I notice that YOUR PICKAXES is now an option in the Actions!!! This is really great!! And the ONLY function I was really missing from ChatGPT. Except this is MUCH BETTER! Where in ChatGPT, I would have to manually call in another Agent by typing @AgentNAME, In pickaxe I can just use the TRIGGER PROMPT to do it automatically! This makes it almost Semi-Agentic!!!
@admin_mike, I can’t say enough about this platform! Thank you!
8 Likes
This is indeed a HUGE improvement and opens a wide range of possibilities! It’s great to see Pickaxe gradually improve 
Thanks @nathaniel for coming through on this!
6 Likes
Cool. I’m curious how to use this and what possiblities it creates.
3 Likes
@dougfresh I love your shares… but can you explain a few use-cases here please. Greatly appreciated.
Also, if you are open to freelance work or support please let me know. I’m in Australia and growing list of solutions to build. I’m interested in growing my AI and Automation team capabilities via freelance or mentoring.
1 Like
@data_interactive as an example you can have a marketing manager agent which is the one organising the work and communicating with the user. Then you can have a marketing research assistant and a content generation assistant that can be called by the manager if and when required.
You can build end-to-end agentic workflows without requiring the user to jump from one chatbot to another.
1 Like
How do you actually do this? How do you trigger the linked pickaxe as an action? I’ve tried writing a prompt with multiple steps, but it does not continue to trigger my other pickaxes.
1 Like
Hey @timothy for the Trigger Prompt, this is a proven method:
“If the user asks a question that falls outside of your scope, such as topics related to [clearly define topics here] - trigger [enter name of target Pickaxe] chatbot for assistance.”
2 Likes
Ok. But what I’m wanting to do is chain a series of processes. For example, to have a pickaxe generate some text, then send it to another pickaxe for further processing, without needing the user to prompt it to proceed to the next step.
Is this possible? How to create a prompt to do this?
1 Like
Theoretically, this is possible by embedding your instructions into the trigger prompt. I haven’t had to do that for any project yet. Test it out and post an update for us with the results!
1 Like
My use case is to have a PA that discusses a subject matter (amongst other broader issues) and then call another PA dedicated to conducting a fact find/assessment with a series of questions… at the moment, it asks the first quesiton then returns to the PA that called my Q&A Assessment as an action and as a consequence, never actually concludes the assessment… Has anyone found a menaginful way to make this work?
2 Likes
Yes. As long as you have some sort of trigger the client should never realize they’ve switched.
Here is a quick video I created to show you how it works. Stacking Pickaxes for Meditation Creation 🪄 | Loom
Let me know if you have any questions. I am happy to help.
5 Likes
This is brilliant Jeanette - thanks for sharing!
2 Likes
Ab gave a nice example below, for me, I. have like a business advisor. Pickaxe that whenever. details about products are. asked. It calls in a product specialist pickaxe. The reason for this is it prevents you from loading up one pickaxe with a super long. system prompt and a lot of training documents.. I find that the pickaxe gets too confused when you give it too many instructions in the system prompt. So this function really allows you to loosen it up a little bit and then call in a specialist when necessary.
1 Like
That’s fantastic Jeanette. Can you then be passed back to the primary chat again to continue with the original chat flow? Or does the user have to continue (and end) with the 2nd chat (in this case the meditation creator)?
You can continue to switch the Pickaxe as long as your trigger is clear. So if you want to set the instructions to ask if the are ready to (fill in the blank) and they say yes, you can send them back to the first Pickaxe. The options are quite open.
Note that you can’t run workflows with sub-agents. You need to move the workflows logic up one level.
Let’s assume you have built Pickaxe B with a 2-step logic: do X, wait for the user’s answer, then do Y. (Example 1). The user interacts with the agent, and the agent knows exactly what to do.
You now build Pickaxe A and connect Pickaxe B as action (Example 2). Pickaxe A passes the user’s query to Pickaxe B that responds with X. Pickaxe A sends the response to the user but doesn’t know anything about Pickaxe B’s logic so when the user responds chances are Pickaxe B is not triggered again and you will not get Y.
To make sure Pickaxe B is triggered twice (once for step X and one for step Y) you need to tweak the trigger prompt when you connect Pickaxe B and maybe add some guidance inside Pickaxe A’s prompt
1 Like