We worked with USAID/CVP to design innovative analytical services and learning opportunities for the agency’s dynamic global peacebuilding and conflict prevention portfolio. This includes scalable violence and conflict assessments, technical assessments in key issue areas such as atrocity prevention, performance and impact evaluations, learning exchanges, context monitoring, and scenario planning support. We leveraged our data science and GIS capabilities to augment our work with prediction and forecasting, natural language processing and text analysis, including social media and sentiment analysis, data engineering, analysis and visualization, remote sensing and satellite imagery analysis, and remote perception surveys and data collection.

As part of PEARL, Integrity was tasked with helping CVP leverage the real-world potential of AI and large-language models (LLMs) to understand and analyse decades of USAID’s peace and security programming. We analysed over 40,000 long-form PDF reports stored on USAID’s Development Experience Clearinghouse (DEC). These documents often lack abstracts or usable thematic tags, making them difficult to locate. We applied a combination of extract, transform, and load (ETL), machine learning, and generative AI methods to: bulk extract documents from the DEC; identify those related to peace and security; group them into primary themes; and generate usable summaries. We delivered fully annotated programming scripts and detailed code walkthroughs, enabling CVP to maintain the analysis independently. Read more about this piece of work here.

Through PEARL, Integrity supported USAID’s ability to identify the systems that facilitate or diminish armed conflict and violence, provides USAID with actionable recommendations for providing assistance that effectively prevents, manages, or mitigates violent conflict, and contributes to the evidence base for effective peacebuilding and violence prevention development interventions.