Genuinely rigorous research, grounded in real industry experience.
Our story
We started Onsite Intelligence after years working with tier-1 construction contractors in the UK and overseas. Watching capable AI tools get blocked at the door because the legal and compliance risks of cloud platforms were simply too high. We also saw the other half of the problem: enterprise document control systems like Thinkproject and InEight, built for tier-1 contractors with dedicated document-control departments, and priced and shaped to match. Brilliant tools, wrong fit for everyone else.
Small and mid-sized UK businesses face the same bad choice over and over: ignore AI completely, or hand their most sensitive information to a server farm on the other side of the world.
Neither option was good enough. So we built the alternative. Proper, private AI systems that sit in your office, run on hardware you own, and answer to you.
Backed by research, not just enthusiasm.
The thinking behind Onsite Intelligence didn't start with AI. It started with an MSc dissertation at the University of Aberdeen examining how standardised data management could improve efficiency in project delivery across multi-disciplinary teams. Research conducted with 22 industry professionals, analysed in SPSS and NVivo, and grounded in peer-reviewed literature going back two decades.
The findings were clear. The biggest barriers to efficiency in project-based businesses weren't technical. They were organisational. Lack of training. Fragmented "islands of automation". Manual re-entry of the same data into multiple systems Research has long shown this accounts for around 25% of total business transaction cost, with as much as 70% of business data re-entered between systems. And it was smaller firms, without dedicated document-control departments bearing the worst of it. They were stuck in what the data-maturity literature calls the "Emerging" stage, while larger organisations had moved on to "Defined" and "Managed".
Onsite Intelligence is the productisation of those findings. Every design decision. The browser-only interface, the structured filing, the no-new-software-for-staff principle, the audit-logged answer trail, and the plain-English retrieval that pulls answers straight from existing documents instead of forcing staff to recreate them all traces back to a research-evidenced barrier the system was deliberately built to remove.
Pulling an answer from a document that already exists is the practical antidote to the 70% re-entry problem the research identified. The result isn't a startup MVP rushed to market. It's a structured solution to a problem we spent years documenting before we ever wrote a line of code.
King's College, University of Aberdeen
Years of research, before a line of code
The full dissertation is available to read in its entirety
We're not in the habit of hiding the work that went into our thinking, so if you want to verify the research, dig into the methods, or just satisfy a healthy professional curiosity, you're welcome to read it. No form, no email gate, no follow-up.
University of Aberdeen · School of Engineering · MSc Project Management · September 2025