Feedback and Questions Regarding Avatar Programming on the Professional Plan

Dear Convai Team,

A cordial greeting to the entire forum.

We are writing on behalf of TANK, a company using the professional Convai plan, to seek your guidance on a few points regarding avatar programming for more accurate feedback.

We have encountered some challenges that we would like to bring to your attention:

  1. Response Consistency: We have noticed that the avatar occasionally generates inconsistent responses, which can lead to confusing feedback for our users.

  2. Feedback with Long Inputs: We have observed that feedback is not provided when a person’s speech exceeds approximately 500 words. This is a concern as the system handled longer inputs correctly at the beginning of our subscription.

  3. Depth of Analysis: Additionally, we feel that the depth of the feedback analysis was more comprehensive when we first started using the service.

To help us better understand these points and optimize our use of the platform, we would like to request a conversation with your team.

Furthermore, it would be incredibly helpful if you could provide us with an example or a case study of a highly developed avatar. Seeing a best-practice example would give us a much clearer understanding of the full scope and potential of what we can achieve with our own avatar.

Thank you for your time and attention to this matter. We look forward to your response.

Best regards,

The TANK Team

Hello,

Welcome to the Convai Developer Forum!

We’re sorry to hear about the issues you’re experiencing.

  • Response consistency depends on the LLM you’ve selected, your character description, as well as the LLM and Temperature settings in your Core AI Settings. For more details, please check our documentation.
  • For your second and third points, I didn’t fully understand the issues. Could you please share more details, along with examples, screen recordings, your character ID, and session ID? That will help us investigate and provide better guidance.