Syracuse applied in July for a $1.5 million HUD grant to put AI on architectural-drawing code review, with the Common Council approving the application on July 13. The pressure is concrete: the city issued just 14 single- and two-family permits last year, even as Micron’s ~$100 billion Clay campus is projected to drive demand for roughly 30,000 new homes. Home HeadQuarters CEO Kerry Quaglia told Central Current that permitting delays “trigger a cascading effect” of consulting and engineering costs that developers pass through to renters. HUD is steering applicants toward a shortlist — PermitFlow, Blitz Permits, CivCheck, Permitify — with six municipalities eligible for $300K–$1.5M each. The slowest actor in the lifecycle, the AHJ, is now the one getting federal money to adopt AI — and an industrial mega-anchor is what forced the issue. Central Current →

No fresh independent signal in pre-design today. Two scans of the trade and practitioner press (Aug 6–10) surfaced only vendor blogs and SEO floor-plan tools — nothing that clears the bar. Naming the gap keeps the briefing honest; padding it would not.

Starting September 1, Honolulu’s Department of Planning and Permitting will require residential permit applications to pass through CivCheck, an AI plan-review tool, before a human reviewer sees them (commercial to follow). Clariti, the vendor, reports 55% faster decisions and 58% fewer review cycles from a small Q1 pilot. The signal isn’t the speed — it’s the profession’s response. AIA Honolulu EVP Julia Fink told Aloha State Daily that member firms find value as “a pre-screening and quality control tool” but want answers on data: architectural drawings “represent significant professional expertise and are protected works… Greater transparency is needed regarding how submitted plans are stored, secured, retained, and whether they may be used to train artificial intelligence models.” Mandatory AI at the permit counter turns every firm’s drawings into a training-data question — the summer’s data-sovereignty fight, arriving at the scale of a single submittal. Aloha State Daily →

In an ENR Q&A, Brett Poulos, national director of preconstruction and estimating at Burns & McDonnell, argued the constraint on useful construction AI isn’t models — it’s structure. “Until that data is structured, it’s very hard to manage and it’s very hard to leverage and ingest into an AI ecosystem,” he said. On a live, multimillion-dollar animal-health manufacturing expansion running under near-continuous regulatory oversight, the firm leaned on reality capture, continuously updated models, and AI progress-tracking to catch issues before they reached the field. His hot take: the pull toward integrated project data could accelerate mergers between contractors and design firms, since owning only one data silo caps what AI can decide. This week’s quiet consensus, said again by a practitioner — the bottleneck is data plumbing and ownership, not model IQ. ENR →

KODE Labs won FedRAMP High authorization for KODE OS, its AI-driven building operating system, via Palantir’s FedStart program — clearing it for the government’s most sensitive facilities. It is already running under a $14.35 million GSA Public Building Service contract across 150 GSA-managed buildings in the Washington, D.C. area. Co-founder Edi Demaj told Facilities Dive the platform lets teams “prioritize based on real data” rather than paper inspections, aimed at GSA’s estimated $26 billion deferred-maintenance backlog; the company is now eyeing DoD and state/local/education work. Security certification is a boring milestone that matters — it is the gate that lets AI building ops into the public estate, where the maintenance backlog is largest. No energy or savings outcomes were reported yet. Facilities Dive →