Hello,
can I plug the result of one pickaxe into another?
that is, can I use both the inputs and the responses of a pickaxe processing within a second pickaxe that I have inserted into the study?
Why maybe in the first one I define an element that must be the input or part of the input of the second pickaxe?
Thank you.
2 Likes
To be able to do that would be great.
I’d take a look at this video we just posted on the subject.
What you’re describing could also be covered by our User Memory feature, but that’s Pro only.
Oh wow! This is so awesome. I wonder why this isn’t an announcement?
This is one of the best and most-demanded features. We have been extensively testing Pickaxe and dify.ai. For some of our chained use cases, we used Pickaxe calling a dify.ai workflow by creating a custom dify action. However, the entire workflow was slow and unpredictable as sometimes it would run and other times it would fail (often at Dify’s end).
Chained Pickaxes and actions solve that. I am going to test this extensively and if chaining works well, we’ll move to the pro tier of Pickaxe. Kudos to the whole Pickaxe team for listening to the user feedback and continuously iterating on features! With this feature set and pricing model, you are going to slay it!
I’ll request the pickaxe team to provide a provision for a trial of the pro plan (even a 14 day trial would do). Trials let us create use cases and show them to the leadership and influence buying decisions.
2 Likes
I did some hands on with prompt chaining. It works very well! However, there is a limitation that right now you can chain only one pickaxe (which in turn may be chained to another) but you can’t have multiple chained pickaxes to pick and choose what to run.
For example, in some cases, I want to chain 2 pickaxes (say Pickaxe B or pickaxe C) and let the user choose which chain to run. For instance, if Pickaxe A creates an Article, Pickaxe B can create a summary and Pickaxe C can create a tweet. Sometimes, I may require both a summary and a tweet (A->B->C) and at times only one of them (A->B or A->C). If this could be done, it would be very helpful!
1 Like
I too would be interested in a workaround for this. Currently writing a bot that is more of a choose your own adventure, and the Pickaxe chaining depending on user input would be spectacular. 
1 Like
I would also be interested, I have not joined the meeting because of the language problem. sorry.
We introduced some basic prompt chaining via the Actions. It may not satisfy your full appetite for chaining, but I’d check it out.
Seems simple enough, but I can only connect one such action. What I’m trying to achieve is this:
Pickaxe_Main offers the user 3 options (A,B,C)
If user selects A, it branches to PickaxeA which uses chatGPT 4o mini
If user selects B it branches to PickaxeB which uses Claude 3.5
If user selects C it branches to PickaxeC using Claude and some knowledge
If this isn’t possible (yet), perhaps there’s a way the studio can be used to steer the user?
1 Like
Importantly the prompt chaining we have (done via Actions) does not redirect the user to an entirely new Pickaxe. Rather, it makes a call to a different Pickaxe and returns the results.
What may be much simpler is to create these 3 Pickaxes (a, b, & c), then give Pickaxe_Main the link to each one with an explanation. Then instruct it to give out the link to the appropriate Pickaxe.