MoD Assurance Assessment Build delivered a data-driven assurance assessment build that automates evaluation of project documents against GovS 002 criteria, producing structured ratings, scores, and commentary at scale.

Local RAG Assurance Engine delivered a fully local, offline-capable assurance analysis engine using retrieval‑augmented generation (RAG) to identify and surface evidence from project documentation and return structured, machine‑readable outputs.

Risky presented a concept prototype focused on applying agentic AI to proactively interrogate risk registers and surface actionable insights for regulators and delivery teams, emphasising clarity of narrative and decision support.

The team developed a proactive behavioural analytics solution focused on surfacing risky resourcing and forecasting behaviours early. Using Python-based analysis and clear visual storytelling, they translate schedule and resource data into behavioural flags and practical recommendations that support early intervention and constructive, blame-fre...

WBS Cost Estimation Tool developed a desktop‑based Work Breakdown Structure (WBS) and Cost Breakdown Structure (CBS) estimation tool that supports structured cost entry, versioned change tracking, and comparison of estimates against actuals across the project lifecycle.

RIO Co‑lab built the RIO Co‑lab, a multi‑agent risk‑register analysis and visualisation solution that applies specialist AI agents to identify themes, assess data quality, summarise change, and surface actionable insights across project and portfolio risk registers.

The team designed a context-aware Lessons SME Agent that builds on an existing Lessons Library to deliver targeted, actionable insights from historic MOD Gateway Reviews. Their work shows how semantic retrieval and persona-driven prompts can surface the most relevant lessons and recommended actions for different roles and project phases.

The team developed a scalable Lessons Library pipeline that ingests historic MOD Gateway Review documents and converts them into a large, structured lessons dataset. Their solution focuses on high‑volume extraction, semantic classification, and sentiment analysis to rapidly surface reusable lessons for assurance and organisational learning.

Forecast Input Cost App delivered a Power Apps and Power BI based cost‑forecasting solution that enables controlled forecast entry, integrates actual spend data, and provides clear visibility of cost performance against estimates across projects.

PEAT Document Assessment System developed an interactive assurance evidence assessment solution that applies large language models to analyse project documentation, score maturity, and surface assurance evidence and gaps aligned to recognised governance frameworks.

The team produced a data‑driven risk heuristics analysis pipeline that combines Python analytics with large language model feedback to assess and enrich existing risk registers. Using Jupyter notebooks, they analyse risk and mitigation data, apply SME heuristics via an LLM, and output annotated spreadsheets and summary datasets designed for do...

The team proposed a comprehensive behavioural analytics dashboard to expose hidden patterns in project scheduling and resourcing that undermine delivery confidence. Using defined schedule and resource integrity metrics, the solution highlights chronic under‑ and over‑utilisation, forecast inaccuracy, ignored dependencies, and reliance on gen...

The team developed an AI-enabled risk management solution that integrates SME heuristics with automated evaluation to improve the quality and actionability of project risk registers. The approach combines Microsoft Power Platform components with LLM-driven analysis to identify weak risks and mitigations, prioritise critical issues, and support c...

The team explored persona‑driven behavioural analytics to address risky resource planning practices. By combining detailed persona definitions, behavioural metrics, and deep analysis of forecasting and utilisation data, they designed a dashboard concept that highlights over‑optimistic planning, generic resource use, and weak feedback loops,...

The team built Jim‑E, an interactive AI‑assisted risk review tool that applies SME heuristics to project risk entries. Using a lightweight Streamlit interface and encoded heuristic rules, the solution helps users identify weak risks and mitigations, capture structured feedback, and generate clear audit‑ready reports.

Risk Review Regulator designed a set of agentic AI personas to automatically review risk registers, analyse data quality and exposure, and drive action through targeted, human‑centred email nudges. The solution demonstrates a closed‑loop risk‑management cycle spanning project, portfolio, and enterprise views.

Team 1B applied structured prompt engineering with Microsoft Copilot to automate assurance evidence identification and scoring across multiple personas, aligned to PEAT success criteria.

The team focused on establishing strong data quality and analytical foundations for a Project Health and Behaviour Monitor. Using a structured synthetic dataset, they demonstrated how task-level schedule, cost, and resource attributes can be cleaned, validated, and analysed to identify volatility, critical path risk, forecasting accuracy issues,...

Project Overrun Predictor built a machine‑learning driven schedule‑forecasting prototype that predicts the likelihood of project overruns by analysing feature trends across completed and in‑progress energy projects, supported by an interactive Streamlit application.

The team built a rule‑driven risk assessment system that converts SME survey responses into structured, validated heuristics. Using LLMs, fuzzy matching, and human‑in‑the‑loop review, they generate, deduplicate, and govern high‑quality risk and mitigation rules that can be applied consistently across risk registers.