I’m currently working on a project that aims to eliminate matching meetings for niches of consulting / support, etc… in which the quality of the lead is very significant for sales representatives.
The idea is to create a chatbot that replaces the lead forms and knows how to do a high-level matching meeting and save it for the sales representatives. (If the lead is suitable, it sets up a Zoom meeting for him straight away)
For the most part, a good matching conversation both gives value to the interested customer and also verifies through indirect questions whether he is suitable? What is the level of maturity? Does he meet the criteria of a customer that we would like to have, etc…
Because the skills of sales and consulting are intertwined, at first I approached the project and decided to create a consulting agent who knows how to lead to meetings. His goal was to be both a salesperson and a consultant…
As I progress in the project, I receive feedback that it may need to be divided differently.
The division that was proposed to me:
Manager - an expert sales agent who knows how to lead the conversation based on scripts. (The manager’s training is salesperson training + basic adaptation to the knowledge of a consultant and knowledge of lead matching - so that he knows when to use each sub-agent if necessary)
Consultant - has all the knowledge about the specific solution for the client. (The manager can consult with him if necessary.
Lead matching specialist - he knows how to analyze the client’s answers to see if he is ripe and ready for a meeting. (Works based on the company’s lead matching knowledge)
Meeting coordinator. An agent who knows how to actually match meetings.
Quality control agent - reviews the quality of the answers.
Communication improvement agent - responsible for creating curiosity and interest in the conversation through communication that sounds human.
Since I make these calls myself right now, it’s hard for me to imagine how a team of several agents could manipulate such a conversation together.
It sounds like what you need might just be a more advanced version of the examples in this thread. So I’d suggest starting there — take a look at the approaches folks have shared, and then gradually adapt them to fit your use case, one step (or complexity level) at a time.
Yes, my intention is to ask whether the thinking about dividing into agents is correct where the main agent must know at least basic knowledge of all the sub-agents or the approach I took at the beginning? (As stated at the beginning of the question)
Or in simple terms, should we split it into 6 agents or just two?
Since these will be a crew then splitting them into smaller agents (as you said) is best. Agents which have smaller more focused tasks often perform better.
The truth is that I looked at the thread and tried to understand… but since I’m new to the field, I had a little trouble… Can you explain more about what you mean? Thank you very much.
Hey @user11, it’s all good – learning is part of the journey to success. In that case, I’d definitely suggest getting comfortable with this Brave New World.
Here are some excellent resources that will help you get more comfortable and confident, so you can eventually tackle real-world, challenging problems using agentic systems:
Wow… I can’t explain how helpful this is for me. I found a case study on Lennex Zinyando’s exactly the kind of thing I want to build… It’s amazing. Thank you so much for this.
