This first “recipe” introduces the concept of democracy-in-the-loop—a counter-model to automation and optimisation that prioritises collective sense-making, friction, and meaningful human agency.

It is the first in a series of five practical “recipes” developed to help practitioners design democratic processes that integrate AI tools without compromising democratic values. These short guides are drawn from the Digital Democracy Lab Handbook and tested in real-world use cases.

Where much of civic tech leans into speed, scale, and seamlessness, this model proposes something different. Democracy-in-the-loop suggests that slowing down, making design decisions visible, and building in moments of reflection and contestation can create more legitimate, trustworthy and inclusive outcomes—especially when digital systems are involved.

Instead of seeing AI as a replacement for human judgement or discussion, this recipe invites you to think of it as a tool within a democratic process—not the driver of it. It asks: how can you design a process where human deliberation remains central, even when digital tools are involved?

This recipe is part of the Digital Democracy Lab Handbook — a practical resource for facilitators, technologists, and public sector actors exploring how AI can support meaningful democratic exchange. You can find further practical guides and recipes in the handbook.