Getting answers to the wrong conversations

[Explosion-Scratch]

Settings are at default

[clement2026]

Thank you for pointing this out!

The talk.proxoar.com showcases intuitive UI interaction designs. It uses pseudo data to prevent token consumption. It’s important that this information is clear on the site, and an update to reflect this will be deployed shortly.

For genuine AI responses, consider setting up your own instance with the following config sample.

talk.yaml:

llm:
  chat-gpt: open-ai-01

creds:
  open-ai-01: "[Your OpenAI Secret Key]"

(Note: The actual OpenAI secret key should replace the placeholder above for functionality.)

To configure a voice-interactive instance, please refer to the README.md documentation.

[Explosion-Scratch]

@proxfly

Still happening

[clement2026]

Could you provide additional information to assist in diagnosing the issue more effectively?

• Were you using talk.proxoar.com or your own deployed instance?
• Could you please share the content of your talk.yaml file (ensuring the API key is sanitized)?

If you were using talk.proxoar.com, it's important to note that it is a demo and does not utilize the real ChatGPT, therefore, the response is expected to be incorrect.

[Explosion-Scratch]

If you were using talk.proxoar.com, it's important to note that it is a demo and does not utilize the real ChatGPT, therefore, the response is expected to be incorrect.

Oh yeah, I was using the demo. Maybe mention that in the responses lol

[clement2026]

If you were using talk.proxoar.com, it's important to note that it is a demo and does not utilize the real ChatGPT, therefore, the response is expected to be incorrect.

Oh yeah, I was using the demo. Maybe mention that in the responses lol

It's a good idea to mention that in responses👍. An update should be rolled out later today.