Inside LLMs mind is a small app to demonstrate that LLMs are not essentially black boxes that gives out smart anwers out of the fog. They can be carefully crafted and designed to answer specific needs.
In this app, two LLMs have been chained. One can retrieve info on Internet, the other perform math calculations. Depending on your prompt it will use one of them.
Details about the app on my research notebook : https://dataflow.hypotheses.org/1068