Why I left a job I loved

Three years ago, I never would have believed I’d leave my role as a lawyer, because few things engage me the way construction law and the construction industry do. But that’s exactly why I chose to give up my title as a lawyer and join a tech company, combined with the realisation of what properly implemented AI can do to solve the construction industry’s shared challenges.

I’ve worked as a lawyer for more than fifteen years. As your experience grows, your focus shifts from handling individual cases to also spotting structural patterns and systemic failures. In construction, one such structural problem is the combination of too much data and too little information. The relevant information is often there, but it’s hard to find and navigate, even for people in decision-making positions.

There are several reasons for this. Construction contracts consist of many long documents, and understanding them properly and assessing their risks requires both legal and technical expertise. Projects run over many years and generate huge amounts of data: change memos, meeting minutes, formal notices and email correspondence spread across multiple parties, systems and project phases. Key people with hard-won experience retire or change jobs. Assessments made at the tender stage often aren’t passed on to whoever runs production, and lessons from a finished project don’t reach the next one. On top of this, interpreting the contracts requires a good knowledge of relevant case law, which is often hard to access for companies without an in-house legal department (and often for us lawyers too).

What makes this especially frustrating is that the issues keep recurring. The same kinds of shifts in liability, penalty clauses and unclear provisions on change orders come up project after project. Tender documents are unclear, so bidders interpret them differently. The result is bids that aren’t comparable in any real sense, which ultimately hurts the industry (and taxpayers) as a whole. The industry needs structured tools so that companies stop solving the same problems from scratch, again and again.

I’m convinced that agentic AI will be that tool. That is, the kind of artificial intelligence where systems don’t just answer questions (like a regular chatbot) but act independently, make decisions and carry out complex tasks to reach a specific goal. Today, though, the construction industry’s level of digitalisation is low. The digitalisation that has happened has focused on individual tools for technical project management, BIM and bid analysis. These are important applications, but they cover only part of the construction process. Employees are probably also using generative AI tools on their own initiative, including tools IT hasn’t approved, so-called “shadow AI”. But that kind of use lacks company-wide coordination, quality control and access to important data such as case law.

What the industry has largely been missing, and still is, are agentic AI tools built around how the industry actually works across the whole project cycle. For individual companies, the gains are better decision-making in the early stages, more systematic risk management, more efficient handling of project issues, and an institutional memory that doesn’t disappear when key people leave a project. From a societal perspective, the long-term potential is just as clear: better procurement with clearer allocation of risk, less information asymmetry between client and contractor, and more effective contract follow-up that reduces the risk of costly disputes. Put simply, an industry that builds more value for every krona invested.

Earlier this year, Newground Alliance and Brickanta released the report The Cost of Standing Still, which looks in depth at the productivity challenges facing the construction industry and how AI can, among other things, improve and strengthen decision-making and assessments across the sector. Implementing AI will be a big shift for many organisations, but the alternative, standing still, is simply too expensive. Companies and individuals willing to test and evaluate new solutions and ways of working are a prerequisite for progress, even if change can chafe in the short term.