Activity performance prediction
We use the largest database of as-planned and as-built construction projects in the world (>750,000 programme files) to train deep learning models to predict the performance of all activities in any construction programme. Our deep learning models use a variety of features that ensure each activity is uniquely described, and go through a rigorous testing process to ensure the highest accuracy.
Graph Neural Networks
Harnessing the power of GNNs, we analyze relational data among tasks, resources, and timelines to extract actionable insights. This relational understanding enables us to optimize resource allocation and uncover potential bottlenecks before they impact project timelines.
Advanced simulation engine
We saw the industry standard for Monte Carlo simulation and said "it's not good enough". We developed an in-house advanced simulation engine that works on programmes with tens of thousands of activities and can produce forecasts in minutes. This way, we can ensure that the forecast information you get is reflective of your actual programme, and not a summary.
Interested in ML? Join our Paper Club!
Founded in 2018, it’s open to everyone with an interest in groundbreaking research into ML - with regular guest presenters and an audience drawn from companies and universities across the UK and beyond.
Large language models (LLMs) are the state-of-the art when the task at hand is content generation. We harness their power to create the most advanced generative applications for risk professionals.
We have all seen ChatGPT, and we all think it's great. However, accessing the breadth of internet's knowledge requires an involved process and iterative querying. We are developing our own LLMs (ICE-LM) that will be fine-tuned for the risk management case. Whether it's sifting through project reports, contracts, or correspondence, our LLMs ensure that crucial information is never overlooked, aiding in informed decision-making and proactive management across all stages of your projects.
One of the most challenging tasks in the early phases of a project is creating a schedule from scratch. At nPlan we are developing generative AI models to create schedules from scratch. These will enable exploring a vast array of project strategies, ensuring a well-informed and resilient project roadmap.
Handling multiple data-sources
Project data is messy. A risk professional needs to be able to access risk registers, schedules, cost estimates, H&S reports, early warning systems, and more. Somehow, the risk professional needs to be able to understand and derive insight from all these. We are developing LLM-powered applications that will make this process simple, with a clean interface that will empower you to see what's happening in all aspects of your project in real time.
Reporting... made easy
Regular reporting is one of the most dreaded activities in the risk industry. Countless risk mangers spend 70%+ of their time reporting on risk, instead of managing risk. We are developing LLM-powered reporting applications that will make report generation a 1-hour job at most, regardless of the report and the data required to create that report.
Risk Analysis Algorithms
Because not all AI is done with Machine Learning, we also do research in automated risk analysis methods to provide the most relevant information for a risk professional to manage.
Forecast-driven schedule summaries
When working with schedules of tens of thousands of activities it is inevitable that the schedule will need to be summarised. We have developed schedule summarisation methods that work hand-in-hand with our forecasts. So even if the schedule is summarised, you won't lose any important information.
Optimal risk management strategies
Finding the right risk management strategy is one of the most complex tasks in mega-construction projects. With 10s of thousands of activities contributing to the potential delay of a project, which ones should you focus on first? nPlan has developed several state-of-the art ranking methods that show you what are the top activities to manage in the schedule.