From ambition to scalable business value with generative AI
Salling Group, one of the leading retail groups in the Nordics, saw generative AI as the next step in unlocking new business opportunities and improving both customer experiences and internal decision-making. Through a co-build collaboration, Salling Group and Twoday worked to design and implement scalable generative AI solutions.
Client: Salling Group
Industry: Retail
The challenge: How to use generative AI to unlock new opportunities
After several years of successful AI initiatives, Salling Group saw generative AI as the next major level for business impact. The organization recognized that it would be able to help them reimagine customer experiences and improve decision-making across its retail brands.
To turn this vision into reality, Salling Group needed to identify the use cases that could convert generative AI’s promise into measurable business outcomes. That ambition led Salling Group to partner with Twoday.
“We wanted a partner who truly understood generative AI using the Microsoft Azure platform – both the technology, as well as the security and governance requirements. Twoday had exactly that skill set and experience from a wide range of industries. They’ve helped us accelerate our AI ambitions on a number of projects.”
Alan Jensen
CIO, Salling Group.
The collaboration: Co-build ensures learning and ownership
Rather than delivering ready-to-use solutions, Twoday worked side by side with Salling Group in a co-build model. Twoday became an integrated part of the internal teams, contributing their generative AI expertise and knowledge, and guiding architectural and security decisions. This co-build approach ensured that Salling Group could fully own, scale, and evolve the solutions after delivery.
“Co-building was ideal for us. We wanted to build something new and further develop it ourselves. The partnership with Twoday has been based on mutual trust and open dialogue, and Twoday has used their generative AI knowledge to lead the way and support our learning.”
Senior Manager AI, Salling Group.
The results: From experimentation to scalable AI capability
Over the past years, several projects have shown how generative AI can deliver real business value for Salling Group employees and customers.
- A secure and scalable AI assistant
One of the first projects Twoday delivered for Salling Group was the implementation of an internal AI assistant called Sallie. Initially rolled out at the company’s headquarters, the solution quickly proved its value by enabling employees to retrieve policies, procedures, and operational information in seconds instead of searching through multiple systems.
- Support and guidance for customer service agents
Customer service agents have to navigate complex information related to thousands of products across multiple brands with service cases that often involve several stakeholders, such as freight forwarders or manufacturers. This is precisely where generative AI has made a difference by synthesising information from multiple sources and helping customer service agents to respond faster and with greater confidence.
- Inspiring customers in the store
Together with føtex, Twoday helped develop a proof of concept for an AI-driven meal inspiration universe. A key insight driving the project is that many customers enter grocery stores without a clear idea of what they want to buy or prepare for their meals. To address this indecision, the proof of concept leverages generative AI anchored in the product intelligence of føtex’s assortment. It offers customers meal and ingredient suggestions in real-time in the store.
Read the full case
Services delivered by Twoday
-
Advisory
-
System development
-
Project management
Faster and more accurate responses with generative AI
One of the first projects Twoday delivered for Salling Group was the implementation of an internal AI assistant called Sallie. Initially rolled out at the company’s headquarters, the solution quickly proved its value by enabling employees to retrieve policies, procedures, and operational information in seconds instead of searching through multiple systems.
“We wanted to use generative AI to create an internal tool that was secure, robust and could grow and adapt with us. Twoday helped us create a secure foundation for our AI assistant that could easily expand and be used across brands and employee groups.”
Steen Kronborg
Chief Enterprise Architect, Salling Group.
The team decided to build Sallie using Microsoft’s Azure Kubernetes Service (AKS) to ensure secure, scalable deployment, and Azure AI Foundry to develop and manage the generative AI capabilities. Integrated with Azure OpenAI services and supported by Microsoft’s enterprise-grade security and governance framework, the platform provided a reliable and compliant foundation.
Maintaining consistency and accuracy across multiple brands
Scalability was soon needed as Sallie’s success meant it was expanded beyond headquarters for use by employees across Salling Group’s portfolio of retail and franchise brands. It was enriched with brand-specific knowledge, enabling employees to access relevant handbooks, operational guidelines, and internal processes tailored to their brand. This significantly reduced time spent searching for information and improved consistency across the organization.
A particularly valuable enhancement was the integration of standard operating procedures used by the customer service teams. By embedding approved workflows and guidelines into the AI solution, customer service representatives could receive instant, accurate responses to support customer inquiries. This led to faster handling times, more consistent communication, and an overall improvement in service quality.
Simplifying the work of the customer service team
Customer service agents operate in a complex environment with information related to thousands of products across multiple brands, and service cases often involve several stakeholders such as freight forwarders or manufacturers. In that context, generative AI has shown potential as a supporting tool by helping agents access information more easily and draft responses in selected types of cases.
The Twoday team worked closely with Salling Group to develop a customer service AI solution designed to support agents by summarising dialogue with customers and other stakeholders, and by assisting with responses to common first-level interactions.
Supporting the customer service experience
The solution is built and hosted on Azure Kubernetes, combining LLMs from Azure AI Foundry and Redis cache for semantic retrieval and storage. It integrates with Zendesk and was developed with a strong understanding of Salling Group’s customer service processes and ways of working.
Throughout the project, both Salling Group and Twoday gained important learnings about applying generative AI in a highly complex information and system landscape. While this complexity has limited the extent to which AI can simplify workflows end-to-end, the solution has demonstrated value as a support tool, particularly in onboarding new colleagues and helping agents in selected situations.
“We have learned a lot in Salling Group since we started working with Generative AI in Customer Service. Our information and system landscape is very complex, and that has made the journey both valuable and challenging. Together with Twoday, we have used those learnings to shape a tool that can support our colleagues in selected situations — especially new colleagues, who can benefit from guidance directly in the workflow. The solution has shown potential as a helpful support tool, while also making clear that in a complex setup like ours, AI is only one part of the bigger picture.”
Improving the customer in-store experience
Salling Group has also been exploring ways to use generative AI to inspire customers in their everyday shopping decisions. Together with føtex, Twoday helped develop a proof of concept for an AI-driven meal inspiration universe. A key insight driving this use case is that many customers enter grocery stores without a clear idea of what they want to buy or prepare for their meals.
To address this indecision, the proof of concept leverages generative AI anchored in the product intelligence of føtex’s assortment. It offers customers meal and ingredient suggestions in real-time in the store. By enabling tailored, interactive conversations, this solution helps alleviate the mental burden of indecision around meal planning, guiding customers toward options that align with their purchasing interests, health and dietary needs, and recipe preferences.
Personalised meal inspiration for shoppers
For example, if customers are in the store considering buying tomatoes, they can instantly receive inspiration for a tomato-based meal, see the required ingredients and adjust the recipe based on personal preferences. Generative AI enables them to get meal plans and recipes in clear, easy-to-understand language, which makes daily shopping simpler and more inspiring.
“This proof of concept has shown us the potential of generative AI to inspire customers where it matters most – in the store. It can strengthen customer engagement while supporting shopping decisions. I’m excited to see where it leads and how we can scale it throughout føtex.”
Malene Pinstrup Rytter
Brand Manager, føtex.
Are you ready to unlock your data’s full potential?
Get in touch!