You’re a leading SME contractor and want to improve your construction management systems, but you can’t afford enterprise-level software from the big software developers. So what do you do? If you’re Steve McKenna MCIOB, director of chartered contractor Gemstone, you use AI to build your own construction management software.
With no coding background, McKenna used Claude Code to develop Construction AI – a production-grade, multi-tenant SaaS application comprising more than 700,000 lines of code, 186 database tables with row-level security, 596 API routes, and 60-plus AI-powered tools spanning 22 modules (from drawings management and tender analysis to cost control, contract administration and programme tracking).
He launched Construction AI into the marketplace just under two months ago. He has already secured nearly 20 customers and will be exhibiting at Digital Construction Week in June.
McKenna started in construction at the age of 16, training at the CITB’s Bircham Newton base some 30 years ago. He trained as a civil engineer with Balfour Beatty, before becoming a QS. His career highlights include working on the White City Westfield development, being involved in the construction of a Tesla factory in Germany and working for ISG. In 2019, he set up Gemstone.
CM Digital caught up with him to find out more.
CM Digital: What drove you to develop Construction AI?
Steve McKenna: There are some great construction management platforms out there, like Procore or Aconex, but they’re serving the bigger contractors. Where I was two to three years ago, I couldn’t afford those enterprise solutions. Thus, Construction AI makes enterprise-level software available to SME contractors at an affordable price.
“As AI improved, the process became easier, and I basically followed the foundations of how you build a building: get the core right and then, as you build it out, you follow the same principles all the way through.”
[As AI started to explode] Gemstone was growing, and we started to implement management systems to be structured and organised with our documents and communications with clients.
It was late 2024, and I began implementing some simple AI tools, such as a spelling and grammar checker, a toolbox talks generator, and tools that could help the team prepare RAMs – that sort of thing. We were having some success with it and it just kept on growing and growing. In early 2025, I incorporated Construction AI and continued building AI tools. AI was improving and it snowballed. Then we launched the business earlier this year.
The thought process was: “What holds us up in our business? What can we improve? What would free up more time for us?”
The main areas of focus were creating project programmes, creating RAMs and tender processes: from getting an inquiry into the business, how do we manage that process? How do we get it out to supply chains to price, and then, once we get the pricing back, how do we standardise it and just make the whole process efficient? Those were the key pain points that we addressed first.
How technical are you? Have you coded in the past?
I’ve always been keen to adopt technology in the business. I’ve seen the value in it. I’m kind of techie, but not techie to the extent that I can design and build software. As AI improved, the process became easier, and I basically followed the foundations of how you build a building: get the core right and then, as you build it out, you follow the same principles all the way through.
Once you’ve got the set-up right and it’s working, it’s just a case of adding little building blocks and building it out, and maintaining quality control throughout the process as well.
Early on, I tried to find a co-founder who was more techie than me, but that didn’t work out.
Before I touched any code, I spent a lot of time designing the architecture first, probably a good two or three months. After I started to get some success with the AI tools I was creating, I sketched out the plan and how I wanted it to all look – the framework – then I designed and wrote specs for each section, and iterated with the AI. Once I was settled on a spec, it went into production in the AI. That was followed by recursive testing.
You didn’t just build the platform with AI, did you? AI is embedded throughout
Construction AI is built with AI throughout the product, rather than an AI process strapped on. For example, if you upload a specification into the application, it chunks the document into small sections, no more than 2,000 words, stores them in a database, and then that’s semantically retrievable by the AI agent.
“We’ve built a full knowledge base, which includes the likes of UK contract structures, CDM regulations, building element classifications and building regulations, that the AI looks into.”
We’ve built a full knowledge base [which includes the likes of UK contract structures, CDM regulations, building element classifications and building regulations] that the AI also looks into. So if it’s building a [project] programme, for example, we’ve got a full suite of documentation for the AI to reference. When it gets a prompt from the user, it will check that knowledge base and then structure the program accordingly, using best practice.
What’s the pricing structure?
It’s £100 a month per seat. So, for a smaller contractor with one or two employees, it’s quite affordable compared to an enterprise solution.
We’ve already got 20 customers on board, some with three or four seats. The launch has gone well, so we’ve recruited a product manager, who’s reaching out to our customers and supporting them through the onboarding process, which isn’t too tricky. The platform is quite straightforward and intuitive to use.
What’s next?
We’re carrying out some development work to make the tendering process a bit more fluent for estimating.
We’re working with the customers to fine-tune the platform, within reason, to suit their needs. Because we’re agile, we can be flexible and work with them to make it work for them.
I know how hard it is [as an SME contractor] – you don’t get much time and you’re always chasing your tail. We want to help SMEs manage their processes better.
