We probably haven’t posted about this yet, but we’re working on an exciting product for 2025.
An agent that builds your automations, and I don’t mean a copilot, but a full fledged agent that takes ownership in how it builds.
Almost like this:
If you have access to this, what would you prompt it to do? We’ll take your prompts and work hard to make the agent capable of handling them, so feel free to post as many prompts as possible.
WOW, this looks amazing! I’m really excited to start testing.
I think I have an interesting use case that might be a good fit for your solution. Here’s the first step in a multi-step process I’d like to automate.
Short Introduction
I run a recruitment business (a rather “lazy” one). My website hosts around 200 different job posts. Thanks to SEO automation, these posts attract about 10 responses daily.
Here’s how my process works currently:
I review each job post response to ensure the candidate’s resume and, if provided, their motivation letter meet the knock-out criteria.
If a candidate doesn’t meet all the criteria, I send them an email asking for more information about how they match the criteria I couldn’t confirm.
If the candidate does meet the criteria, I create a Google Doc with two tables:
The first table lists the knock-out criteria in the first row, with a second row beneath each criterion for explanations.
The second table contains the scoring requirements, following a similar structure.
The process continues beyond this, but this step alone seems complex enough to start with.
System Workflow Details
Here’s an outline of the automation I envision:
When a new opportunity is added to the Reacties op opdrachten sales pipeline in Flowlu, the workflow should:
Retrieve the details of the opportunity, the assigned contact, and the project associated with this opportunity.
Download the CV and (optionally) the motivational letter attached in Flowlu’s custom fields.
Upload these files to Straico.
Extract the job post description (associated with the project) and send this to Straico to verify if the candidate’s resume and motivational letter match the knock-out criteria.
If the candidate does not meet all the knock-out criteria:
Generate an email using Gmail, asking the candidate to provide details on how they believe they meet the criteria.
Include a step for human review to check the email before it is sent.
If the candidate does meet the criteria:
Create a Google Doc with:
A table listing the knock-out criteria (first row) with space for explanations underneath each item.
A second table for the scoring requirements in the same structure.
Save the document as a .DOCX file and attach it to an email for the candidate, asking them to provide explanations for the criteria.
Include a human review step before sending the email to ensure everything is accurate.
I hope this provides enough clarity for you to assess the use case. Let me know if you need further details!
I would love for the agent to be able to build custom HTTP request blocks from API documentation. For example, I might ask it to build a custom HTTP request that will create a new record in a Bika.ai database. I give the agent the prompt and tell it the documentation is at https://bika.ai/en/help/guide/developer/openapi
The agent then pulls the API documentation from that site but it also is smart enough to crawl / follow links as needed (or use existing blocks / knowledge) to build the custom http request block. In this case, that would mean referencing the APITable API documentation / blocks as well.
It might be a heavy lift, but it would make it immensely easier to integrate APIs that don’t currently have pieces in Active Pieces.
@Riley Thanks for your input! Definitely, will take that into consideration (not sure to which extent yet), but it’s definitely an important aspect. We have MUCH more similar functionality that will help with the same goal in the new version!