Replied AI Assistant configuration options

[Anishek]

Grant Account Access for Support

1. I authorize the Pabbly Support Team to access my account for troubleshooting purposes.

WhatsApp Number

+918340182405

Email Address

[email protected]

Hi Pabbly Chatflow team,

I have two questions related to AI Assistant configuration and these are interlinked with each other:

1. Under AI Configuration, where we have to choose the AI tools to use, can we select some other LLM model other then Open AI? And if yes, then what's the way to do it?
2. I build the WhatsApp agent and deployed it for my first client. The usage cost that I'm paying has been close to $10 for just 4 days and I have used models like GPT 4o, still the costs seems high. What do you think what should I do and how can I navigate this (as I kept the margins thin, while selling to client)?

Looking forward to your answers.

Regards
Anishek

 

[Preeti Paryani]

Hello @Anishek,

Thank you for reaching out.

Regarding your first question, currently it is not possible to use LLM models other than OpenAI models inside the AI Assistant configuration of Pabbly Chatflow. At the moment, only OpenAI-supported models can be used while creating the AI stage/assistant.

Regarding the second point about higher-than-expected GPT usage costs, this mainly depends on how the AI assistant is configured and how frequently/prominently the AI responses are being used in conversations. Models like GPT-4o can increase costs quickly depending on the message volume and token usage.

In such cases, our recommendation is to use the client’s own OpenAI API key instead of your personal API key. This way, the OpenAI usage charges will be billed directly to the client’s OpenAI account rather than yours, helping you avoid absorbing recurring AI costs from multiple client deployments.

You can also consider:

• Using lighter/cheaper models where possible
• Reducing unnecessary AI prompts/responses
• Limiting long conversation contexts
• Using AI only for specific stages instead of every message

Please let us know if you need any further guidance regarding optimization strategies.