Recap ICA x Women in Tech Sweden meetup

From Vision to Everyday Impact

Annie von Heijne, Manager, Data Analytics & Transformation, ICA Sverige, opened the evening by connecting ICA’s 100+ year legacy with its modern digital transformation. She shared ICA’s Vision 2030, which centers around four customer promises:

• Personalised experiences
• Value for money
• Saving time and energy
• Enabling sustainable choices

AI, she explained, is a core enabler of these promises. From dynamic pricing to personalised offers and smart assortment planning, AI is helping ICA improve both operational efficiency and customer experience.

“Vi vill nyttja AI för att skapa värde för ICAs kunder, handlare och bolag – både genom ökad effektivitet och förbättrade kunderbjudanden.”

She also highlighted how ICA balances analytical AI (like demand forecasting) with generative AI, and the importance of investing in skills, processes, tech, and data to make it all work.

The Reality of Building AI That Works

Sofie Lindblom, AI & Data Transformation Lead, Apotek Hjärtat, gave us a behind-the-scenes look at Apotek Hjärtat’s live AI assistant, now handling up to 20,000 customer queries/month. She covered everything from how data is pulled, cleaned, and transformed, to how human and bot interactions are managed.

Her team used customer service queries + e-commerce content to train the chatbot, ensuring it could handle common topics like:

• Product availability
• Order status
• Payment & returns

Key insight? It’s not just about tech. Success hinges on change management, clear internal communication, and deep process alignment.

“Det är inte själva AI:n som är det svåra – det är att förändra våra arbetssätt, roller och ansvar så att AI:n faktiskt kan göra nytta.”

She also reminded us that AI is a spectrum—from skeptics to utopians—and most of us land somewhere in the messy (but exciting!) middle.

Solving the ‘What’s for Dinner?’ Dilemma

Hanna Brännström, Data Scientist, ICA Sverige, wrapped the evening with a fun, practical demo of ICA’s upcoming AI-powered recipe assistant. It helps customers:

• Find meal inspiration (“Jag behöver fika!”)
• Adjust for preferences, allergies, or portion sizes
• Add matched ingredients directly to shopping lists or carts
• Avoid overbuying (hello, 5 packs of salt)

The system uses LLMs (like Google Gemini), vector databases, and custom logic to match vague human requests with real-time inventory and personal preferences. And it’s smart enough to suggest offers and avoid mismatches (no, sylt is not sylta).

“När vi kopplar våra recept till verkliga produkter och personliga preferenser, då kan vi verkligen förenkla vardagen.”

She also emphasized the collaborative effort across designers, data scientists, engineers, and product managers to make this work.