Hey ActivePieces Team (e.g. Ash and team),
You’ll see us using ActivePieces Cloud for workflows involving OpenAI. As you will already know from the community and, possibly, from your own experience, some OpenAI tasks take ages to execute - especially for long texts.
Some of our workflows will inevitably include multiple OpenAI steps stacked together.
OpenAI doesn’t provide ASYNC processing options, so, unfortunately, the only way for AP Cloud to work with OpenAI is to sit patiently and wait for the HTTP response to come back with the output.
This brings me to the four requests, which, I believe, anyone who ever built a multi-request workflow with OpenAI, will support:
- We need more time for workflows. 10 min is not enough. Please consider extending to 40 min, in line with Make.
- We need error handling options for pieces that error out - specifically, an option to re-execute a failed piece (immediately or after a delay) with the same
- For flows that time out, we need to be able to see how far the flow was able to go before it timed out (whether 10 min or 40 min or anything else). This will allow us to, perhaps, create a hand-over webhook to another flow.
Thanks in advance for considering these requests. AP is a superb product, so all feedback and requests are offered with love
Community - please signal your support if these are relevant to you.