Scan-to-BIM in 30 Minutes, Agentic Project Management, and the Supply Chain You Didn't Audit

NavLive, developed by academics from the University of Oxford, has won Best Use of AI at the Digital Construction Awards 2026. Their AI-powered handheld LiDAR scanner generates RICS-grade 1:100 surveys, 2D floorplans and 3D point cloud models directly on site - no cloud upload required.

The standout case: BW Workplace Experts scanned an eight-storey bank in London in under 30 minutes, generated floorplans instantly and had a full BIM model within hours. That's a process that would traditionally take days or weeks compressed into a single site visit.

But, the adoption story is what makes this significant. AWW Architects - one of the UK's leading multi-disciplinary practices - has adopted NavLive specifically to meet Building Safety Act compliance requirements. The technology isn't being trialled in a lab. It's being deployed on live projects to meet regulatory obligations. NavLive's GBP 4m funding round confirms serious investment interest beyond the award itself.

In January 2026, Procore acquired Datagrid AI, an agentic AI platform built specifically for construction. Datagrid's agents can perform deep searches across fragmented data sources, validate RFIs, run scope checks, handle prequalification, conduct audits and generate daily reports - all autonomously.

Procore Agent Builder is now in open beta. Teams can build custom agents via natural language prompts. The pre-built RFI Creation Agent reportedly reduces RFI resolution from days to seconds.

This is the platform-level shift the industry has been waiting for. Not AI features bolted onto existing software, but true agentic workflows embedded into the project management layer. The distinction matters. A chatbot that answers questions about your project data is useful. An agent that validates an RFI against the latest drawings, checks the spec, and drafts a response is a fundamentally different proposition.

AVEVA has announced a new lifecycle digital twin architecture built on the NVIDIA Omniverse DSX Blueprint. The collaboration targets large-scale data centres - AI Factories - using domain-specific simulations, digital visualisation and collaborative design tools to maximise GPU efficiency. The methodology mirrors approaches used in major engineering, procurement and construction projects.

Digital twins aren't new. But, this collaboration signals something important about where they're heading. These aren't passive replicas. They're AI-driven systems that actively orchestrate decisions across the asset lifecycle - from design through construction and into operations.

AMC Bridge has built an MCP Connector for Autodesk Platform Services and Procore - a demonstration of how the Model Context Protocol enables AI agents to work across both platforms simultaneously. Natural language queries can now span project documents, drawings and cost data across systems, replacing brittle point-to-point integrations.

This sits alongside the Figma MCP story from the wider AI world this week. Figma launched its own MCP server in open beta, and Cursor immediately extended it to generate full front-end components from existing design systems. The pattern is consistent: professional software tools are opening up to AI agents. Construction platforms are already part of that shift.

The Building Safety Act 2022 is increasingly cited as a practical accelerator for AI adoption in the UK - particularly for audit trails, decision documentation and compliance evidence. Construction firms are deploying AI project management tools not primarily for productivity, but to produce structured, auditable records.

This connects directly to the NavLive story above. AWW Architects adopted scan-to-BIM technology specifically to meet BSA compliance requirements. The pattern is emerging: regulatory pressure is becoming a stronger commercial driver for AI adoption than productivity gains alone.

Despite the award wins and platform launches above, only around one in five UK construction firms are currently using or planning to use AI. The adoption gap is most pronounced in smaller contractors and specialist trades.

That's a striking number set against the stories in today's brief. NavLive is winning awards. Procore is shipping agentic workflows. MCP connectors are bridging platforms. But, 80% of the industry hasn't started. The gap between the firms that are moving and those that are waiting is widening - and it's becoming a competitive gap, not just a technology one.

The AI development world had a proper security scare this week. LiteLLM - a widely-used Python library that routes requests to AI models - had versions 1.82.7 and 1.82.8 on PyPI compromised via the CEO's hacked GitHub account. The malicious payload was designed to exfiltrate cloud credentials, SSH keys, Kubernetes configs, CI/CD secrets and shell history.

The package was quarantined quickly. But, the blast radius was significant. LiteLLM sits as a dependency in a large chunk of Python AI projects. The security community's reaction was clear: in an agentic AI world, the entire filesystem becomes an attack surface, because any file that enters an AI agent's context can become a vector.

For construction technology teams, this is a directly practical issue. If your organisation is using AI-integrated software - project management tools, document automation platforms, AI assistants - those tools almost certainly sit on a stack of Python dependencies. The practical response is straightforward: pin dependency versions, audit what your AI tools depend on and make sure your vendor has a clear security posture.

OpenAI has confirmed that Sora, its video generation product, is shutting down. The stated reason is resource reallocation - OpenAI has finished initial development of its next major model, codenamed "Spud," and is redirecting compute.

Sora launched to enormous fanfare in early 2024. It didn't survive the cost-versus-engagement maths. The lesson is familiar: side products are expendable when the core model race demands resources. OpenAI is narrowing its focus around general models and infrastructure. The tools built on top of frontier models will increasingly be third-party - which creates more room for specialist construction AI at the application layer.

Anthropic has shipped Memory 2.0 for Claude Code. AI coding tools can now retain project context across sessions - architecture decisions, naming conventions, prior discussions - without the developer having to re-explain the codebase each time.

For construction technology teams and the software vendors serving them, this is a practical upgrade. AI coding assistants used on multi-week projects are meaningfully more useful when they maintain context throughout. And given the LiteLLM story, it raises a good question: as AI tools remember more, what exactly are they retaining, and for how long?