A mini interview with the BeebleBot team
We sat down with the team behind Beeblebot, an AI assistant designed to streamline access to information. They share insights into the creation process, challenges faced, and their vision for AI in the future.
As a digital service studio, we’ve implemented Beeblebot as our new internal assistant tool. It provides easy access to our company handbook, covering everything employees need to know about life at Humblebee.
Hey! How did you come up with the idea for the AI Assistant?
Benny: Not sure if I should take the credit for the Beeblebot. But since Olga, Andreas, and Johan were looking into creating services with AI, we needed a project to try out and test the new technologies. I recently looked up some information on our Beeble website and thought there must be an easier way to quickly access this information. Feeding an AI with the data and connecting it to a tool like Slack that we use every day was a typical "let's spend a few hours automating something that takes 5 minutes."
What was the most enjoyable part of working on this project?
Olga: The coolest part was transforming a boring document into an interactive interface. Seeing that relevant information can be accessed more easily is satisfying. Especially since it doesn’t require any additional tools and can be done through Slack, a medium that we all use every day.
Did you encounter any challenges or unexpected twists during development?
Benny: A common problem seems to be getting the latest data in a structured way. If we have more time to spend on the Beeblebot, it would be nice to ensure it automatically gets the latest updated information from our Beeble.
What technology did you use to create the assistant? Andreas: This project uses Azure AI Search to find information and Azure OpenAI to generate responses. Key technologies include Python, TypeScript, Azure AI Search, and Azure OpenAI.
What is the main goal of the AI Beeble Assistant?
Johan: Main goal was to learn about this new technology so we are prepared when customers start asking for solutions involving it. Turns out that happened sooner than later and we are now already running our first client project integrating LLM and RAG technologies into a customer solution.
Share your opinion on the future of building services with AI technology/smartness.
Johan: As experts in building digital services, our role is to make AI technology tangible and practical for our clients. It’s crucial to demonstrate its real value through direct application in their operations. This requires us to identify and implement solutions that offer the most significant benefits, a challenge we’re already familiar with in our work at Humblebee.
AI is an incredibly powerful tool in our technology toolbox. But if misapplied, it can lead to skepticism and confusion. Therefore, our current focus should be on mastering the correct use of AI. By doing so, we can seamlessly integrate these insights into creating innovative digital services. And thanks to our great team of Technology and Design experts at Humblebee we’re already ahead on that journey. Let's go create!