Home

Clay Announcements
Clay Support
Discuss Anything
Feedback & Requests
GTM Career Advice
Introduce Yourself
Member Spotlight
Share Content and Events
Share Full-time Jobs
Share Part Time Jobs
Share Tables, Tools, and Workflows

Community Guidelines

Explore Events
Clay Cup

Powered by Tightknit

Issue with Claygent Web Researcher Tool Call Limits | Clay

Issue with Claygent Web Researcher Tool Call Limits

·Dec 05, 2025 05:11 PM·

When using Claygent web researcher, I'm having a new issue today where a decent percentage of prompt responses are mentioning not being able to perform web searches and hitting tool call limits. I have ran identical prompts (usually includes visiting an account's site and assessing firm location and product offerings) with the same model (GPT-5-Mini) in the past and have not experienced this. Has Clay made any changes to LLM call settings that are causing this? If so, is there any user-configuration I can do on my end to allow it more tool calls? Thanks

17 comments

· Sorted by Oldest

Clay Support
APP
·
·
Yes, there have been recent changes. We updated Claude Sonnet 3.7 integration which doubled the estimated token usage per Claygent run. This more conservative approach minimizes failures but can cause rate limit issues.
For your OpenAI API, you need 450,000 TPM for reliable Claygent web research. Check your API usage tier at https://platform.openai.com/settings/organization/limits to ensure you meet this requirement.
If you're hitting budget limits, you can increase the "Clay Credit Budget" in your AI column settings to allow more credits per row for complex prompts.
Clay Support
APP
·
·
Is that what you were looking for?
Aubry S.
·
·
No. I use my own OpenAI API key and am Tier 4 so should not be hitting any rate limits. I am not hitting budget limits on these prompts. If anything, I'd love for them to use more tokens to respond more consistently/accurately.
Clay Support
APP
·
·
I understand your frustration, and I’ll connect you with a human agent who can assist you further with this issue.
Clay Support
APP
·
·
Our support team has got your message and we'll get back to you soon!
If you’re dealing with a specific table, drop the URL below so we can help you quicker. Otherwise, someone from our team will be in touch soon!
image.png(93 kB)
Aubry S.
·
·
Some examples of this
Sayli G.
·
·
Hey Aubry! Apologies for the experience you had. Do you mind sending over the table link URL so that I can take a look!
Aubry S.
·
·
Sayli G. Sure thing! https://app.clay.com/workspaces/835672/workbooks/wb_0t6pr04zHam5QWGUMWv/tables/t_0t6r09ad7dDqieCQjRr/views/gv_0t6r09aoTbAEW96B9JP
Aubry S.
·
·
I've been trying to whittle down the number of failed rows. I made another "Use AI" column named "Web Search Error" that analyzed the Reasoning output from the Tax Prep US Status Use AI column and parsed if the web search failed or not. After filtering to the failed ones, I've re-ran the main LLM call and repeated a few times. Have now gone from 2.4k failed rows to only 129. I've ran nearly identical prompts in the past though with nowhere near this high of a failure rate, so I feel like something funky is going on
Aubry S.
·
·
Sayli G. Further info supporting my theory that Clay has modified the system prompt or some other parameter to constrain the model
LuisArturo
·
·
Hey there Aubry thanks for sharing jumping in for Sayli here, taking a look for you.
LuisArturo
·
·
Hey there Aubrey looking through this are there any particular rows you are still seeing this in so far, In the Rose I've checked where the collegiate failed to find a result I'm not seeing any mention of web search errors or call constraints occurring.
Aubry S.
·
·
LuisArturo Hi, here's an example of one that is still having this issue
Aubry S.
·
·
Row 11 with current filtering
Aubry S.
·
·
Is it possible that the recent product updates on 11/26 modified the system prompt of these LLMs to constrain tool calls to only 1? I never saw these issues on runs prior to this