Hello Project AI enthusiasts,
This week’s story is about Kier Group putting Microsoft Copilot into its everyday working environment, from office teams through to field-based roles. The useful signal is not simply that another major contractor is buying AI software. It is that Kier is treating AI as part of how the business operates, not as a side experiment.
The Editor’s Picks widen the lens to frontier-model honesty, humanoid robot production, AI contract risk and the overdue rebuild of AEC design software. Alongside that, this issue includes practical signals from AI in AEC and project delivery, five curated links directly in the newsletter, xAI’s new coding CLI, and a quick guide for using AI well.
📌 Featured This Week
Kier Group Embeds Microsoft Copilot Across the Business
Kier has expanded its Microsoft partnership to roll Copilot into its everyday working environment, with AI positioned as part of the contractor’s wider growth strategy. CEO Stuart Togwell says the integration is already supporting productivity and client delivery, while the move follows a set of senior digital and technology appointments that frame Kier as a “naturally digital” business.
The useful lesson for project delivery leaders is that adoption is now moving from innovation teams into operating models. A 10,000-user rollout at a Tier-1 UK contractor is the pattern to watch: AI stops being a demo when it is embedded into daily workflows, supported by leadership, and connected to the way projects are actually delivered.
🔗 Editor’s Picks
This week's must-read stories on AI, project delivery, and infrastructure:
Anthropic Launches Claude Opus 4.8 With Gains in Coding and Honesty
Claude Opus 4.8 is the model story with the most practical project-controls hook this week. The headline is not only better coding and reasoning; it is the claim that Opus 4.8 is much less likely to let flawed code pass without comment, which matters for teams asking AI to review schedules, cost logic, assumptions and delivery plans. Read the blog
Figure Ramps Humanoid Production to One Robot Per Hour
Figure’s move from one humanoid a day to one per hour is another sign that physical AI is shifting from demos to fleets. Construction should still be cautious about jobsite claims, but the direction is important: repetitive logistics, inspection and controlled offsite environments are becoming more plausible early adoption zones. Read the blog
EU AI Act Pressure Meets a Contract Gap
The EU AI Act’s high-risk obligations are approaching, but construction contracts still do not clearly allocate responsibility for AI tool selection, configuration, data inputs or output errors. That gap matters because design, safety, workforce monitoring and automated decision-making use cases can quickly become legal and commercial risk, not just technology risk. Read the blog
a16z on Revit: AI’s AEC Opportunity Is Review, MEP and Rework
a16z’s essay is useful because it names a frustration many AEC teams live with every day: the built environment is still designed on software foundations shaped decades ago. The AI opportunity may not be instant generative buildings, but the less glamorous work of review, coordination, MEP reasoning and rework reduction. Read the blog
🔨 AI in AEC and Projects
Here are some intriguing reads, specifically for AEC and project delivery professionals:
Figure AI runs humanoids for 200 continuous hours: Five Figure 03 robots reportedly processed nearly 250,000 packages with zero mechanical failures, a useful signal for anyone tracking future site logistics and repetitive material-handling tasks.
PMWEB unveils AI-driven platform upgrade: PMWEB 2026 adds an AI Assistant and an “Intelligence Control Framework” for capital programme owners, showing how AI is entering project information management rather than staying in generic productivity tools.
NextEra and Dominion point to the power constraint behind AI: The proposed utility mega-deal underlines how AI data-centre demand is becoming a power, grid and infrastructure story, with direct consequences for energy and construction pipelines.
Stop treating AI in construction as a bolt-on: PBC Today’s piece argues that AI should be embedded into business structure, data governance and workflow redesign rather than managed as another isolated innovation project.
How drones and AI are mapping heat loss across UK neighbourhoods: Drone-based thermal mapping is a practical example of AI being used to turn building-performance data into neighbourhood-scale retrofit intelligence.
🧭 Curated Links
Five stories worth reading before you get back to delivery:
California’s AI Bill Wave Clears a Key Deadline
California’s latest AI legislative update shows how quickly state-level regulation is becoming part of the operating environment for technology users, not only model developers. For project organisations with US exposure, the practical watchpoints are transparency, automated decision-making and any AI use connected to critical infrastructure. Read more
Microsoft reportedly drops Claude Code after token costs spiral
Microsoft has reportedly cancelled its internal Claude Code pilot after token costs ran ahead of expectations. It is a useful warning for every project organisation trialling AI tools: if usage-based tools become more expensive as they become more useful, then spending caps, adoption modelling and weekly cost monitoring need to be in place from day one. Read more
NVIDIA launches verified agent skills
NVIDIA’s Verified Agent Skills framework points toward a more auditable way to manage enterprise AI agents, with human review, automated scanning, skill cards and cryptographic signing. For AEC firms connecting agents to project systems, this is the governance shape to watch. Read more
Cisco warns that single-turn AI safety tests are misleading
Cisco’s research suggests frontier models can look far safer in single-turn tests than they do under multi-turn attacks. That matters for project teams giving AI access to contracts, financial systems or project data, because real misuse rarely happens in one obvious prompt. Read more
DeepSeek’s permanent price cut escalates the token price war
DeepSeek’s 75 per cent V4-Pro price cut keeps pressure on AI pricing just as enterprise usage rises. The strategic question for AEC firms is whether lower model costs translate into cheaper tools, or whether integration, governance and workflow redesign remain the real cost centres. Read more
🔧 Tool of the Week
xAI ships Grok Build
xAI’s Grok Build agentic coding CLI is now in early beta for SuperGrok and X Premium+ subscribers. It runs in the terminal, creates implementation plans for approval, edits files, executes shell commands and manages dependencies, giving AEC technology teams another option to compare against Claude Code, Codex and internal automation tools.
💡 Quick Win
Use the “10 Golden Rules” before trusting an AI output
The “10 Golden Rules for Using AI Well” is a useful reminder that better AI work still starts with clear instructions, context, constraints and checking. Treat it as a lightweight pre-flight checklist before using AI to summarise project information, draft client material or support delivery decisions.
👀 Also This Week
Industry expert Phil Chell discusses the transformative impact of AI on construction, project management, and professional services [Podcast]
📰 Event of The Week
RICS Wales Conference 2026, Cardiff, 16 June
RICS Wales Conference returns to Sophia Gardens in Cardiff for a full-day built-environment programme covering housing, infrastructure, business rates, energy efficiency, sustainability, mental health, economic trends and cost management. It is a timely one for teams watching how regional policy, infrastructure priorities and commercial pressure are shaping project delivery after the Senedd election cycle.
This week's newsletter is sponsored by:
Movar Reply is helping the construction industry make smarter decisions with AI. Their tools connect your planning systems, cost data, and risk registers into one unified data environment—then use AI to surface insights you'd never find manually. They even offer free AI tools for project reporting.
If you're curious about what AI can actually do with your project data, they're worth a look.
👉 Explore Movar Reply's AI tools: https://movar.group/
Till next time,
Project Flux
All content reflects our personal views and is not intended as professional advice or to represent any organisation.