Seems like everyone is building a ChatGPT thing right now, doesn’t it? Well we are too! Inspired by so many others, we decided to see what AI could do with our simplified analytics and observability data.
Turns out, it can do quite a lot.
I’m thrilled to share that we’ve shipped our first AI insights chatbot, Professor Sloth. With this ChatGPT-powered AI Assistant, Request Metrics is taking data analysis to the next level by helping you spot anomalies, suggesting courses of action, and guiding you towards valuable insights.
But this wasn’t without it’s challenges…
Prompt Engineering Challenges
Integrating OpenAI’s ChatGPT into Request Metrics proved to be an interesting challenge, particularly when it came to the emerging problem of prompt engineering. To get correct and interesting results, we had to talk to this supremely powerful AI like it was a first-grade math student. Phrases like “go slow”, “check your work”, and “let’s take this step by step” litter our prompts.
Without these statements, we found that ChatGPT would often outright lie to us, reporting median values that weren’t in the dataset, or reporting trends that were not statistically relevant. Much like guiding a young learner, it was essential to provide clear instructions and set expectations to ensure that ChatGPT could understand and analyze the data effectively.
With a little care and patience, this approach allowed us to harness the power of AI while maintaining a high level of accuracy and relevance in the insights generated by Professor Sloth.
Professor Sloth and Data Analysis
Incorporating an AI Assistant like Professor Sloth into your data analysis workflow offers numerous advantages. Here are some benefits we’ve observed while using Professor Sloth:
Professor Sloth can quickly spot anomalies in your data, helping you identify potential issues and areas for improvement. This can save you time and effort by automating the initial data exploration process. Here’s an example where the professor found an API Endpoint anomaly, and pointed out that they are all SVG file requests.
Our ChatGPT-powered AI Sloth suggests courses of action based on the data it analyzes. This helps you make informed decisions on where to invest your resources and how to improve your website’s performance. Here’s an example where it found the largest web analytics sources of traffic, and recommended additional focus.
Professor Sloth recommends additional data analysis steps to help you dive deeper into specific areas of interest. This can be particularly useful when you’re unsure where to start or need help identifying new avenues for exploration.
Here, Professor Sloth suggested some geographic filters to get more relevant data, focusing on US and German traffic.
Ongoing Training for Professor Sloth
Currently, Professor Sloth is available in a handful of locations throughout the Request Metrics UI. However, we have big plans for our AI Assistant. In the coming weeks, we’ll be expanding its presence within the platform and teaching it about more types of data.
As the AI APIs continue to evolve, we hope to expand Professor Sloth’s chat capabilities, allowing you to ask further questions about your data and receive even more personalized insights. With this ChatGPT-powered feature, we aim to make data analysis more accessible, efficient, and enjoyable for all users.
Professor Sloth shows off the power and potential of AI integration in data analysis. By overcoming the challenges of prompt engineering and leveraging the capabilities of OpenAI’s ChatGPT, we’ve created a valuable tool that can revolutionize the way you approach your website’s data.
With Professor Sloth, you’ll gain access to anomaly detection, actionable recommendations, and guided analysis, all powered by cutting-edge AI technology. As we continue to expand and evolve this feature, we’re confident that Professor Sloth will become an indispensable asset in your quest for data-driven insights and website optimization.
Ready to try it out? Check out our live sandbox and see what Professor Sloth has to say about the data. Or sign-up for Request Metrics free and get AI-powered insights on your own analytics and observability data!