This is the second of three articles that provide practical considerations for anyone submitting an application to the Futurescot AI Challenge.

The steps below are designed to be undertaken in a 0.5 day workshop or in a series of meetings and based on the Storm ID established AI Discovery workshop.

Related articles in this series: 

Part one: becoming familiar with that AI can and can’t do 

Identifying business problems or opportunities

Solving real-world problems starts with recognising what they are. To find use cases, begin by understanding the business and user needs.

Some useful questions to ask are:

  • What business problems do you want to tackle?
  • What are your users’ main frustrations? Where in your service do they get stuck? What needs do they have that you aim to solve, but are struggling to meet?
  • Do you have routine business processes that could be improved?
  • Do you have complex, multi-step processes which could be made more efficient?
  • Do you have manual paper work processes that act as a bottleneck?
  • What could you do if you had better access and understanding of your data?
  • How could you use AI as a way to address problems that haven’t yet been solved?

Finding answers

Data and evidence tell the users’ stories. Turn to existing research, user feedback and website analytics.

Business needs are in your team. Bring the team together. Get stakeholders in a room and give them the space to talk about process, pain points and aspirations.

Brainstorming ideas

Once you find a list of problems, it can be helpful for the team to work alongside the experts to brainstorm ideas. They should focus on how AI could solve the problems identified. Some simple and quick techniques to help generate ideas include:

SWOT analysis

Conduct a SWOT (strengths, weaknesses, opportunities and threats) focusing on a particular business or service area. This helps to surface broader issues and specific areas for improvement.

Thinking without constraints

LLMs can still benefit a business if plugged into existing workflows. However, to get the most from LLMs, organisations may need to rethink their processes.

We find it can help to surface ideas by thinking, at least initially, without constraint. By this, we mean ignoring current barriers, like budget, organization, and culture. It encourages the organization to explore ideas freely, without worrying about limits.

Human in the loop

Often, you cannot rely solely on AI to fully automate a process. Humans and AI working together is likely to deliver the best outcomes for public sector organisations in the short term.

So, in generating ideas, rethink the role that humans can play in a business process. You may think about ways to limit the role that humans take in certain processes. For example, they may do validation, editing, or overseeing of AI responses, rather than generating them. This opens up new possibilities for what AI could do.

Following these steps should provide you with some use case cases which you can further refine as you assess their feasibility.

Prioritise based on value

Once a list of possible use cases is generated you can then prioritise based on value. Some evaluation criteria to consider are:

  • How does each idea affect citizens? Does it improve their experience or service quality?
  • What potential does the idea have around efficiency gains or resource optimisation?
  • How well does the use case align with the organisation’s strategy and priorities?
  • Does this use case have broader appeal? Could it be applied in other parts of the public sector?

Consider a basic ranking system to give weight to each of these factors. This will help to align your idea with the judging criteria, while focusing on positive impact and alignment to your organisation’s priorities.

I hope the above has provided some clarity of thinking on how to go about exploring use cases for AI in your organisation. If you wish to get in touch please send me an email or check out on the AI Challenge website for further information.