The Retail Trust was seeking new ways to innovate and support retailers to nurture a more resilient and sustainable retail sector. CGI had previously supported the charity with the creation of a ‘first of its kind’ Happiness Dashboard, powered by Generative AI, which provided retailers with actionable insights on their employee wellbeing. This included recommendations on wellbeing content and initiatives for employees.

The Retail Trust wanted to evolve the solution to streamline the process of sharing campaign content with their customers and reduce the need for time-consuming manual intervention. The goal was to automate 80% of the campaign process including both campaign creation and the extensive feedback gathered from their retailers. They also sought to create more user engagement.

Automating the campaign process with Agentic AI

Agentic AI is a network of multiple AI agents that interact or collaborate to solve complex tasks. CGI leveraged the capabilities of Agentic AI to develop a dynamic Campaign-in-a-Box solution that gave the Retail Trust a ready-to-go pipeline of high-quality content. All content adhered to strict editorial guidelines and was tailored to each of their retail customers and end user employees. The product was based on a series of AI agents, built into the Happiness Dashboard and using existing Azure and Databricks architecture, to harness current and emerging capabilities.

The use of Agentic AI to take advantage of autonomy and emergent behaviour, brings the Retail Trust to the forefront of technology evolution in the sector
The solution allows Retail Trust to serve highly personalised, auto-generated content that can be easily curated, adapted and improved

Key features of Campaign-in a-Box

  • Wide-scale content curation - the primary agentic workflow has the capability to digest multiple sources of information. For example, looking at hundreds of lines of corporate guidelines, mental health guides, demographic data and feedback, to create more engaging and well-validated content.
  • Automatic data-driven feedback - for each campaign, feedback could automatically be captured from retailers and sent back directly to the Campaign-in-a-Box product. This functionality enabled the AI agents to produce more appropriate content for subsequent campaigns.
  • Independent validation systems - the content validation system worked independently of the generation process so there was greater quality control. This reduces the likelihood of the models automatically falling into the same way of thinking and validating each other.
  • Human-enabled review - the outputs were logged, monitored and validated against the Retail Trust’s content rules and retailer feedback. Content was validated for safety, accuracy, cost, and compliance. Finally, there was a human review of all outputs to ensure strict compliance.
  • Future-proofing - the solution was built with scalability in mind with a low amount of hard coding for future enhancements. Its modular and extendable design will allow the use of future generative AI models and the ability to add more agents to the framework to adapt to new challenges and opportunities as they emerge. 

 

Outcomes

  • Highly personalised campaign content to drive engagement 
  • Ready to roll out to retailers at scale across the UK
  • Reducing the need for human intervention on manual tasks
  • Ease of sharing content on the key topics that matter

 

Transforming efficiency by up to 80%

The use of Agentic AI to take advantage of autonomy and emergent behaviour, brings the Retail Trust to the forefront of technology evolution in the sector. As a result, they now have an end-to-end Campaign-in-a-box service that has transformed both quality and efficiency, reducing manual load by 80%.

From a user engagement point of view, the solution allows them to serve highly personalised, auto-generated content that can be easily curated, adapted and improved. 

Absolutely blown away by the Agentic AI platform running on Databricks that the CGI team have built - from concept to fully deployed into UAT in just six weeks.

Tim Walpole Head of Data and Creative Technologies, Retail Trust

retail shopper uses tablet device in clothes store

Next steps

The Retail Trust plans to further extend the system to refine models over time and introduce other agents as required to improve performance and end-user relevance.

By the end of the year, the goal is to extend the product to over 200 retail clients to generate regular wellbeing campaigns.

 

As the responsible AI experts, we can help you harness the transformative power of AI. Contact us to find out more.