64% In, 20% Deep: Architecture's AI Reality Check

RIBA Journal's latest assessment of AI in architecture found that nearly two thirds of practices have experimented with AI, but only a fifth have embedded it in daily workflows. That gap between trying and committing showed up everywhere this week. New data puts measurable AI impact among contractors at 38%, up from 17% a year ago, yet the majority of AEC professionals remain on the sidelines. Meta released an open-source model that optimises concrete mix design for strength and sustainability. Huawei demonstrated real-time AI control of cement production lines. And Anthropic assembled twelve of the world's largest technology companies around a model it considers too powerful for public release. The question for project delivery leaders is whether the distance between those embedding AI and those still experimenting is becoming too wide to close.

RIBA Journal's 2026 assessment, authored by Phil Bernstein and Vincent Guerrero, finds 59% of UK practices now using AI, up from 41% in 2024 and above 80% among larger firms. Only around 20% have moved beyond ad-hoc prompting into structured workflow integration.

Four areas are accelerating: workflow automation for specification and visualisation (with tools like Autodesk's Forma enabling earlier design exploration), building performance engines in structural engineering, LLM-based code generation for computational design, and broader access for smaller firms where adoption sits at around 48%.

The consensus from RIBA's 2026 AI Summit is that AI should function as a design companion, enhancing creativity without replacing judgement-led design. Hamza Shaikh at Gensler cautioned against letting tool adoption outpace intentional approaches. Fewer than one in five practices have invested in R&D, and only 15% have formal AI policies. For project delivery professionals, architecture's story is a useful proxy: adoption is rising, but the distance between experimentation and integration remains the defining challenge.

Read the full analysis →

This week's must-read stories on AI, project delivery, and infrastructure:

Meta Built an AI That Designs Concrete. It Signals Where Construction Is Headed.

Meta’s BOxCrete model shows how AI is moving into core construction workflows, optimising concrete mixes to improve performance while reducing risk and carbon. In a Minnesota data centre, it helped concrete reach structural strength 43% faster. In parallel, Anthropic’s Project Glasswing signals how quickly frontier AI is advancing, with models already uncovering thousands of software vulnerabilities. Different domains, same takeaway: AI is moving beyond experimentation into real-world impact at speed.

Read more →

Huawei Deploys AI in Cement Production: Coal Consumption Down 1% Per Line

At MWC 2026, Huawei showcased its AI deployment at Conch Cement, where an optimisation solution analyses over 100 production parameters in real time to control free calcium range. The result: a 1% reduction in coal consumption per production line, saving approximately US$250,000 annually per line. According to Huawei, AI is shifting from a supporting function to a core role in construction materials manufacturing. As more producers adopt similar technology, cost and carbon savings should flow through to project budgets.

Read more →

AI Adoption in Construction Doubles: 38% Now Report Measurable Impact

New data shows 38% of contractors now report measurable business impact from AI, up from 17% a year ago. Usage is concentrated in estimating (24% of firms), bid management (22%), and safety monitoring, with automated estimating achieving 85-90% accuracy. Pilot firms report 30-50% decreases in admin hours through automated field reports and invoice processing. But only 27% of AEC professionals overall currently use AI, with data security (42%) and cost (33%) as the top barriers. The gap between early adopters and the rest is widening.

Read more →

Anthropic Launches Claude Managed Agents Public Beta

Anthropic opened a public beta for Claude Managed Agents, letting developers deploy production-ready AI agents in days. Users configure the task, tools, and guardrails; the platform handles execution, security, and access control. Agents can operate independently for hours without losing state, with multi-agent coordination also in preview. Pricing is $0.08 per agent session-hour on top of standard API usage. Early adopters include Notion, Rakuten, Asana, and Sentry. According to Anthropic, Rakuten deployed agents across five departments in approximately a week each.

Read more →

Qonic: Native Drawing Generation

Qonic has rolled out native drawing generation in its browser-based BIM platform. AEC teams can produce fully annotated floor plans and sections directly from BIM or IFC models, with dimensions, room tags, and annotations generated automatically, and export as DWG without leaving the platform. Documentation updates automatically as the model evolves, removing the repetitive manual sheet-building that has been a persistent pain point in BIM workflows for years. Qonic is also presenting at NXT BLD 2026 in London, 13-14 May.

Learn more →

Caveman Mode: Cut Claude API Token Costs by Up to 75%

Developers found that forcing Claude into extreme brevity rules (3-6 word sentences, no articles, no preamble) cuts output token usage by up to 75%. Output tokens on Claude Sonnet cost five times more than input ($15 vs $3 per million). Add these rules to your system prompt for any API integration where verbose responses are unnecessary. This is a developer/API technique not applicable to Claude.ai chat.

How AI Will Reshape Construction: The View From Silicon Valley, London, 21 April 2026

Bricks & Bytes brings Professor Martin Fischer, Stanford's CIFE Director and the pioneer of Virtual Design and Construction, to London for a closed-door evening with senior construction leaders. Spaces are limited and approval is required. Powered by Cogital and Speckle.

Details →

Catch up on other exciting stories shaping AI and project delivery:

Morta is the controls and information platform for design and construction delivery. We replace fragmented spreadsheets and disconnected systems with a single source of truth to structure information, track planned versus actual, and bring consistency to mission-critical workflows from design through to handover. Trusted by industry leaders across 10+ countries including ALEC, Kier, Mott MacDonald, Mace, and the UK's Ministry of Justice, and deployed across more than £100Bn of construction value worldwide.

Try Morta for free. Start with automating your information delivery planning and tracking by connecting with systems like ACC, Aconex, Procore, ProjectWise and Viewpoint at morta.io

That's it for this week. See you next week

Yoshi & James

_{All content reflects our personal views and is not intended as professional advice or to represent any organisation.}