Built AI tools for my surveying practice, now helping others do the same

The short version of why GroundMark exists is that I couldn't find anything built for someone like me. I'm a quantity surveyor, which in construction means the professional who manages the cost and the contract on a building project. I run a small practice on my own.

I started looking seriously at AI in July 2025. I was aware of it, but had no real idea how it applied to my line of work. I took Mindvalley's Amplify with AI Quest, which was genuinely eye-opening on what AI can do, from generating images and music to turning a photo of your fridge into dinner ideas, and it taught some useful frameworks. But there was nothing in it about applying any of this to a surveying practice. I then did the RICS course on AI, and came away no clearer. It was only after starting AI Mastery in April 2026 that it clicked. Not because AI Mastery is about quantity surveying, because it isn't, but because it gave me enough range and confidence to build the tools for my own job myself.

There are firms building specialist software for specific surveying tasks. I'm not trying to match or copy those, and I couldn't. My tools focus on the everyday repetitive work that can be made quicker and safer. The idea behind GroundMark came from a simple realisation: if I, as a small sole-trader owner, couldn't find help that fit how the work actually runs, then nobody else in my position could either. That's a niche, but in the UK it's a substantial one. The tagline is 'AI built in practice', and the credibility behind it is that I've built AI into two real fee-earning practices of my own, a surveying practice and a Passivhaus consultancy, including the dead ends. I've done it, not just studied it.

Achievements (progress to date)

•    Brand built and documented: a GroundMark brand book (version 1.0) and a full logo kit.

•    Offer defined: a three-tier model, from short workshops through a full implementation programme to a toolkit and retainer, with the implementation programme as the anchor.

•    A workflow programme mapped across the practice, built in the right order with the foundations first. The brand and setup work is complete, and the document-processing area is proven (a specialist tender comparison down from most of a day to under an hour).

•    Supporting assets built: a CPD deck, a one-pager, a workflow audit and workflow map, an ideal-client profile, a business plan, and a written case study.

•    Positioned deliberately as a brand separate from my surveying practice, to avoid a genuine conflict of interest. It advises at the level of tools and systems, and is designed so it never needs access to a client's live project data.

Tools: Claude (projects, skills, memory, and connectors to my real stack), Claude Design for the site and prototypes, Airtable, Microsoft 365, and image-generation tools for the brand assets.

Methodology: The principle GroundMark is built on is that the foundations come first. Most attempts to bring AI into a practice fail because people chase the shiny output before doing the unglamorous setup: the context, the brand, the structure. So every engagement starts by finding where the practice actually loses time, then builds and embeds tools they keep and run themselves. It's enablement, not a done-for-you service. I'm helping a firm build the capability in-house, not adding another thing they outsource and depend on.

Learnings: Credibility in this market comes from having done it, not from theory, and that includes being honest about what didn't work. I document the dead ends, such as a tool to populate energy-modelling spreadsheets that hit the limits of how those spreadsheets are built. Staying tool-agnostic is a deliberate trust signal, not a limitation.

Direction of travel (deliberately not yet complete): 2026 is a proof-of-concept year, not a revenue chase. The goal is to have the full workflow programme documented with real, conservatively measured time-savings, and at least one implementation engagement completed or underway, by December 2026. Several workflow areas are mapped and in build, with real examples replacing the placeholders as I run them. The endgame is a repeatable, evidence-led offer that a small practice recognises immediately as built for their world, rather than another generic AI course.