While creating quick suite agent, is it okay to have a single ~ 20 page long prompt document? Will it have issues with agent performance because of it’s long nature which might increase the token numbers to be processed by the model?Is there a best practices doc for prompts while building QS agent?
Hi @revsp and welcome to the Quick Suite Community!
In regards to a longer prompt document; it may alter the response time but in terms of an ‘increase in token numbers processed’; there is very little documentation available that explores this.
When adding reference documents however, I believe the chat agent may take that information and save it to it’s ‘persona’. The agents are continuously adapting their personas and response styles based on your prompts and documents regarding that.
So from this, I would think that it should not create an increase in token numbers processed.
Regarding best practices; while the information available is still fairly light, I would suggest checking out this documentation:
2 Likes
Hi @revsp - No, the prompt should not be long enough. at end of the day, prompt should be crispy with limited line so that agent can take it and performed well. if you have a big document, this can be used for ingestion and used for reference. What is the exact use case you are trying to achieve.
Regards - San
Hi Sanjeeb,
I am trying create a chat agent for a domain specific analytical use case which involves users asking questions about metrics/ KPIs, insights about some underlying data, etc.
One thing we are striving to do is that, we are creating a standardized prompt input through reference document which can be used across multiple domains with portions of doc being common across domains (such as general fabrication rules, etc.) while others being customizable for that specific domain such as metric formulas. how should we structure the single prompt doc? or will there be a difference in the agent’s performance if the standard prompts are put in one doc and customizable ones are kept in another? Or any suggested best practices for this use case?
Hi @revsp,
Hope everything is well! I also agree with Sanjeeb that your long prompt should be reformatted into something like a PDF to act as a reference document. With that, the structure I believe does not matter too much. I would just make sure to follow suggestions as stated in Brett’s linked documentation:
Hope this helps!
Hi Luis,
Yes, I am using a pdf document as of now as the reference document. The number of pages in the pdf document (if >20 pages, will it make a difference to the chat agent’s performance?)
Hi @revsp,
I do not think that will matter at all! As long as the info in the document is consistent throughout, everything should be fine. In terms of approach, if there’s anything you want it to specifically target in terms of responses, I would keep trialing your chat agent it until it meets your goals.
1 Like
Hi @revsp,
Just checking back in since we haven’t heard from you in a bit. I wanted to see if the guidance shared earlier helped resolve your question, or if you found a solution in the meantime.
If you still have any additional questions related to your initial post, feel free to share them. Otherwise, any update you’re able to provide within the next 3 business days would be helpful for the community.
Thank you
Hi @revsp,
Since I haven’t received any further updates from you, I’ll treat this inquiry as complete for now. If you have any additional questions, feel free to create a new post in the community and link this discussion for context.
Thank you