Fix the Workflow First. Then Add the AI.

Bluebeam's Steve Smith wrote one of the more direct pieces on UK construction AI this week, and it's worth reading in full. The headline findings from Bluebeam's own research: 84% of AEC firms plan to increase technology investment in 2026. Yet only 11% describe themselves as fully digital across all project phases. AI adoption is strongest in design. It drops sharply on-site.

The argument Smith makes - and it's a good one - is that AI can't rescue a workflow built on broken processes. The Get it Right Initiative puts the cost of avoidable errors in UK construction at somewhere between £10bn and £25bn per year. The root causes are familiar: poor information flow between site and office, inconsistent naming conventions, mixed paper-and-digital handovers. These aren't technology problems. They're behaviour and process problems.

The practical implication is uncomfortable for a lot of firms. You can invest in AI tools and see zero improvement if the underlying workflow is broken. Inconsistent data in means inconsistent decisions out. The technology doesn't fix the culture. The culture has to be ready for the technology.

But, the firms that do have their processes in order - the 11% who are genuinely digital across all phases - are in a very different position. For them, AI is an accelerant, not a solution to a problem they're still diagnosing.

On 18 March 2026, the UK government published its long-awaited Report on Copyright and Artificial Intelligence, alongside an Impact Assessment. The short version: no major reform. The government is maintaining the current legal framework for now, declining to introduce new legislation on AI-generated content, training data, or IP ownership.

For AEC firms, the implications are specific. If your firm is using AI to generate design concepts, produce BIM content, process drawings, or create technical documentation, the question of who owns the output is still legally murky. The existing copyright framework wasn't designed for AI-generated work. And with the government declining to clarify it, that ambiguity stays in place.

It's not a crisis. But it's the kind of thing that creates problems when it intersects with client contracts, novation agreements, or professional indemnity cover. Fieldfisher's analysis is worth reading if your legal team hasn't already flagged it.

The most significant UK product launch in recent weeks is one that deserves more attention than it's received. Construction AI went live on 4 March 2026 - a project management platform built entirely around AI and targeting the 98% of UK construction firms that have been priced out of enterprise software like Procore or Autodesk's full stack.

The founder is Steve McKenna: Chartered Builder, MCIOB, 30 years of industry experience, and - notably - no software development background. He built the platform entirely in collaboration with AI tools. The result is over 700,000 lines of code, 186 database tables, and more than 60 AI-powered tools across 22 modules.

The capabilities are squarely aimed at the daily pain points of smaller contractors:

- AI-generated RAMS - Risk Assessment and Method Statements produced in minutes, not days

- Automated construction programmes - generates compliant programmes from project inputs

- OCR-powered drawing management - reads, extracts, and catalogues construction drawings automatically

- Industry knowledge base - trained on UK construction standards, CDM regulations, and BSR requirements

- Intelligent tender analysis

Introductory pricing is a £2,000 lifetime licence for two seats. That's a founding-member price, not the long-term model - but it signals who they're trying to reach.

The PlanOps relevance here is significant. This is the clearest evidence yet that practitioners with deep domain knowledge - and no software engineering background - can build production-quality AI tools. The model of practitioner plus AI is exactly where the market is heading. And it's arriving faster than most people expected.

The Archdesk analysis of agentic AI in construction is worth reading alongside the Bluebeam piece above, because both arrive at the same conclusion from different angles.

Agentic AI - systems that plan, reason, and sequence tasks without constant human input - is being applied to construction workflows in five meaningful ways right now: dynamic scheduling that monitors real-time progress and automatically reschedules dependent tasks; change order automation that detects scope creep and routes approvals; predictive procurement that tracks material consumption and triggers just-in-time purchase orders; safety intelligence that analyses incident patterns; and continuous financial oversight that flags margin risks before they become problems.

The critical qualifier from Archdesk's analysis: the AI is only as good as the data feeding it. Firms with fragmented systems - estimating in one tool, scheduling in another, costs in a spreadsheet - will see limited benefit from agentic AI. Data unification isn't a nice-to-have. It's the prerequisite.

This connects directly to the Building Safety Act compliance picture covered in this brief on 31 March. The Golden Thread requires a verifiable, auditable evidence chain. Agentic AI can help build and maintain that chain - but only if your underlying data is structured and connected. Firms that have invested in data infrastructure are now seeing that investment compound. Firms that haven't are facing two problems at once.

OpenAI closed a funding round of $122bn this week - reaching a post-money valuation of $852bn. The company reports 900 million weekly active users for ChatGPT and $2bn in monthly revenue. Strategic partnerships with Amazon, NVIDIA, and Microsoft remain central to its infrastructure plans.

The number is staggering. The more useful signal for construction firms is what it tells you about where investment is concentrating: large-scale AI infrastructure, enterprise AI applications, and the compute required to run increasingly capable models. The firms building on top of these models - including construction-specific tooling - are working with an increasingly capable foundation.