Secure Internal Chat with Azure OpenAI
Customer
Swiss Chemical Company
Provide a secure internal chat based on Azure OpenAI, guidelines for developing LLM solution and secure prototyping environments for the organisation.
Challenge
- Large Language Models are a significant business demand but can pose various risks
- Being able to adequately review and evaluate solutions built on top of the new LLM technology is challenging organizations
- Significant risk exist due an estimated 11%-25% of sensitive company data being submitted to public OpenAI
Solution
- Deploy an internal secure and private ChatGPT
- Develop a set of guidelines for the solution architecture review board on how to evaluate LLM based applications
- Deploy an internal sandboxed environment for developers to safely build prototypes of LLM applications
Business Impact
- Significantly reduced risk of leaking organizational information through public ChatGPT tools
- Enabling the tech organisation to evaluate LLM applications faster, consistently and establishing a unified architecture
- Prevention of shadow IT through establishing an easy-to-use unified architecture