For over 30 years, FREITAG has been synonymous with bags steeped in history \u2013 made from recycled lorry tarpaulins, with every piece being one of a kind. The company has been using Salesforce since 2011 to offer customers a consistent, high-quality brand experience. For a long time, digital customer service relied on a chat-based support model which, whilst feeling personal, involved a high administrative burden, had limited scalability and allowed only limited analysis of customer interactions. When Salesforce announced Agentforce at the end of 2024, Katharina Hornschuch, CRM & Analytics Product Owner at FREITAG, immediately began evaluating the new solution. \u201cWe wanted to develop an AI agent for our customer support that could reliably answer typical customer enquiries whilst being fully integrated into the Salesforce platform,\u201d she explains. \u201cOur aim with Agentforce was to reduce the growing support workload, gain more control over responses and ensure a consistent, high-quality customer experience \u2013 based on robust data and a scalable system architecture. \u201cWith FRIDA, we have introduced an AI agent into our support team that continuously learns and helps us to better understand customer concerns and optimize our service in a targeted manner.\u201d<\/p>\n\n\n\n
\u201cIt quickly became clear that an AI agent\u2019s answers are only as good as the information it uses. That\u2019s why we first laid the groundwork and built a structured knowledge base. Previously, a lot of information was scattered across various documents or people\u2019s minds. For the first time, we\u2019ve brought everything together in one place, structured it and organized it clearly. FAQs, service information and campaign-related content were transferred to Salesforce, supplemented by clear responsibilities and processes for maintenance and approval. In addition, we were able to integrate further data sources via Data 360. For example, FRIDA retrieves product information directly from our website via a sitemap connector. This ensures that FRIDA always has access to up-to-date information.\u201d<\/p>\n\n\n\n
\u201cWe deliberately launched with a clearly defined pilot project. Germany and Switzerland were our first markets, with real customers and genuine enquiries, so that we could gain insights based as much as possible on practical experience. FRIDA was introduced there transparently as an AI assistant and took over the task of responding to incoming website enquiries. At the same time, we defined clear boundaries, such as everything FRIDA should not answer. For topics outside the stored knowledge base, such as press enquiries or partnerships, the agent reliably refers the matter to our support team, ensuring responsibility always remains with the correct contact person. We made intensive use of the first few weeks of live operation to further optimize the system\u2019s deployment. We analyzed every conversation in detail, reviewed tone and response quality, and fine-tuned the escalation process. The results were very positive right from the start, with satisfaction rates of over 95 per cent. At the same time, FRIDA is in a continuous learning process: recurring topics and knowledge gaps are identified, transferred to the knowledge base and thus directly improve the agent\u2019s performance. This iterative approach is a key success factor.”<\/p>\n\n\n\n
\u201cFor us, FRIDA is a digital colleague in the Customer Support Team. It handles the many recurring standard enquiries and provides our customers with quick and reliable assistance. This gives our team noticeably more freedom to focus on more complex, consultation-intensive or even more emotionally charged issues. This is changing the way we organize our service. It is no longer just about automation, but about the interaction between humans and AI. We deliberately think in terms of hybrid teams, in which both sides contribute their respective strengths \u2013 and this is precisely where we see the true potential of Agentforce. The launch of FRIDA \u2013 a name chosen by our service team and derived from the English translation of our company name \u2013 was just the first step for us. Looking ahead, we want to further develop the AI agent and integrate it more deeply into our systems, for example by linking e-commerce data. This would allow FRIDA to be used in future for the somewhat more complex repair process, which plays a central role for us. We see great potential for AI-supported processes and a significantly better customer experience as a result.\u201d<\/p>\n\n\n\n