DCW landed: asset intelligence, the lifecycle battleground, and a calmer conversation

A useful week. The kind that gives you a clean line to walk through the next quarter with.

Digital Construction Week opened at ExCeL on Wednesday and ran through Thursday. 9,000+ delegates, 230+ CPD sessions across ten stages, 150+ tech brands. The headline numbers matter less than the tone. The breathless "AI will change everything" pitches that dominated the hall two years ago have been replaced by a calmer, more evidence-led conversation about succession, governance and what an audit trail actually looks like. Sessions on workforce succession (Nemetschek's James Chambers), succession-proofing BIM through retirement (Tektome's Naoki Kitamura), and design-phase collaboration (Paul Drayton with WSP, BDP and AECOM) all shared a thread - how do you retain knowledge, decisions and value as the tooling layer accelerates and senior people retire. The "Would I AI To You?" session with Vicki Reynolds (ONE Creative) and Dan Rossiter (BSI) on the Inspire Stage was the most pointed example: a deliberate format for stress-testing AI claims against reality.

The single cleanest new idea came from Glider's Nick Hutchinson on the Digital Operations Stage. His argument: the next discipline for the built environment is not better information management - it's asset intelligence. The two are related and distinct. Information management is the discipline of the asset's record. Asset intelligence is the discipline of the asset's decisions - using the asset's history, condition, obligations and trajectory as the thing decisions are made against. One records. The other recommends. That framing gives buyers a four-part vendor filter that's already worth more than any RFP template: what decision does this drive, on which asset, with what evidence, and what's the human approval gate?

The big strategic story landed on the eve of the show. Procore relaunched its Common Data Environment on 1 June with Datagrid-powered agentic AI coworkers built in - agents that surface answers from existing project records before a new RFI gets raised, spot discrepancies between approved designs and field execution, and accelerate issue resolution with citation-backed responses and human approval gates throughout. The launch geography is the signal: UK and Ireland first, ahead of broader EMEA, with an EU Data Zone due in autumn. Procore picked the market with the toughest current information-management pressure to launch first. Combined with Bluebeam Max running Claude inside Revu via MCP since 19 May, and Autodesk's MaintainX acquisition pushing them into the Operate stage of the lifecycle for the first time, the strategic battleground has moved decisively. The dominant AEC software vendors are racing to own end-to-end lifecycle, not to compete on point-tool features. For any UK contractor or consultant procuring AI capability over the next two quarters, the question is whether to commit deeper into one vendor's emerging stack or defend best-of-breed flexibility and accept the integration cost. Neither answer is obviously right. Procrastinating is no longer defensible.

Underneath all of that, the wider AI stack delivered. On Friday 28 May, Anthropic shipped Claude Opus 4.8 with native Dynamic Workflows - up to 16 subagents in parallel and 1,000 per run, with the model verifying outputs before reporting back. It tops GPT-5.5 on SWE-Bench Pro by ten points and Anthropic claims it's four times more honest about its own uncertainty. Multi-step orchestration that previously needed bespoke plumbing is now a model-native feature, and the build-vs-buy maths on document and compliance automation just shifted. The day after, Anthropic raised $6.5bn at a $965bn valuation - ahead of OpenAI on private-market value for the first time - and confirmed Mythos-class frontier models for wider release "in the coming weeks." And in the middle of DCW week, Anthropic made Claude Platform generally available on AWS, billing through AWS Marketplace in Claude Consumption Units, with managed agents, webhooks and self-hosted sandboxes that keep tool execution inside the customer's own AWS account. For any UK enterprise already on AWS, that quietly removes most of the procurement and data-residency friction that's been blocking enterprise rollouts.

A few smaller items worth holding onto. The EY × University of Cambridge × NVIDIA "intelligence layer" report - published on 6 May but the right thing to read on a train into DCW - reviewed 100+ construction productivity studies and put rework at up to 15% of total project cost. That's the number to use when a finance director asks why bother. AI takeoff and estimating quietly became the most defensible use case for UK SME and mid-tier contractors: Togal.AI reports 50-80% takeoff-time cuts at up to 98% accuracy; Cambridge-based Kreo claims 85-92% accuracy on simple geometry. xAI expanded Grok Build to all SuperGrok subscribers and shipped the cleverest architectural idea in mainstream agentic coding for months - 8 subagents working isolated Git worktrees in parallel. MiniMax M3 arrived with frontier-pricing claims and a 1M-token context window, though the weights aren't out yet and the benchmarks are vendor-reported. NVIDIA unveiled Nemotron 3 Ultra at Computex - 550 billion parameters, genuinely open-weights, the first credible mid-frontier on-prem option. The week ending 1 June also saw NavigateAI raise $25m and LightTable $22m for AI preconstruction - the capital is going where the demos are.

Pull all of that together and the test for June is the same one that ran through every session at ExCeL. Pick one workflow. Defend the quality bar against a sceptical board. Keep a human approval gate. Measure time saved against a realistic manual baseline. Avoid the temptation to run three pilots in parallel; the firms reporting the strongest gains are the ones that did one workflow well and only then scaled. And then write it up in your own voice for LinkedIn - specific named moments, named conversations, named numbers. The slop-throttling point from a fortnight ago still matters. Authentic, expert-led case studies are reportedly outperforming where they used to. Be the case, not the announcement.