FAQ
Who made this? Technically Sharon Machlis, but in large part this was coded with generative AI 😅 using Google's Gemini 2.5 pro large language model with help other LLMs including Claude Sonnet 3.7 and various ChatGPT models and custom GPTs. It uses technology similar to that behind ChatGPT specifically to answer questions only about the Framingham recycling program, but the idea could apply to a lot of other government services and information. This is not an official city of Framingham app.
Where does the information come from?
Data come from one page on the city's website and a few posts by Framingham Recycling Coordinator Eve Carey, but the app is not affiliated with the city. You can see official information about the Framingham Recycling program at Framingham Curbside Recycling.
What can I do with this? You should be able to ask Recyclebot things like
- Can I recycle pizza boxes in Framingham?
- What types of plastic can I recycle in Framingham?
- Can I recycle shredded paper?
How does this work? It first analyzes your question and 'translates' it into a series of numbers called embeddings, and then checks the documents to find excerpts with embeddings that are most similar to your question. Then, your question and those releative chunks are sent to a Large Language Model (same tech as behind ChatGPT) to generate an answer.
What LLM does this chatbot use? It currently uses OpenAI's gpt-4o-mini, which performed a bit better than Gemini Flash 2.0 and is less expensive than Anthropic's very capable models.
What software does it use? This app uses the R Shiny Web framework and a lot of cool R packages including shiny, shinychat, ragnar, ellmer, and reactable.
Can I see the code behind this? Not this specifically, but I've got code for a similar app on GitHub at https://github.com/smach/QuartoTips