Turning an h2oGPTe Notebook into a Wave Web App
- Generative AI
- Enterprise h2oGPTe
This closing lesson shows how to turn a notebook that analyzes survey responses with Enterprise h2oGPTe into a shareable web application built with H2O Wave, an open source Python and R framework. The app connects to h2oGPTe with an API key, sends each question's answers to the model to extract the top topics and their occurrence counts, and plots the results as a dynamic chart a non technical user can explore. It also covers the prompt engineering needed for a consistently structured output so the charts render reliably, and closes by pointing to a more built out review analysis app with filtering, trends, and downloadable reports.
Key takeaways
- How connecting to Enterprise h2oGPTe from an H2O Wave app follows the same client and API key setup as the original notebook
- Why the app prompts the LLM to return its topic analysis as a Python dictionary rather than free form text
- What a non technical user can do with the finished app: pick a survey question and see its top topics plotted automatically
- How H2O Wave apps can be published to the H2O Cloud App Store once built
- The kinds of features a more developed Wave app can add, including trend views, data filtering, and downloadable reports
Start owning your AI