This post is from a suggested group
I’ve been trying to get a clearer picture of how AI is moving beyond just text generation into actually performing structured actions inside software systems. Recently I read this breakdown about Large Action Models and how they work in real applications: https://www.trinetix.com/insights/what-are-large-action-models-and-how-do-they-work . What caught my attention is the idea that models are no longer just “answering” but actually executing multi-step workflows across tools and APIs. I’ve seen basic versions of this in automation scripts, but not at a truly autonomous level. Do you think we’re actually close to reliable action-based AI in production systems, or is it still mostly experimental?
I think we’re in a transition phase right now. Some parts of Large Action Models are already practical, especially when the environment is controlled—like internal tools, structured APIs, or well-defined workflows. I’ve worked on automation systems where AI decides the next step in a process, but it’s always bounded by strict rules and validation layers. Without that, things can go wrong quickly. So in my opinion, we’re not at full autonomy yet, but we are at a point where “semi-autonomous” systems are becoming genuinely useful in production if they’re carefully constrained and monitored.