feature request: it would be great if you could use Generate using chat for the og prompt as well as the json structure. Often there are subtle updates that need to be made to the prompt after testing, and because the Generate flow is limited and non-deterministic, the corresponding output often changes the output entirely, when only a minor change is needed. It would be great to be able to shoot back and forth with a chatgpt bot. It's also annoying when you lose the variables in chatgpt because you cant copy-paste them (unless im missing an easy way to do that)
Thanks for the feedback! I understand you'd like a more conversational way to iterate on both prompts and JSON schemas without losing variables.
Before I connect you with someone who can review this feature request, have you tried using Sculptor? It's our built-in AI tool in the top-right corner that helps build tables, generate workflows, and set up automations. It might help with iterating on your prompts and structures more conversationally.
Let me know if Sculptor helps with your workflow, or if you'd still like to discuss this feature request further with our team!
If you still need help with your feature request or using Sculptor, I’m here to assist. Would you like to share more about what you’re trying to achieve or any specific challenges you’re facing?
hmmm dont know if I would want to use sculptor for this specifically, would be handy do just do it from prompts with existing tables
I get it - you want to iterate on prompts directly within your existing table workflow rather than switching to a separate tool.
For prompt optimization within tables, you can use the "Help me" button (metaprompter) in AI columns to refine your current prompts. This acts as a built-in prompt optimizer that helps improve your existing prompts for your specific use case. [1]
You can also structure your prompts with variables using curly brackets to reference other columns, making them more dynamic and reusable within your table. [2] For JSON schema work specifically, you can build detailed prompts with structured outputs and edge case handling directly in your AI enrichment columns. [3]
This keeps everything contained within your existing table structure without needing to jump between different interfaces.
Was that helpful?
If you still need additional help with refining your prompts or working within your existing table, I’m here to assist. Would you like to provide more details about what you’re trying to achieve or any specific issues you’re encountering?
still request the feature plz - chatgpt is the handiest format
Got it! I understand you'd like a ChatGPT-style conversational interface for iterating on prompts and JSON schemas directly within your existing Clay tables.
This would be a valuable addition to our current prompt optimization tools. While we have the metaprompter and "Help me" button for refining prompts, a full conversational interface like ChatGPT would definitely make the iteration process more intuitive. [1]
Your feedback about wanting to maintain variables and make subtle adjustments without losing context is really helpful. A human agent can review this feature request and share it with our product team for consideration.
Would you like to speak with a human agent to discuss this feature request in more detail?
If you still need help with your feature request or have more to share about your workflow, I’m here to assist. Would you like to provide more details or continue discussing this with me?
