I recently talked with Chris Brady, an AI adoption consultant based in Birmingham, UK, who has spent 18 years working in construction. Two years ago, he began integrating AI into his work with contractors and SMEs, initially as an add-on service, and it has since become his main business.

Chris now runs Metrix, an AI consultancy focused on UK construction companies, alongside two other ventures: Trade Upskill, an education platform for construction professionals, and ctrldash.ai, a compliance-automation SaaS for construction SMEs, both of which are soon to launch.

What struck me most in our conversation was how grounded his approach is, built on years of direct industry experience rather than arriving from outside with a technology solution looking for a problem.

Faster means more work

I mentioned a study finding that firms failing to adopt AI risk losing contracts to faster, safer, and more sustainable competitors. I asked Chris whether he is already seeing a real competitive advantage emerge from being faster. He acknowledged that to be the case.

Companies that use AI for tendering, quoting, and pricing can turn work around much faster, enabling teams to handle more opportunities and grow revenue with existing headcount. Chris likened this to Jack Dorsey’s recent comments about using AI to run smaller, faster organizations, arguing that the same dynamic is now emerging in construction.

Faster output doesn’t translate into fewer jobs in construction right now, Chris argues. Demand is so strong that efficiency mainly frees up capacity to take on more work rather than making people redundant.

Construction is a project-based industry where many designers are still paid by the hour. Investing in new tech that reduces the hours needed for the same job looks counterintuitive at first sight. Chris acknowledged this but rephrased it: a designer who works faster shows competence, earns repeat clients. He believes the short-term financial impact of working more efficiently is outweighed by the long-term benefits of client loyalty.

Chris made an interesting analogy to BIM, which he considers to be essentially the same transformation as AI, just at an earlier stage in its development. Both require a shift from manual, siloed work to structured, digital data.

A bottom-up approach to adoption

Chris’s consulting model begins with understanding the business from within. He develops AI-powered surveys, built in Claude Code, that he asks employees at all levels to complete, not just the owner or director.

For one client with 18 staff members, 15 participated in the survey. This provided a detailed snapshot of where time was being wasted and where frustrations were most intense. That data then informs a phased roadmap, addressing quick wins first to build confidence and trust before moving on to more complex automations and custom-built tools.

This contrasts deliberately with top-down adoption, and Chris used the Klarna story as a cautionary tale. The company moved quickly to replace workers with AI, only to have to rehire people when it became clear the tools weren’t ready for the full scope of the work.

Chris’s framing resonated with what I’ve seen in Finland, too. The companies that gain lasting value from AI are the ones that involve people rather than forcing change from the top. Chris clearly states that his aim isn’t to scare people away but to show them immediate value and build from there.

The 10-80-10 model

Chris shared a helpful framework called the 10-80-10 model: spend 10% of your time giving context and guidance to the AI, let it handle 80% of the work, and use the last 10% to review, refine, and ensure the output meets professional standards.

He demonstrated this using a reporting tool he developed for a London-based quantity surveying firm. The firm was spending between four and eight hours creating four types of recurring reports for clients.

Chris reverse-engineered the structure of a finished report and built a tool that automatically pre-populates about 73% of each report. The remaining fields are filled using structured drop-downs instead of free-form input. The time saved per report is significant.

Structured data, kept private

Data quality and structure emerged as recurring themes. Chris uses Obsidian to maintain his personal knowledge base in markdown format, which multiple LLMs can access, so he doesn’t have to re-explain who he is and what he does each time he starts a new task.

Chris recommends a similar approach for clients: start by standardizing the document management system they already use, connect it to an LLM, and set up version-control monitoring.

Data security is the biggest obstacle Chris encounters. It’s less about a company’s own data and more about client and contract information. He manages this by implementing redaction protocols and custom guardrails within project workspaces. He has also developed a free, open-source tool called Winnow that converts any document to Markdown without requiring sign-up.

The outlook for the next few years

Chris was frank about the outlook for AI consultants. In the coming years, the tools can become capable and trusted enough that companies won’t need a guide to use them as they do today.

That foresight partly explains why he’s been building his own products in parallel: a training platform based on real construction scenarios and a compliance SaaS that originated from years of watching smaller firms struggle with document governance.

Chris admitted that his days don’t feel like work at all. His enthusiasm was contagious, and I hope more people in our industry will see AI not just as a threat or hype, but as something truly worth exploring.