplier workflows has the potential to transform the broader homebuilding supply chain." Founded to modernize residential construction through AI and automation, Higharc works with homebuilders across North America and has been recognized among the Deloitte Technology Fast 500 and Fast Company's Most Innovative Companies of 2026.