Luke Roberts works at the intersection of restorative practice, systems change, and complexity. He supports schools, justice organisations, charities, and public sector leaders to navigate conflict, harm, and culture change in complex environments. For more on Complex Systems Luke’s book – Leading Schools and Sustaining Innovation: How to Think Big and Differently in Complex Systems is available on Amazon or https://routledge.pub/Leading-Schools-and-Sustaining-Innovation
In the previous blog, I argued that restorative practice struggles in complex systems not because of a lack of goodwill, but because organisational defaults are shaped by management logics that prioritise control, standardisation, and risk reduction. In systems like schools, restorative practice is often absorbed as a veneer rather than allowed to reorganise how harm is understood and responded to.
If that diagnosis is correct, then the question becomes: how does restorative practice actually evolve in complex systems? Where does learning happen when institutions themselves struggle to learn?
The answer, I want to suggest, lies less in programmes and policies, and more in networks — and increasingly, in the tools that help those networks see, understand, and adapt.
From organisations to networks.
Complex systems rarely change through hierarchical structures. Innovation tends to emerge in the informal, relational spaces that run alongside formal structures: between committed practitioners, across boundaries, and through shared problem solving around real cases of harm.
These networks matter because they carry learning faster than institutions can formalise it. They allow experimentation without permission, adaptation without redesign, and sense making where policy offers only categories.
Restorative practice has always depended on this kind of relational infrastructure. What is new is the growing role of AI as a support for networked learning and connection making — not only for young people experiencing harm, but for the practitioners supporting them.
A case of complexity: Georgei
George was a 15 year old student on the edge of exclusion. He experienced frequent emotional outbursts, repeated peer conflict, and escalating disengagement. He showed strong academic potential, but was easily dysregulated by competition, sensory overload, and perceived slights. Traditional behaviour systems responded reactively and left staff feeling ill equipped.
Initial restorative work focused on conflict resolution. Using a structured restorative process, George was supported to articulate emotions, understand impact, and explore alternative responses. This helped with individual incidents, but it did not alter the conditions that repeatedly produced them.
To move from reaction to prevention, the lead conflict resolution practitioner at the school team expanded her teams focus beyond George himself to the relational network around him.
Insightbox AI as a tool for seeing the system.
AI supported personality assessments called Mindsight were used to explore personality and sensory profiles across George and the Conflict Resolution Team. Patterns quickly became visible. George and two members of staff shared high intensity profiles characterised by drive, competitiveness, and emotional energy, while another colleague had a contrasting, calming profile that appeared to reduce escalation.
What previously looked like individual behaviour now appeared as a relational pattern. The system adapted by reconfiguring support rather than intensifying control. The AI did not replace professional judgment; it strengthened it by making relational dynamics visible.
This learning belonged not only to George, but to the practitioners themselves. AI supported reflective practice, team awareness, and more intelligent allocation of relational labour.
Why this matters for restorative practice:
The significance of this case is not the technology itself, but what it enabled. AI increased the team’s capacity to perceive patterns and respond with greater variety. It supported learning rather than stating the obvious, sense making rather than surveillance.
This matters because restorative practice depends on judgment, context, and relationship. Used well, AI can extend these capacities of practitioners of teams rather than undermine them.
From behaviour to capability and future pathways
One unexpected insight from the work with George emerged as the team began to sit with the data more reflectively. The AI supported assessments were not only revealing patterns that helped stabilise relationships in the present; they were also identifying strengths — dispositions, skills, and ways of being that could have real value beyond school.
Traits that had previously been framed almost exclusively as “risk factors” — intensity, competitiveness, leadership instinct, high energy — appeared differently when removed from a purely behavioural lens. In the right environments, these qualities could translate into creativity, entrepreneurship, advocacy, problem solving, or roles requiring drive and presence. What the technology surfaced was not a diagnosis, but a capability profile.
This moment matters. Too often, systems designed to manage behaviour narrow a young person’s future rather than expand it. Restorative practice, at its best, resists this narrowing. It asks not only what went wrong, but what this tells us about who this person is becoming.
Here, restorative practice and AI intersect in an unexpected way. When technology is used to support sense making rather than prediction, it can help practitioners shift from remediation to anticipation — from preventing exclusion to preparing for contribution. The work with George began to inform conversations about subject choices, vocational pathways, and future work environments where his strengths could be assets rather than liabilities.
This is where restorative practice has a unique and under recognised role to play in preparing students for the future. It situates skills, identity, and aspiration in relationship — with self, with others, and with systems. It refuses to reduce a young person to past behaviour, and instead treats conflict and difficulty as information about how environments need to adapt, and how futures might be supported.
Restorative practice is fundamentally concerned with connection, understanding, and the repair of harm. When AI is used ethically — as a tool for sense making rather than control — it can strengthen these aims rather than undermine them. The question is no longer whether technology belongs in restorative systems, but whether we are willing to ensure it serves relationship, learning, and belonging for the networks students need now and in the future.
Author bio
Luke Roberts works at the intersection of restorative practice, systems change, and complexity. He supports schools, justice organisations, charities, and public sector leaders to navigate conflict, harm, and culture change, with a particular interest in how emerging technologies reshape relational practice.
InsightBox is a UK based organisation supporting schools, educators, students, and families to better understand how relationships, sensory needs, and personality dynamics shape behaviour and belonging. Their work combines restorative practice, practitioner judgment, and AI supported insights to help teams see relational patterns that are often invisible in traditional behaviour systems. By focusing on sense making rather than prediction or surveillance, InsightBox aims to strengthen decision making, reduce exclusion, and support more adaptive, relational responses to complexity.
