Area Development’s Q3 site-selection report makes a quiet but consequential argument: AI has moved the front end of location decisions weeks upstream. A pre-AI shortlist — 200 communities filtered to 25–30 by a team of analysts over two weeks — now happens “in an afternoon,” before an RFI is drafted, before a community knows a project exists. The model reads what economic-development offices do not advertise: city-council minutes, zoning disputes, utility-capacity debates, permit timelines. The sharpest point reframes the whole data-readiness conversation: sites presented as PDF brochures or interactive GIS are harder for AI to parse than plain HTML tables — and a site the model cannot read may be invisible to the shortlist entirely. Source →

Fourth honest gap in two weeks. Targeted scans returned only vendor launch pages, competitive comparison microsites, and “top AI tools for architects” listicles — nothing that clears the non-advertorial bar. The vendor-marketing monoculture at the drawing board’s front edge is itself the story: the reporting and the money are upstream in feasibility and downstream on the jobsite, not here. We would rather say so than pad.

Architosh’s AIA26 show-floor coverage puts Chaos — which it ranks among AEC’s “big four” software houses alongside Autodesk, Trimble and Nemetschek — at the center of the visualization story. V-Ray 7 Update 4 embeds the Veras AI engine natively in the V-Ray Frame Buffer, so architects can test design and material ideas by text prompt inside the render tool, plus a machine-learning feature that turns a single 2D reference photo into a physically accurate, fully textured PBR material. The more consequential move may be the least glamorous one: Chaos Glyph, a Revit plugin automating sheet and view creation, tagging and dimensions — documentation labor, not pretty pictures. Source →  ·  V-Ray 7 detail →

Two days after this digest noted capital migrating away from jobsite robots toward back-office agents (F-Prime’s chart: ~$211B into AI versus ~$18–21B into applied robotics in 2025), the robots answered. Amsterdam’s Monumental raised a $32M Series B led by Khosla Ventures for autonomous bricklaying — 150+ electric robots now laying brick to millimeter precision on live sites in the Netherlands and UK, coordinated by its “Atrium” platform, with walls built for 100+ structures and a first US pilot this year. Founders Salar al Khafaji and Sebastiaan Visser sold their last company to Palantir and carried its forward-deployed-engineer model into robotics; contractors pay for a finished wall, not the hardware. A day later, Austin’s TerraFirma banked a $115M Series A for AI-run autonomous heavy equipment. One week’s “money is leaving robotics” is the next week’s counter-signal. Source →  ·  TerraFirma →

Writing in Propmodo, GVA Real Estate Group CEO Alan Stalcup lays out how his multifamily firm automated four full workflows — leasing, collections, maintenance and renewals — and cut payroll roughly 20%, not through layoffs but by not rehiring roles as they turned over. He reports the leasing loop alone lifted signed leases 67% (37 to 62 a week). It is a rare named-operator account with hard numbers in a segment starved for them. Two caveats keep it honest: the figures are self-reported, and Stalcup has published a near-identical version of this playbook before — so read it as an operator making a case, not an independent audit. Still, it is a concrete data point on where owner-side automation is landing: the leasing office, not the building’s systems. Source →

Gemini 3.5 Pro slipped its rumored July 17 launch again; Bloomberg reports the model fell short of internal coding benchmarks, and Alphabet’s stock fell ~4%. For AEC this is the flagship flagged for two-million-token whole-model queries and sketch-to-CAD — a coding shortfall tempers the near-term “talk to your entire BIM” promise. The Verge →  ·  Detail →