A sample app for the Retrieval-Augmented Generation pattern running in Azure, using Azure Cognitive Search for retrieval and Azure OpenAI large language models to power ChatGPT-style and Q&A experiences.
When a user invokes Voice Chat to ask a health care coverage question, they get a completion informing them the assistant is there to answer development questions.
Wouldn't it be a better idea to change the system prompt...
new ChatMessage(ChatRole.System, """
You're an AI assistant for developers, helping them write code more efficiently.
You're name is **Blazor 📎 Clippy** and you're an expert Blazor developer.
You're also an expert in ASP.NET Core, C#, TypeScript, and even JavaScript.
You will always reply with a Markdown formatted response.
"""),
... to align with the documents being fed into Cognitive Search?
Minimal steps to reproduce
Run the app, go to Voice Chat and ask a question about health coverage.
Expected/desired behavior
A system prompt instructing the assistant to answer health insurance questions.
OS and Version?
Win 11
Mention any other details that might be useful
Think about changing the context of the system prompt. Someone new to OpenAI will most likely not immediately catch that problem.
Here's an effective bug:
When a user invokes Voice Chat to ask a health care coverage question, they get a completion informing them the assistant is there to answer development questions.
Wouldn't it be a better idea to change the system prompt...
... to align with the documents being fed into Cognitive Search?
Minimal steps to reproduce
Expected/desired behavior
OS and Version?
Mention any other details that might be useful