How nPlan uses machine learning to solve construction’s biggest problem in 2019

Dev Amratia
nPlan
Published in
4 min readJan 2, 2019

The construction industry is in dire and urgent need of improvement. The 10 largest construction companies in the UK made an average profit of 0.5% over the last reporting year. After Carillion, the next giant is now on the brink of failure and it is worth stopping to think if we, as citizens who rely upon infrastructure, can afford to let the industry continue down this path. What will it mean for the development of our nations? What will it mean for the utility we have come to expect of our infrastructure?

Among the many causes behind the performance of the industry, there is one that seems to be common to all parties. Sponsors, Clients, Contractors and Sub-Contractors are unable to systematically understand the risk they carry and, put simply, this risk materialises into wasted time. A delayed construction project has ripple effects in lost productivity, trust and reputation alongside the legal repercussions, financial loss and organisational breakdowns. Stories of CEOs stepping in to firefight a project going off the rails are commonplace and the continual burning of such fires would halt the progress of any organisation.

Our story at nPlan is centred on improving decision quality by learning from the past. We want to improve the way knowledge is retained and change the industries’ understanding of risk and uncertainty for the better. Over the last 12 months, we have built a software product that quantifies risk and improves certainty in predicting project outcomes. We close 2018 with proof points that our technology works where we demonstrated feasibility on the largest infrastructure project in Europe. We now have 18 partnerships with multinational construction companies, which enables the algorithms to learn from the largest dataset of construction schedules in the world. Without a doubt, teaching an algorithm to automatically understand the historical context of a project purely based on its schedule is a gargantuan technical challenge, which frankly, many have thought to be impossible (more on this in a later blog post).

Simplistic schematic showing the input and output behind nPlan

Taking our algorithms to the next level will only be done by working with others. Each of the 18 partners we work with have been instrumental to our journey. Our experienced investors have helped and coached us and our research friends at UCL, Harvard and Cambridge Universities support us as we break new ground. We are also very grateful to have been awarded just over £1m in research grants and that Forbes Magazine thinks we are 1 of 15 Machine Learning Companies to Watch in 2019.

The opportunity in front of us is incredibly exciting and is what motivates the team to excel every day. When nPlan delivers insight on project risk, it will do so faster and more accurately than any human has ever been able to. More crucially, the insight we provide removes human biases which plague the construction industry and is able to perform analyses on thousands of projects, simultaneously. These incredibly accurate, rapidly accessed and independent insights allow clients and contractors to understand their risk like never before. This, in turn, will fundamentally change how contracts are negotiated, increasing competition while protecting margins and will work to create a more financially sustainable industry.

2019 will be a big year for the construction industry as these exciting opportunities begin to play out (with sufficient food, sweat and beers from the nPlan team).

The exciting part about this technology is that the more data the algorithms see, the more accurate and confident our insights will become. This is very much about the power of the collective, working together to improve outcomes for the industry while respecting the need to maintain privacy and confidentiality. If you’d like to hear more about how the technology works, the benefits it could bring and how you could get involved, please drop me an email and I’d be happy to share more: dev@nplan.io.

--

--