I'd like to use OpenAI 4o or Claude Sonnet 3.7 to perform web search for up-to-date generation of text (in my case research on news related to customers). I'm doing this within a Workflow.
In both cases, I start my prompt with the recommended:
You are an expert researcher...{rest of prompt}
OpenAI 4o - Managed
Gives me a very quick response of:
Please hold on while I conduct the search for current information on {client_name}. This will just take a moment.
But the workflow doesn't wait around for the rest of the prompt response...
Any suggestions on how I can "make it wait" for the second half of the OpenAI response?
Claude 3.7 - API Key
I have a Teams key configured, and have enabled Web Search (only available March 2025) as ON. But the result back clearly shows data from 2022/2023, and is not up-to-date with expected "news" for a few test clients occurring in early 2025.
Here's a toy example to reproduce the failure: