By Kim Morrison, Chief Technical Officer, ATC Williams
Artificial Intelligence (AI) is moving fast. Barely a day passes without a new model, capability, or competitor reshaping what organisations can do with technology. The challenge many of us face is not simply one of adoption but also of direction. How do we move quickly enough to stay relevant while maintaining the guardrails that protect our people, our clients, and our reputation?
Over the past twelve months, our organisation has been working through that challenge in a deliberate, structured way. This article reflects on the steps we have taken and looks at what is ahead. The single thread running throughout is that responsible governance and genuine innovation are not opposing forces. Done well, governance supports innovation with greater confidence.
The AI Working Group: Structure, Purpose, and Progress
In July 2025, we formally established an AI Working Group, initially chaired by me as CTO, with the reins recently taken over by Charles Li, Senior Engineer, closely supported by our IT Manager, Sam Wijayasinha, to support acceleration of our internal AI initiatives. This group brings together representatives from across the organisation to provide oversight, prioritisation, and strategic direction for our AI programme.
The AI Working Group was deliberately designed to avoid two failure modes we often observe elsewhere: the purely theoretical committee that produces frameworks no one uses, and the purely technical committee that optimises for capability without considering broader impact. Charles is actively working to keep the group grounded in practical outcomes.
About the AI Working Group
Working closely with IT and the CTO, the AI Working Group supports:
- Setting strategic direction for AI investment and experimentation across the organisation
- Reviewing and approving use cases before they move into production, applying a consistent risk and benefit assessment
- Monitoring developments in the external landscape (e.g., regulatory, competitive, and technological) and advising on their implications
- Coordinating knowledge sharing so that lessons learned in one part of the business benefit the whole
- Providing escalation pathways for teams encountering novel or ambiguous AI-related decisions
Since its formation, the AI Working Group has reviewed a growing pipeline of proposed AI applications, established shared evaluation criteria, and begun building the community of practice that will sustain this work over the long term. These are early days, but the infrastructure is taking shape.
Setting the Foundations: Our AI Policy
Alongside the AI Working Group, we developed our inaugural AI Policy in 2025. Policy documents have a reputation for being shelfware, always produced with good intentions and often ignored in practice. We have aimed to avoid that fate by designing the policy around the decisions people actually face, rather than abstract principles.
The policy addresses three core areas.
1. Acceptable Use
The policy draws a clear distinction between AI applications that are approved for use, those that require CTO and IT sign-off, and those that are out of scope. The goal is not to create a culture of permission-seeking for every small experiment, but to ensure that high-stakes applications that involve sensitive data, inform consequential decisions, or operate at scale receive appropriate scrutiny before deployment.
2. Data and Privacy
Any use of AI that involves personal data, confidential client information, or commercially sensitive material must meet defined requirements. This includes restrictions on which tools and platforms are permitted, requirements for data minimisation, and obligations around transparency when AI is being used in client-facing contexts.
3. Accountability and Transparency
We are establishing clear lines of accountability for AI-powered systems to determine who owns the outcome when a tool produces an error, and how we detect and correct that error. The policy also addresses transparency. Both internal and external stakeholders who have a legitimate interest in decisions should know when AI is playing a significant role in a decision or output.
The policy is a living document that continues to evolve as our experience deepens and as the regulatory environment becomes clearer.
Our AI Think Tank
Earlier this year, we convened a cross-functional Think Tank to ask and consider a blunt question: ‘Where does AI create real value for our organisation, and what stands in the way of capturing it?’
The session brought together leaders from across our organisation, including client-facing teams, operations, finance, risk, people and culture, among others. The participation was deliberately broad, because AI does not respect functional silos.
The conversation surfaced recurring themes. First, the gap between AI enthusiasm and AI readiness: many parts of the organisation were eager to experiment but lacked the shared vocabulary, tooling, or governance scaffolding to do so responsibly. Second, the risk of fragmentation: without coordination, individual teams would build in isolation, duplicating effort and creating a patchwork of incompatible practices. Third, the opportunity cost of inaction: the broader market was accelerating, and standing still was not a neutral position.
A clear mandate was established to accelerate internal AI adoption beginning with personal productivity and then create a knowledge base built on robust AI policy. Targeted training and measurable goals are tied to business outcomes rather than technology milestones.
A Final Reflection
AI is not a technology project, but instead an organisational transformation that touches how decisions are made, how clients are served, how risk is managed, and ultimately how we think about the nature of work. Our AI Think Tank, AI Working Group, and AI Policy are the scaffolding that allow the real work to happen by embedding AI in ways that are genuinely better, not just faster.
I acknowledge Charles Li, Sam Wijayasinha, Laura Tipping, and the rest of the AI Working Group for the thoughtful, rigorous way they are approaching this mandate. Creating robust governance is important work. I am optimistic that the foundation we are laying will make the decisions ahead easier to navigate.