From ‘AI Working Group’ to an AI working group that works

When used well, AI reduces manual effort, improves service delivery, enhances decision‑making, and frees people to focus on value-adding work. It creates new opportunities for innovation and crosses teams, disciplines, and job roles.

With that opportunity comes a responsibility to guide adoption safely. As AI use accelerates, so does the need for strong governance and clear guardrails. Without it, things can go wrong.

A recent example comes from West Midlands Police (WMP), where AI-generated evidence mistakenly referenced a non-existent football match between Maccabi Tel Aviv and West Ham, a detail introduced as a result of using AI tool Microsoft Copilot. The Acting Chief Constable has now taken the decision to block access to Microsoft Co Pilot on WMP systems until further notice.

This is exactly why working groups are so important. But the value doesn’t simply appear because a group meets to talk about AI, and that’s where many of them fail. You meet, discuss exciting new technology, debate definitions, explore risks, review frameworks, imagine future possibilities… and then nothing changes. Months pass with no decisions, no direction, and no meaningful implementation. Sound familiar?

A successful working group cannot be only a consultation forum. It must be a mechanism for action that turns AI from an interesting topic into something practical, responsible, and genuinely useful.

So, what does an AI working group that truly works look like?

1. A clear purpose, grounded in organisational priorities

Start by defining why the group exists and what it is responsible for delivering. A strong Terms of Reference might include:

• exploring and assessing AI use cases and opportunities
• conducting AI risk assessments
• preparing for emerging regulation and legislation
• developing and implementing AI policies
• defining and owning the organisation’s AI strategy
• maintaining an AI application catalogue
• sponsoring responsible and ethical AI adoption and deployment.

Crucially, this work must be anchored to the organisation’s broader vision and priorities. An effective group isn’t just about exploring technology, it’s about understanding how AI can enable the organisation’s mission, values, and strategic goals. Whether the vision is improving student experience, strengthening operational resilience, enhancing public service outcomes, or driving innovation, the group must operate with that direction in mind. Clear alignment ensures that AI activity is purposeful, coordinated, and able to demonstrate real value rather than becoming a standalone initiative.

2. Action oriented meetings instead of endless conversations

A working group earns credibility by delivering outcomes, not by meeting frequently. Track progress openly and be clear about what has been piloted, what has been adopted, what has changed, and what the organisation can now do as a result.

Crucially, the group needs enough decision-making authority to keep momentum. In many organisations, formal approval often sits with committees or boards, but AI evolves too quickly for a group that can only recommend and then wait months for signoff. A high performing group should have delegated permission to make lower risk decisions, greenlight pilots, and move quickly within defined guardrails.  

3. Prioritising your AI deployments

Many organisations begin with the question “Where can we use AI?”. A more valuable question is “What’s the challenge?” Consider what is slow, expensive, or frustrating today and where you are performing repetitive manual work. As you explore opportunities, consider the quality of your data – feasibility depends heavily on having access to clean, well‑governed, reliable data. A structured decision matrix that considers feasibility, risk, value and alignment to organisational objectives ensures AI is deployed where it genuinely matters. Use AI on what slows you down, not on what makes you special.

4. Assign clear ownership

If everyone owns AI, no one owns AI. Every initiative should have a single accountable owner, a clearly defined scope, an implementation roadmap, and a measurable outcome.

Just as important is selecting the right people. Combine technical expertise, operational insight, legal and ethical oversight, and the authority of senior leadership. AI working groups fail when they are composed only of senior leaders or only of technologists. AI delivers value when these perspectives work together. Give the group a clear chair, clear decision routes, and visible sponsorship at executive or board level.

The real measure of success

A successful AI working group is not measured by the number of meetings, the length of strategy documents, or number of AI tools bought or deployed.

It’s measured by:

• Challenges solved
• Time saved
• Decisions improved
• Responsible and effective adoption.

When the group becomes a driver of practical, responsible, meaningful change, that is when AI starts creating real value.

How can we help:

We’ve been helping organisations across multiple sectors to design, deploy, and deliver AI initiatives that actually work in practice. We understand that most organisations sit somewhere between two concerns: not wanting to miss out on the opportunities AI brings, and not wanting to mess up by getting it wrong.

Our AI specialists can join your AI working group to provide grounded, practical insight shaped by real delivery experience. We help groups align around a shared organisational vision, translate ambition into actionable work, and make sure conversations lead to decisions, direction, and measurable progress. Whether you’re just getting started or refining an established programme, we can support you to set the right focus, avoid common pitfalls, and build the confidence and capability needed to use AI responsibly and effectively.

Want to learn more? Get in touch at ai@waterstons.com