jamie.ai

• 

  Logic flow

Inspiration

Initially we thought of the role Young Jamie plays in the Joe Rogan podcast. Having someone who is constantly searching the web for information based on whatever the co-hosts are talking about to provide fact-checking and additional context.

However, this same flow of combining a Search Agent and Conversation Agent could be applied to fact checking political debate or sports commentary.

What it does

1. Browser's built in Text-to-Speech converts audio to transcript which its then chunked.
2. Chunks are sent to OpenAI for entity and topic extraction. If entities are found, a search query is constructed.
3. That search query is then run by Perplexity to surface relevant information.
4. That information is uploaded to the ElevenLabs Knowledge base.
5. At the request of on of the podcast co-hosts, the Conversation agent will present the latest information it has in its knowledge base.

What's next for jamie.ai

We would like to extend the role of the Conversational Agent to allow the co-hosts to ask followup questions to what is presented, thus creating a cyclical flow. Using the client-side tooling, the conversation agent could also make actions on the browser to navigate through the links within it's knowledge base.

Built With

• elevenlabs
• openai
• perplexity
• python