Creating a Waterfall with Claygent and GPT Text Generation Templates

Richard B.

I want to create a waterfall out of my templates (claygent + GPT text generation) - is that doable? Thx!

15 comments

Channeled
APP
·
·
You can build a custom waterfall combining Claygent and GPT text generation by following these steps: 1. Click "Add column" and select "Waterfall" to open the custom waterfall panel. 2. In the waterfall panel, you can configure your sequence of data providers, including Claygent and GPT text generation. The waterfall will run these providers in the order you set. 3. For each action in the sequence, specify: • The required inputs you want to search by • The waterfall output • Your endpoint mappings You can use Claygent as a catch-all failsafe in your custom waterfall sequence. Please note: Currently only a subset of integrations and actions can be added as custom waterfall stages, but full support for all providers and actions is coming soon.
Sources: Building a data waterfall | Documentation | Clay University Custom waterfalls Custom waterfalls
Channeled
APP
·
·
Was that helpful?
Channeled
APP
·
·
Hi Richard B.! This thread was recently closed by our Support team. If you have a moment, please share your feedback:
Channeled
APP
·
·
Thank you so much for sharing your feedback Richard B.!
Channeled
APP
·
·
Thanks! We've reopened this thread. You can continue to add more detail directly in this thread.
Channeled
APP
·
·
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!
Richard B.
·
·
I could not find a way to rename the sequence steps, what is the trick?
Bo (.
·
·
Hey Richard! You're right - the sequence steps can't be renamed right now. I'm curious what name you were thinking of using? For the waterfall setup with Claygent and GPT, can you confirm everything is working? If not, could you share the table URL so I can take a look? Let me know and I'll help set this up for you.
(52 kB)
Richard B.
·
·
Hi Bo, thanks for your quick reply! You clay support guys rock! So, I want to create a sequence of steps that you would call a news recency waterfall: 1) check pers. Linkedin post, → if not found → 2) check company Linkedin post → if not found → check company news on google → if not found → etc. so, the question is the waterfall the right approach or is an auto update of some columns better? Thx!
Bo (.
·
·
Hey Thank you Richard!!!! I'd recommend using standalone enrichment instead of a waterfall and use a Conditional Formula between them. I'd also recommend you to group related checks together (one for people, one for companies) which saves credits too. The AI models should be able to handle that kind of search. P.S Make sure you specify your outputs in the prompt, then use output > JSON > Generate from prompt to structure your data properly. Let me know if you need help setting this up!
Richard B.
·
·
great!
1.
how to group related checks together?
2.
I don't have to option to use output > JSON (see screenshot)
Bo (.
·
·
Instead of using templates, you'll want to create a new AI action and combine both prompts into one. That can be done by copying both and asking the metaprompter to re-write it or simply doing this manually by editing it. I'd update for something like "Find X, if no Y is found for this, then look for Z." You'll also be able to structure the output using JSON as well as Let me know if you need help
Richard B.
·
·
sounds complicated but will give it a try - for sure the right thing to do on a Saturday night! thx!
Bo (.
·
·
I'd recommend to check out Clay University (clay.com/university) - especially the 101 and outbound sections. They've got some great guides to help you get started and very fun to do on a Saturday night!! :)
Channeled
APP
·
·
Hi Richard B.! This thread was recently closed by our Support team. If you have a moment, please share your feedback: