Depolarization Toolkit: An Agentic AI Framework

In collaboration with UNDP and DW Akademie, the project aims to develop an AI assistant with a multi-agent architecture for depolarization-related programming, capacity building, and training, targeting development practitioners, media professionals, and public communicators.


Project Team:
Semuhi Sinanoglu
Armin von Schiller

Financing:
IDOS

Time frame:
2026 - 2026 / ongoing

Project description

Designed for practitioners, the toolkit provides cutting-edge knowledge on polarization to those working to counter it. An AI-powered assistant with a modular, multi-agent architecture will be developed and user-tested as a conversational interface for generating adaptive, context-sensitive content. The assistant, grounded in a vetted knowledge repository, will serve as a dynamic toolkit, publicly available to organizations and individuals, to help co-design their own depolarization programming, capacity-building strategies, consulting processes, and training. Scalable thanks to its multi-agentic structure, the AI model can reason about the user’s context, produce structured outputs such as action plans and risk assessments for programming, and dynamically generate simulations and branching scenarios for capacity building and training.

Publications