Curious how you approach cases where an AI step decides to trigger an action, like calling an API, sending data onward, or kicking off another flow

[mdashrraf]

Hey, I came across your Node-RED OpenAI nodes while looking at agent/action workflows. Curious how you approach cases where an AI step decides to trigger an action, like calling an API, sending data onward, or kicking off another flow. Do you usually recommend human review points, or is it more about how the Node-RED flow is wired?

[allanbunch]

I approach this case-by-case, depending on the application. Though, I almost always start with human-in-the-loop so that I can see exactly wat's happening at those touchpoints before I bring AI into the picture.