Honolulu's Department of Planning and Permitting this week launched Priority Review, routing applications pre-screened by its AI plan-check tool, CivCheck, straight to the front of the permit queue. The reported numbers are real: average review time down from 73 days to 32.5, plan-review cycles from 3.4 to 1.4. The honest part is the denominator. Spokesperson Davis Pitner told the Star-Advertiser that participation is voluntary, roughly 130–140 users have enrolled, and just 19 completed applications sit behind those figures — a sample he calls "relatively small." CivCheck becomes mandatory later this year, residential first. The tool isn't the bottleneck; getting anyone to use it is. Honolulu Star-Advertiser →

The programming-and-massing beat again produced only vendor blog posts and syndicated press releases — nothing clearing our non-advertorial bar. That absence is itself a data point: early-design AI remains a category marketed far more than it is independently reported. We would rather name the gap than pad it.

On the AEC Business podcast this week, Endra co-founder Niklas Lindgren explained why the a16z-backed Stockholm firm built its own granular 3D data model rather than a Revit plugin: Revit's underlying model, he argues, is too coarse for deep automation. Endra's "spatial AI" routes conduit and ductwork clash-free and models entire electrical systems — receptacle to transformer — in a single source of truth, working through RIBA Stage 4 with engineers kept in the loop rather than handed black-box output. The provocation worth sitting with: Revit doesn't disappear — it becomes an "orchestration layer" that specialist tools build around. An electrical module ships September 14. AEC Business →

Brasfield & Gorrie — Birmingham-based — took ENR's 2026 Southeast Contractor of the Year on a record $7.7 billion in 2025 revenue, and tucked into the profile is the AI line: the firm's internal platform, Magnus 2.5, logged more than 70,000 interactions in its first two months, with a broader rollout still to come. Around it, 631 active projects worth $29.4 billion, more than $2 billion in data-center and hyperscale work, and a 0.50 experience-modification rate across 220-plus jobsites. The interaction count is self-reported, and ENR doesn't say what Magnus actually does — but the tell is that a regional GC now treats internal AI as standing infrastructure, not a pilot. ENR →

Writing in International Airport Review, DFW Airport's Paul Puopolo detailed a multi-year overhaul of operations and maintenance across 6 million square feet and 26.9 square miles — a footprint drawing 150,000-plus work orders a year inside a $12 billion expansion. The target: cut maintenance cost per square foot 20–25% by 2030, built on an "Asset Health 360" platform and dispatch layer developed with BCG. What makes it worth reading isn't the AI; it's the candor. Puopolo reports frontline adoption above 90% while admitting the digital-twin ambition had to be scaled back first, that legacy data was fragmented, and that the operating model — not the technology — "is the harder part." International Airport Review →