Construction AI: the bankable wins - robot bricklayers, reality capture, and a cheaper Flash frontier

The WLTR ("Walter") wall-laying robot is building 27 homes in Durham as part of a scheme led by JT Lifestyle Homes. Developed in the Czech Republic with GreenBuild, Walter lays up to 200 m² of masonry a day, builds walls to 3.5m (with plans to extend to 5m) without scaffolding, and operates in rain, wind and temperature extremes. It uses DryFix adhesive foam rather than traditional mortar. Future deployments are lined up at a former Post Office redevelopment in Hull, new homes at Haywood Park in Southampton, and an apartment refurbishment in Nottinghamshire. The machine has been in UK trials since 2025 - what's new this week is the renewed traction and fresh deployment news.

The framing is the part to study. With the trade's average worker age at 46, too few young entrants, 35,000+ unfilled roles and a 1.5-million-homes target, the pitch is "automation complements human labour" and "fills the gap" - not headcount reduction. That is a far more durable narrative with clients, unions and planners than a productivity-and-redundancy story.

The most bankable AI use case on UK sites remains reality capture and progress tracking, and the case data is now substantial. Sir Robert McAlpine and Vinci's joint venture (IHP) ran Buildots - 360° site video automatically mapped to BIM and the programme - on the Royal Bournemouth Hospital project, using it for QA, billing and delay-risk mitigation; Sir Robert McAlpine has adopted Buildots as a preferred partner across large areas of live work. Separately, Vinci Construction UK reports saving more than 6,000 hours annually on documentation alone using OpenSpace for automated progress capture and remote verification.

These are the unglamorous wins that pay back. They cut admin load and reduce disputes by giving every party the same visually-verified ground truth - rather than promising autonomy or replacing judgement. They are also the easiest pilots to justify to finance, because the time-saved baseline is measurable against the manual process they replace.

Launched at Google I/O 2026, Gemini 3.5 Flash is now the default engine behind the Gemini app and AI Mode in Google Search globally. The headline is the price/performance inversion: Flash now out-scores the prior Gemini 3.1 Pro flagship on several agent benchmarks (76.2% on Terminal-Bench 2.1, strong on MCP Atlas, Toolathlon and Finance Agent v2 - where it edges Claude Opus 4.7 and GPT-5.5), runs at roughly 289 tokens/sec (about 4× faster), and costs $1.50 input / $9.00 output per 1M tokens - around 40% cheaper than 3.1 Pro. The caveat: on deep academic reasoning (Humanity's Last Exam, ARC-AGI-2), Gemini 3.1 Pro still leads.

For construction, the relevance is in high-volume, repetitive workflows - document triage, classification, RFI summarisation, compliance-record search at scale. The cost per task on those workloads just dropped meaningfully. The strategic implication is the one we keep returning to: build a routing layer that swaps to the cheapest model clearing your quality bar, rather than hard-wiring one frontier model into every workflow.

Among the I/O launches, Gemini Spark is the one most relevant to everyday operations: a personal AI agent that carries out multi-step tasks on your behalf, including asynchronous, background execution - it keeps working when you're away from the keyboard. Independent hands-on coverage (PCMag) tempered the marketing but confirmed the core capability: queue a multi-step task, walk away, come back to a result.

A useful counter-current in this week's construction-AI chatter: a recurring reference to the Builder.ai collapse - the "AI" app-building service later reported to have leaned heavily on hundreds of human engineers behind the scenes. It's resurfacing as shorthand for a healthy instinct: interrogate any "AI does it end-to-end" claim and ask precisely what is automated versus quietly human.

This pairs directly with the trust theme from the UKCW debrief. The fastest way to lose credibility with a sceptical construction audience is to oversell autonomy. The fastest way to build it is to be specific about where the AI helps, where the human stays in the loop, and what the audit trail looks like. Vendor and survey numbers (including the reality-capture and Houzz figures above) should be cited as case-study or self-reported, not as audited fact.