Hi everyone, I am new to Pickaxe. I have setup a Pickaxe tied to Claude (mini structured coaching session with framework, knowledge files, etc) and I have it embeded and everything is working, but i have a few questions
-
I need to find a way to limit what the user asks outside the prompt instructions. For example, if someone goes through this, they can continue to ask questions and ring up the charges. How can I prevent someone to continue after we are done with my prompt.
-
How can I ask and get the users name and email? My goal is to use this as a lead magnet, and it they want more detail outside the guided prompt, to email them.
If you change it to form only they can only respond to your questions. Then you can add a name/email question as well.
I use it exactly the same way - you need to create hard and fast rules to input into the model memory that makes your guardrails really clear. Make sure you test it repeatedly even after you let it go live (you can just pull in the conversations from the users to see where it’s veering off course). For getting their name and email, I say “in order to generate your decoder results, I’ll need your name and best email.” You’ll need to make it really clear in your model instructions that that is the trigger to generate some sort of response (for my users it’s their Decoder Results). I also set up webhooks so that they get an automatic email asking if they’d like to talk about anything the decoder revealed with an actual human being with a link to my calendar. Hope that helps!