Hi, for row 41, why does it not get the right info? https://app.clay.com/workspaces/271261/tables/t_NnKycXajwhga/views/gv_m5ve9do8inMn If you go to their site, you'll see they have jobs posted. They also didn't raise money.
This discrepancy could be happening for two reasons: 1. If you're looking at job listings, the table might be set up as a static source rather than a dynamic one. To get updated job information, you'll need to change it to a dynamic source and set up scheduled imports to automatically fetch new entries. 2. For company data like fundraising information, there may be limitations from data providers or privacy restrictions that prevent real-time data access. Our providers don't use active sessions or cookies to scrape data live, which can result in differences between what you see when logged in versus what the enrichment shows.
Did that answer your question?
Looks like you're still looking for help. We're currently outside of our standard office hours (9 AM to 9 PM EST, Monday through Friday), so it may take a little longer for us to respond.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!
Hey there Luka thanks for reaching out, taking a look at this, this may have happened due to the AI being overwhelmed with the prompt. One solution you can try here is using the Meta prompter, which can be found by selecting the help me button in the AI prompt box. What this will do is help rewrite the prompt in a that the AI can better read in and execute. https://downloads.intercomcdn.com/i/o/w28k1kwz/1376266729/676f1933c723578d31c55612a3e1/image.png?expires=1739322900&signature=6b20796a3cfb4172dc9f7f12da92bac2307a554f14e2a487b81a18d59c606ce5&req=dSMgEMt4m4ZdUPMW1HO4zehty1RpJtAeaYZ2hsWG5zGcsoCVPmUbGSie4YYH%0ARv7A%0A
Thanks LuisArturo. I tried it but it doesn't seem like it's working still. If you look at row 41. Hiring: It says they're not hiring but they are: https://utility.agency/jobs#joblist (this is a very easy one - don't understand how it didn't get this one) Non_USengineers: it says there are no non-US engineers but there are many: https://www.linkedin.com/company/utility-agency/people/?facetGeoRegion=103813819 Funding_Round: I can't find any news that they've had a funding round but it says they did Employee_size_growth: I checked LI and it doesn't seem like they've grown _______ Every single field is incorrect. I don't understand how I'm supposed to automate this when I have zero confidence in the data.
A few other unrelated questions:
how do I turn off the automatic running of AI fields - I selected the companies and automatically it started running the AI portion so I used up many credits
I heard that Clay shouldn't be used on getting emails and AI (I should have my own API keys instead). What should Clay be used on then?
Hi Luka! This happens because the AI model as a small token window and is a cheaper model, meaning it can only process limited information at once. When looking up company data like job posts and LinkedIn profiles, it sometimes can't access all the details due to that data being behind a login - leading to wrong or incomplete answers. For more reliable data, I recommend using Clay's built-in enrichments instead: - "Find Jobs" will accurately track open positions - "Funding" a waterfall that provides verified funding rounds - "Headcount by Country" shows employee locations - "Employee Growth" tracks team size changes https://downloads.intercomcdn.com/i/o/w28k1kwz/1377601153/86870a615712d5a17bd4eaccfc32/CleanShot+2025-02-12+at+_48ZqckNTAy%402x.png?expires=1739385900&signature=d6cba07d2c6fe56f82ba0014a9109e9f8af02f947dcff1ec3e332298515b6f30&req=dSMgEc9%2BnIBaWvMW1HO4zaYSMDF6PkiYdHeJmk6E2NZC6M0HYXGcB1BpqHpH%0A5Wmu%0A https://downloads.intercomcdn.com/i/o/w28k1kwz/1377600915/86c2a251abe21a4c1381c4568005/CleanShot+2025-02-12+at+_38nJy8DoQn%402x.png?expires=1739385900&signature=43ad93fdb87c51e659f7f1cb690f507742b65810adda6af239183293a8cc76df&req=dSMgEc9%2BnYheXPMW1HO4zSC46CSeuT7MlDK1rPCd2IHFsatRKifb8PakhNg%2F%0A%2Blsi%0A For your other questions: To stop automatic AI field running: Go to Auto-Update settings and uncheck it. This prevents unexpected credit usage. While Clay works great with emails and AI, using your own API key lets you save on credit usage better and you can then use your credits for data from enrichment like I discussed where they will be more precise and not open to hallucinations Check out clay.com/claybooks and clay.com/templates to explore more use cases. Let me know if you need help switching to these enrichments!
It really depends on how you approach this. It might costs more but using your API with AI you'll save on credits and from there you'll be able to reference exact data while creating more precise and trustworthy messages. Regarding the key not working, it’s often due to needing Tier 2 access with OpenAI’s API services. To check your current tier, usage, and limits, visit the OpenAI dashboard and look under: • Usage • Limits • Usage Tiers We’ve seen users successfully add a pre-paid tier amount to their OpenAI account and resolve the issue. However, since this is outside our control, we recommend troubleshooting directly with OpenAI’s team. Let me know if you need any further assistance! 😊
Also, why is there so much inconsistency when running Claygent? I run the same row twice and I get two different results. And it seems I get worse results if I run multiple rows at the same time. Have others noticed that?
Hey there, thanks for reaching out! ☺️ Jumping in for Bo. Feel free to run the Claude model and compare it to the other AI results. It's considered better and creating creative lines if that's a use case. Inconsistencies when running Claygent can occur due to the nature of AI processing. Each time AI generates a response, it "re-thinks" the answer from scratch, which can lead to variations in results even when running the same row multiple times. This is because AI processes information independently for each request, without remembering past responses unless specifically programmed to do so . Running multiple rows at the same time should not inherently affect the quality of results. However, optimizing the prompt given to Claygent can improve outcomes. You can use the "Help Me" feature in the Claygent menu to generate a prompt that follows best AI prompting guidelines . Additionally, choosing the right AI model for your task can impact results. Claygent Neon is designed for data extraction and formatting, while GPT-4 and Claude Opus offer enhanced reasoning capabilities. Experimenting with different models may help achieve more consistent results. If there's anything else I can assist you with, please let me know! 😊 Have a great Thursday, Daniel
Hi, I tried using Headcount by Country and I still don't see the right data. For example, row 15 says they have 1 person in the US but there are 3 in India and 2 in the US that fall under Engineering. https://www.linkedin.com/company/anoki-inc/people/
Same with row 14, it says 1 from France when there are 7
Hey Luka! I checked the LinkedIn data and saw the mismatch. The issue was that "engineering" wasn't capturing roles with "data" in the title. I added "data" as a keyword and now it's finding the correct countries Let me know if you need help adjusting any other things
.png){kind=logo}
