I’ve been trying to migrate my Zapier AI Agents from another platform to Zapier, and while I like the overall direction, the migration has been quite challenging due to a few limitations I’ve run into.
I wanted to share my experience and see if others are facing the same issues:
1. AI by Zapier payload limit
According to Zapier Support, the current payload limit for AI by Zapier is around 6 MB, while the documentation for v17.2.0 indicates a limit of around 35 MB.
This is becoming a major blocker for some of my workflows. When working with transcripts, images, and other larger inputs, having a much lower payload limit significantly limits what can be done with AI automations.
2. No way to dynamically pass multiple images
Another limitation I’ve noticed is with image inputs.
For example, if the number of images that need to be analyzed varies from one run to another, there doesn’t seem to be a straightforward way to dynamically pass all of those images into a single AI by Zapier step.
The alternatives seem to be either:
-
Attach one image per run, or
-
Use a loop to process each image individually, which increases task usage and cost.
It would be really useful to have the ability to dynamically pass an array/collection of images into a single AI step.
3. Separate System Prompt and User Prompt
I also think the prompt section should be divided into two separate fields:
-
System Prompt — instructions, rules, context, constraints, etc.
-
User Prompt — the dynamic data/input being processed.
This would make it much easier to structure prompts properly and follow best practices, especially for more complex AI workflows.
Overall, I’m trying to move more of my AI automation into Zapier, but these limitations are making the migration difficult.
Is anyone else running into these issues, especially the payload limit and dynamically passing multiple images?
