Consistently Skipping Conversation Stages

[isaacharlem]

Hello. Thanks for the great work on this agent, it has helped me develop a much better understanding of how to leverage langchain and openai.

That being said, I am having some trouble getting the agent to have a full conversation and address essential stages in the conversation. I made my own custom stages but kept the stage analyzer chain the same (with the new conversation stages). Any idea why this is happening?

[filip-michalsky]

@isaacharlem Make sure the Stage Analyzer and Sales Agent have the stages in sync - are you updating the right dictionary ?

additionally, without knowing what stages you mean, it's hard to help, but I would imagine you might need to work on your prompt. Try this: https://www.promptingguide.ai/techniques/fewshot

[ZhanarBaken]

@isaacharlem
Hello! I believe the result of the determine_conversation_stage function is not being used in run.py and in the prompt.
I added sales_agent.determine_conversation_stage() to run.py, included "conversation_stage" in the input_variables of SALES_AGENT_INCEPTION_PROMPT within chains.py, and added Current conversation stage: {conversation_stage} in SALES_AGENT_INCEPTION_PROMPT from prompts.py.
After these changes, it started working as in sales_agent_with_context.ipynb.

@filip-michalsky Why is it like that?