The Claygent Argon model hasn't been "nerfed" - it's still our strongest overall model for deep research and complex analysis. [1]
However, there have been recent updates to Claude Sonnet 3.7 integration that doubled the estimated token usage per Claygent run, making it more conservative but also more stable.
For accessing company and people data (not LinkedIn specifically), Claygent can still pull this information effectively. If you're having issues, try checking if you're hitting rate limits or consider adjusting your prompts for better results.
The reason you may not see some LinkedIn profiles come through is because if a profile has limited public visibility, you're unable to retrieve the full profile content.
I'd recommend using our Enrich Person action for live data on people's profiles.
