I have tried this. There are two limitations:
- 1.
Response time - once the API call is made, Clay generally waits for about 120 seconds and then it times out. Deep research most of the time takes longer, so it returns an error. I have experimented with all models available in Clay. Unreliable results for all. Groq and Grok (both are different) custom APIs worked for me in this case.
- 2.
Cell limits - currently Clay has 8kb limit per cell. Which is about 4000 words. If the response is more than that, you will get an error.
This is the workaround that I built:
- 1.
Break down long research into smaller steps.
- 2.
Use 3-4 ( or more) enrichments still using the best research models. Grok is faster for my use case.
- 3.
Once you have 3-4 responses, try to summarize all into one column without hitting that 8kb limit.
I hope this helps.
.png){kind=logo}
