Turning Fragmented Project Data into Instant Answers
A large national organisation relied on multiple disconnected systems to track project financials, delivery progress, and risk, which made reporting slow and dependent on a handful of specialists.
Skillfield consolidated and modelled the underlying data, then built an AI-powered assistant that allows teams to ask plain-language questions and receive consolidated answers in minutes. Beyond answering the questions asked, the assistant can analyse the underlying data to identify patterns, anomalies, and related insights that users might not have considered, uncovering information that could otherwise have been missed.
The solution replaced hours of manual extraction, reconciliation and analysis with fast, accessible and more insightful reporting.
The Problem
Like many large organisations, our customer’s project information was spread across multiple disconnected systems, with funding, financial demands, delivery tickets, delivery progress and risk in separate tools. Building a complete picture required that data be manually extracted, reconciled, and combined, taking around four hours for a single request. Existing dashboards helped with specific reporting needs but could only answer a limited set of predefined questions.
This created a heavy reliance on a small number of technical specialists who understood how to navigate and reconcile the data – creating a bottleneck. For non-technical stakeholders, getting answers meant submitting a request and waiting for specialist support, resulting in limited access to insights when decisions needed to be made.
The Solution
Skillfield began by establishing the data foundation required for reliable AI. Data from separate systems were extracted, transformed and consolidated into a structured view with consistent fields and metrics.
This data readiness work was critical: without trusted, reconciled information, an AI assistant could not deliver accurate answers.
Skillfield then created a dedicated context layer that defined what each field represented and how the data should be interpreted.
Building on the organisation’s existing generative AI environment, Skillfield developed an assistant that translates plain-language questions into accurate queries.
Teams can now ask questions in a conversational way and receive consolidated answers across multiple data sources without needing to understand the underlying systems or query language. Importantly, the assistant does more than retrieve information. It also analyses the data, interprets results and provides more detailed insights in response to stakeholder questions.
The Outcome
Reporting tasks that previously took around four hours can now be completed in about five to ten minutes.
The assistant is live and used daily, giving teams immediate access to information that was previously compiled only weekly or monthly. Stakeholders can also answer ad hoc questions as they arise, supporting faster, out-of-cycle decision-making.
Rather than receiving data alone and then having to interpret it separately, stakeholders can use the assistant to analyse the information and explore the results further. This means the solution supports both faster access to data and faster generation of meaningful insights from it.
Reliance on specialists has reduced significantly. Teams that previously submitted reporting requests can now find answers directly, removing a key bottleneck and freeing technical specialists to focus on higher-value analysis and more complex work.
