Open source instead of ChatGPT: How BSI harnesses AI’s potential for customer relationship management

Baden, July 14, 2023 – The introduction of ChatGPT has brought renewed attention to the opportunities and potential application areas of Artificial Intelligence (AI); AI also presents entirely new possibilities when it comes to designing digital customer relationships. The customers of BSI, the Swiss software company, already benefit from it today: BSI has been using Artificial Intelligence and language models for customer relationship management in its projects since 2018. By now, these open-source language models have become highly developed and come close to ChatGPT in terms of their capability. With the BSI Customer Suite, companies optimize their marketing, sales and service operations with state-of-the-art technologies – all that in compliance with data protection regulations and on the basis of high ethical standards. In addition, BSI’s AI concept gets high marks for a number of strategic advantages.

Easy and fast deployment of new LLMs

BSI's platform approach makes it easy to connect LLMs (large language models) to the BSI Customer Suite and integrate them into any module. Since BSI consistently relies on open-source AI, the company can quickly turn to the highest-performing model and make it available to its customers when there are new technical developments or compelling providers have entered the market. Just like ChatGPT, BSI's AI is pretrained, which is to say that companies using the BSI Customer Suite can start with specific applications right away. Using open-source models also ensures that a model can be retrained for a different application, if necessary.



High ethical standards and maximum data protection with open source

ChatGPT is a proprietary solution from OpenAI; by contrast, BSI uses open source, an open architecture and transparent training data, providing transparency as to the training data the AI bases its decisions on. This, in turn, makes it easier to examine data-related ethical issues such as bias, toxic language or social stereotypes, since BSI can examine the training data specifically for those matters and respond appropriately, if necessary. Since BSI runs the individual open-source model itself, no data is transferred to the U.S. or other countries with less stringent data protection regulations.

All of this ensures adherence to BSI’s high ethical standards: These standards are firmly embedded in the software company’s Code of Conduct, where BSI commits itself, among other things, to digital ethics considerations in the development and handling of AI systems. Moreover, the Swiss association Swiss Insights, with its “Data Fairness” label, identifies BSI as a company that deals transparently with data processes and actively engages with digital ethics.

Independence from AI models and AI solutions

With its platform model, BSI is not tied to any providers (as ChatGPT is tied to OpenAI) and can therefore incorporate other AI models that are brought to the market as well. This means that BSI’s Customer Suite is always state-of-the-art, and customers do not risk being dependent on a specific model or incurring high switching costs. 

Multiple applications are already available

For example, BSI’s AI automatically categorizes incoming e-mails and generates response suggestions, reducing customer representatives’ everyday workload. The reps are given an AI co-pilot which, for instance, extracts the key items from correspondence, loads context-relevant data from the data model or generates suggested responses. As a result, customer reps no longer need to do their own research yet continue to retain control over their communication with customers. Data analytics provides valuable insights, such as trend development and customer satisfaction. This customized sales engagement, combined with highly individualized product recommendations, gives customers a strong sense of receiving personalized attention, thus strengthening customer loyalty.

Master data management: expansion of collaboration already planned

“When it comes to large language models (LLMs), BSI relies on open-source options. The LLMs are hosted in the BSI cloud and are therefore under our and our customers’ full control. This means that data does not have to be shared – personal data can remain in the same security zone. Open-source models help meet BSI’s high ethical standards, as both the model and the underlying data can be examined for application-specific questions, such as bias. Many companies have counted on us for several years already when it comes to using Artificial Intelligence. We continue to incorporate the feedback we receive about the many implemented use cases into product development – which also benefits our new customers.”
Head of AI at BSI