Heineken
Heineken's Hoppy chatbot cuts information retrieval from 15 minutes to 1 minute
Heineken built Hoppy, an internal generative AI chatbot on SAP's cloud platform that staff reach through Microsoft Teams. Heineken told SAP that the time knowledge workers need to retrieve information fell from 15 minutes to 1 minute. Heineken's annual report also lists it as a tool that supports finance teams.
Key figures
Copied exactly from the sources. Copy any figure with its citation.
from 15 minutes to 1 minute
Time for knowledge workers to retrieve information after Hoppy was introduced, as reported by Heineken through SAP1
What it took
Cost, team size and timeline were not disclosed in the sources.
The problem
Heineken runs in more than 70 countries and operates over 165 breweries. Its business users wanted to find and analyse data quickly, in the language they prefer, instead of searching several systems by hand.
What they built
Hoppy is an internal chatbot driven by generative AI. Staff ask questions in natural language and Hoppy answers from company data and documents. It is available inside Microsoft Teams and other collaboration platforms. The annual report lists it among Heineken's generative AI pilots, describing it as a tool that helps staff navigate global standards, ways of working and processes, and that boosts productivity in finance teams.
How they did it
Hoppy runs on SAP Business Technology Platform, SAP's cloud platform for building and connecting applications. SAP Datasphere provides a single data model across systems. A generative AI hub on the platform handles natural-language questions and more secure access to sensitive information. The sources name no AI model, and they give no cost, team or timeline.
Results
SAP reports that, since Hoppy was introduced, Heineken has seen the time knowledge workers need to retrieve information fall to 1 minute from 15 minutes. With a unified view of the data, Hoppy can also flag missing cost centre information in purchase orders, which SAP links to better accuracy and efficiency.
Lessons for your project
Meet staff where they already work. Hoppy sits inside Microsoft Teams, so people do not need a new tool to ask a question.
Build on one governed data layer. Heineken used a unified data model so the chatbot could query several systems and respect access to sensitive data.
Look for quick data-quality wins. With a unified view, Hoppy can flag purchase orders that lack cost centre information.
Ask the vendor how the time saving was measured before you quote it. The 15-to-1-minute figure comes without a stated method.
Caveats
- The figure comes from SAP's announcement of its customer, so the vendor presents it. Heineken is reported as the source, and no independent party confirmed it.
- The sources do not say how the 15-minute and 1-minute times were measured, how many staff use Hoppy or which markets are covered.
- The sources do not name the AI model behind Hoppy. They name only SAP's platform and Business AI features.
- The trade article on Heineken's 2024 annual report lists Hoppy among its generative AI pilots, so the tool existed by early 2025. No go-live date is given, so the deployed date is the year only.
- SAP says Heineken is still exploring automation of repetitive tasks with Hoppy. This is a plan, not a result.
Sources
- Leading Brands Showcase SAP-fuelled AI, Cloud Innovation at Sapphire 2025
SAP Africa News Center · 2025-05-29 · Press release · news.sap.com
- Heineken placa la sete di dati con l'intelligenza artificiale
SAP News Center Italy · 2025-05-27 · Press release · news.sap.com
- HEINEKEN AI success
PEX Network · 2025-02-25 · News · processexcellencenetwork.com
Summarized in our own words from the sources above. Spotted an error? See corrections.