Public Sector AI Canvas

We are living in an era of rapid technological advancement. Every day, new artificial intelligence tools emerge that are more accessible and powerful than ever before. Consequently, it is natural for governments and public organizations to feel significant pressure to adopt these solutions.

However, caution is warranted: technology alone cannot solve structural problems. Before selecting the latest tool, establishing a clear strategy is essential. Without one, the risks of wasting time, resources, and public trust are high.

Research from MIT, for instance, has found that 95% of AI projects in the private sector fail to deliver meaningful results. The causes range from a lack of strategic clarity to unreliable data and poor governance. The parallel to the public sector is clear: without proper planning, AI risks becoming just another technological fad rather than a true lever for transformation and public value creation.

It is within this context that I developed the Public Sector AI Canvas — a visual framework that organizes the key elements for structuring AI initiatives into distinct sections, focusing on results, governance, and sustainability.

Below is a brief guide to effectively using the Canvas, section by section:

1. Public Problem: Clearly**** define the core problem that needs to be solved. Without an accurate diagnosis, AI becomes a solution in search of a problem.

2. Current Process (without AI): Map how the problem is currently addressed (e.g., manual processes, existing systems). This helps assess the potential gains from adopting an AI solution.

3. AI Solution: Describe the proposed AI-based intervention, such as process automation, prediction/forecasting, content generation, or data analysis. Be specific.

4. Target Users and Stakeholders: Identify who will be impacted — public servants, managers, or citizens. This perspective is the foundation for creating public value.

5. Tools and Technologies: List the proposed tools and technologies (e.g., specific LLMs like ChatGPT or Gemini, APIs, machine learning libraries, no-code platforms). Choose technology based on available resources and strategic goals.

6. Data and Sources: Verify the availability of structured, accessible, and reliable data. Without a solid data foundation, no AI project can succeed.

7. KPIs: Define how results will be measured: time savings, cost reduction, user satisfaction, model accuracy, among others.

8. Risks and Ethical Considerations: Examine potential issues such as bias, privacy, transparency, and digital exclusion. Governments cannot compromise on ethics and public trust.

9. Human–AI Collaboration: Define which tasks remain with humans and which are delegated to AI. Process supervision (human-in-the-loop) is key to mitigating errors and minimizing risk.

10. Governance and Sustainability: Plan how to ensure continuity, including policies, budget, staffing, infrastructure, and ongoing maintenance.

The Public Sector AI Canvas helps leaders structure AI projects with clarity, balancing innovation with responsibility. It transforms the pressure to adopt AI into applied strategic planning — a fundamental condition for technology to truly deliver public value.

I have released the Public Sector AI Canvas under a Creative Commons license (CC BY-NC-ND 4.0). It is available for download and use by public servants, managers, researchers, and anyone interested in advancing digital transformation in government.