AI Hub partners convene for use case scoping workshop

On March 24–25, 2026, the AI Hub partners convened in Berlin for a hands-on Artificial Intelligence Use Case Scoping workshop delivered by Merantix.  

The workshop focused on applying a structured methodology to highlight the foundational considerations needed before adopting AI solutions in a country. Throughout the 2 days, participants applied the methodology to a real-world case: strengthening social assistance delivery (cash transfer) in Morocco. The practical example was guided by insights from the Moroccan National Agency for Social Support (ANSS) team, who joined remotely to share their system, targets, pain points, and how they perceive AI could be potentially used to address these challenges. 

 By the end, participants had: 

1. Mapped the Moroccan use case across all dimensions of the Merantix AI Canvas Methodology
2. Identified key assumptions, evidence, and knowledge gaps  
3. Built a foundation for future evaluation and decision-making on AI adoption  

One key message stood out at the end of the workshop: impact depends on measurable solutions that are well evaluated, effectively targeted, and accountable.

Case study: social assistance in Morocco

The Moroccan case was used as a reference throughout the workshop. In Morocco, social assistance is delivered through monthly cash transfers to beneficiary bank accounts and managed by the ANSS. The system already demonstrates a high level of digital maturity, including a fully digital application and payment process, interoperability with more than 11 institutions, and monthly application reviews. 

However, key challenges remain particularly around data accuracy. These include inaccurate or manipulated application data and undeclared informal income, identified as a major behavioral issue. As a result, targeting can be affected, with some eligible households excluded due to incomplete or incorrect information. 

The AI Canvas: Turning assumptions into evidence 

The workshop was guided by the Merantix AI Canvas, a structured tool designed to guide the application of AI systems through a holistic evaluation of challenges and opportunities. It can identify issues that should be addressed before choosing an AI solution.  

Participants considered the following questions related to the Moroccan context:  

• Assumptions – what claims are we taking for granted?  
• Evidence – what data, systems, and processes already exist?  What is missing? 
• Unknowns – what still needs to be validated to address challenges?  

The AI Canvas process spans five key dimensions: 

1. Value proposition – what problem are we solving, and what value will be created?
2. Performance indicators – how will success be measured?
3. Organization – who owns the solution, and how will it be embedded?
4. Data and technology – what data is available, and what systems are required?
5. Operational implications – what governance, risks, and processes must be addressed?  

Understanding and managing risks 

The final part of the workshop focused on risk identification and mitigation, drawing on frameworks such as the AI Risk Repository, developed by the Massachusetts Institute of Technology. Participants mapped potential context-specific risks in Morocco across several categories including:  

• Bias and discrimination arising from incomplete or unrepresentative data  
• Privacy and security concerns  
• The possibility of system failures

For each identified risk, participants assessed severity, likelihood, and controllability as prerequisites for outlining practical mitigation strategies.  

This structured and integrated approach highlighted a critical insight: risk management cannot be a separate phase or afterthought but should be a core component of designing responsible AI systems from the very beginning. Understanding and managing risks enables effective implementation and reduces potential harm. 

Key reflections and looking ahead

The AI use-case scoping workshop emphasized the importance of considering local and national contexts and capacities, as well as evaluating the current situation through the AI Canvas, before deciding which AI solutions are best suited to address existing pain points within social protection systems. By employing the AI Canvas coupled with the AI Risk Repository, the workshop revealed critical insights:  

• Clarity takes time – investing in problem definition strengthens outcomes  
• Context is everything – local institutional, legal, and social realities shape feasibility  
• AI is socio-technical – success depends as much on governance and processes as on models  
• Participation matters – direct engagement with the Moroccan ANSS team grounded discussions in reality  

The next step is turning these insights into action by testing, validating, and scaling solutions in real-world contexts. The AI Hub will continue to work to advance use cases from concept to implementation, ensuring that AI systems are not only innovative, but also responsible and impactful. 

We encourage you to reach out to the AI Hub at contact@spdci.org if you are interested in collaborating, applying this approach or joining our country support portfolio.