The FlexChat conversation gives you complete control and the most flexibility in your interactions with ChatGPT.

By controlling/editing the User prompts and the Assistant responses, you have full control over what ChatGPT “remembers” allowing virtually limitless conversations.

By controlling Assistant responses, you can tell ChatGPT what it said, then continue your conversation from there. This is much more efficient and effective than trying to correct your prompt to get ChatGPT to respond how you wanted it to.

Click the New Conversation tool on the FlexChat Ribbon to create a new conversation table.

If you want to change the default settings for new Conversations you can do so in Settings. Also, you can use the advanced settings to indicate the exact values that correlates to these drop-down options. For example, when OpenAI comes out with the next AI model, you can indicate that you want to use that as the Advanced model in Settings. See FlexChat.ai Settings Help for more information.

Chat Model

In a conversation you can interact with the Standard or Advanced model.

For advanced users, you can define exactly which OpenAI API model is the Standard model and which model is the Advanced model in Settings. Each model has it’s own capabilities and cost.

Style

The Style value controls the randomness of the response. Style can be Precise, Balanced or Creative. Precise results in more consistent outputs, while Creative generates more diverse and creative results.

For advanced users, you can define what Precise, Balanced, and Precise mean in Settings.

Length

Length refers to the length of the response from the Model. You can set this to be Any, Short, or Long. Limiting the length of the response can be a good thing because you get charged for both the length of what you send, and the length of the response.  How much you’re charged varies based on the model you use, but you’re not going to go broke here.

“Any” means the AI will respond with as much text that it needs in order to give you the full response that it generates. If you select “Short”, you may get an incomplete response.

For advanced users, you can define exactly how long is Long and how short is Short in Settings.

Custom Instructions

Custom Instructions guides the model’s behavior throughout the conversation.  “You are an SEO expert” or “You are Darth Vader” for example.

User Prompt

Contains your prompt (question, statement, or set of instructions) to be sent to ChatGPT. The prompt is a piece of text or a specific instruction given to ChatGPT. It helps guide the AI and provides context for generating a response.

AI Response

This is where you’ll see ChatGPT’s response to your Custom Instructions and User Prompt, but more powerful customizations comes from the realization that you can provide or edit these responses! By controlling AI responses, you can tell ChatGPT what it said, then continue your conversation from there. This is much more efficient and effective than trying to correct your prompt to get ChatGPT to respond how you wanted it to.

Memory

Ok, there’s no “memory” field but ChatGPT’s memory is the Conversation. By controlling/editing the User prompts and the AI responses, you have full control over what ChatGPT “remembers”, this capability combined with having multiple conversations in your word document, allows you to have virtually limitless conversations. This overcomes a limitation of many other ChatGPT tools because when you don’t have control, ChatGPT often forgets key points or earlier discussion and can “lose it’s way” as you carry on your conversation.

FlexChat

When you’re ready for your AI Response, click the Flexchat tool.

Also see Simplifying the Chat Interface by Hiding Convention Rows.