How the AI stack covers the entire construction workflow — and why new projects have the edge

What “AI covers the entire construction stack” means

A modern toolset — given a serious effort from the development team and an adaptive architecture — can cover not one function but a construction company's entire workflow: on-site control, document flow, finance and management accounting, financial models and project analytics. The key word is adaptivity: the system must adjust to the business's processes rather than break them. That is exactly what separates a working rollout from a slick demo wrapper.

Why this is especially underrated in construction

Because construction is one of the least digitized sectors in the world. The McKinsey Global Institute digitization index ranks construction second-to-last; over 20 years the sector's productivity grew only ~1% versus ~2.8% across the economy, and companies have historically spent under 1% of revenue on IT — three times less than automotive and aerospace. Yet McKinsey estimates that digitization could raise construction productivity by 50–60%. In other words, the lag is the size of the upside.

Where AI actually stands right now — the hype cycle without illusions

Looked at soberly, there is no “universal AI for everything” today. Per Gartner's Hype Cycle for AI 2025, GenAI itself has slid into the “trough of disillusionment” — the market has had its fill of universal wrappers and is grasping the limits; at the peak of inflated expectations now sit AI agents and “AI-ready data,” while the focus is shifting to the fundamentals (AI engineering, ModelOps) needed for stable work in production. Our takeaway is simple: bet not on the hype but on the “boring” foundation — data, processes, engineering.

Why incumbents stall while new projects get an edge

The core paradox: 95% of corporate GenAI pilots never reach a return. Per MIT's research (GenAI Divide, 2025) only ~5% of pilots deliver a measurable effect on P&L. The cause is not the models but a “learning gap”: universal tools don't adapt to real workflows, and they get bolted onto rigid legacy. Tellingly, buying from specialized vendors and partnerships succeed ~67% of the time, while in-house builds succeed three times less often. Incumbents don't lose on resources (per McKinsey's State of AI 2025, companies with $5B+ in revenue scale AI more often than small ones) — they lose on friction: too much legacy that is expensive to rebuild. A new project doesn't carry that weight: it can be built AI-native from scratch. That is the edge.

What European regulations (BIM, RIBA) have to do with it

Mature markets have already made digital the norm at the level of law. In the UK, a BIM mandate has been in force since 2016 (now the Information Management Mandate within the UK BIM Framework and ISO 19650 standards), required for public projects regardless of size; digital stages are baked into the RIBA Plan of Work — from design to operation. It's a ready-made process framework that a system simply integrates into. Montenegro operates in a different format and is moving there at its own pace — but digitization will arrive here regardless, and such a framework will carry it. We want to meet that wave ready.

Where the risk is

Sources

• MIT NANDA, “The GenAI Divide: State of AI in Business 2025” — 95% of pilots with no return, ~5% success, buy 67% vs build. Via Fortune (Aug 18, 2025), Forbes (Aug 26, 2025).
• Gartner, “Hype Cycle for Artificial Intelligence 2025” — press release (Aug 5, 2025).
• McKinsey, “The State of AI 2025” — 88% use AI, ~1/3 scale, $5B+ more often.
• McKinsey Global Institute, “Reinventing Construction: A Route to Higher Productivity” — second-to-last on digitization, +1% vs +2.8%, +50–60% potential.
• UK BIM Framework / ICE — BIM mandate, ISO 19650, RIBA Plan of Work.