Developing a strong use case for AI Challenge 2025
This guide will help you create a compelling AI use case for the AI Challenge 2025 and design your own AI Agent by applying pre built public sector AI components developed by Storm ID within their agentic workflow engine platform. It breaks down five key points to shape your idea in line with the judging criteria - Scalability, Innovation, Feasibility, and Trustworthy, Ethical AI.
Start with a clear problem
Begin by identifying a genuine challenge or need. For example, you might target reducing backlogs, improving a failing service, removing bottlenecks or improving experience for citizens. Think about the areas of your organisation where there is scope for improvement through automation and what the expected benefits will be.
A strong use case explains how it will make things better - will it save time, cut costs, improve decision-making, or help citizens. Think about how you will measure success (e.g. 30% faster processing of applications)
Understand the existing process
Once you have a clear understanding of your priority business challenges, hone in on a current process or few involved to gain a deeper understanding of existing tasks involved in the process. This could be done by speaking to stakeholders involved and identify quick wins and major pain points.
Match tasks to AI components
Now you understand the current processes involved in the priority areas you can now choose from a suite of pre-built components to design your AI workflow. Each component represents a specific action or task. You simply do this by determining which pre built AI component is most relevant to each task you are aiming to automate in your workflow.
The pre built components to choose from are:
Assess
Examine input to understand its content, context, or quality.
E.g. check a form submission contains all information needed.
Validate
Ensure input meets specific criteria or standards.
Compare
Identify similarities or differences between inputs.
Rewrite
Rephrase content while preserving its meaning.
Classify
Assign categories or labels to content.
Prioritise
Rank tasks or items by importance or urgency.
Redact
Remove or mask sensitive or unnecessary information.
Generate
Create new content based on input or prompts.
Evaluate
Measure the quality or accuracy of input or output.
Summarise
Condense content into key points or main ideas.
Analyse
Break down content to uncover patterns or insights.
Transcribe
Convert speech or video into written text.
Convert
Transform data into a different format or style. E.g. video or image to text
Define the end to end AI workflow
Once you have assigned AI components to each step in the process you want to automate you can then chain these together to define the end to end workflow.
For example, if your use case was automating public consultation analysis then your end to end workflow and components might look like this:
Summarise
Extract key themes or sentiments from feedback.
Classify
Categorise responses (e.g., support, oppose, neutral).
Analyse
Identify trends or patterns across the citizen input to inform policy decisions.
Feasability Assessment
Once you have a good understanding of your new target AI workflows you need to check if how feasible it will be to implement in the context of your organisation and align to AI best practices. Some things you will want to check are:
Data availability and quality are key factors – ask subject matter experts internally if you have access to the relevant data needed and its of sufficient quality.
Ethics and compliance should be considered - are there risks around bias, privacy, or transparency? If your idea involves processing personal data, you need to consider how to protect it. You must also ensure that you comply with General Data Protection Regulation (the GDPR) and the Data Protection Act.