Add summarisation

[antoni-devlin]

Using langchain, I should be able to prompt Ollama's openhermes (or another, faster, model?) to summarise the transcript created by Whisper.

I think these are the relevant Langchain docs.

As long as the Ollama server is running, basic prompting can be achieved by:

from langchain_community.llms import Ollama

llm = Ollama(model="llama2")
llm("<Prompt here>")