The Agentic Stack, So Far?

This series has quietly become a reference point for understanding the evolution of agentic systems. When we wrote the first piece last September, we didn’t know just how quickly the field would shift – or how much ground we’d end up covering. But week by week, as the agentic stack took shape, we kept pace. We tracked the papers, the protocols, the tools – and the builders working through ambiguity in real time. What emerged is not a commentary; it’s a living record of how the field is learning to build with agency. Knowledge systematized.

We followed agentic workflows from open-endedness (one of my favorites) and creative emergence to real-world infrastructure. We asked what agents are and what they remember, how they plan, how they act, reflect, cooperate, and learn. We traced the rise of MCP and A2A. We looked closely at co-agency, HITL, reflection, reward functions – and the shifting role of humans in the loop. That’s a trove of knowledge that will help you build.

It’s strange, of course, to call this a wrap, because agentic workflows are actively unfolding in front of us. But we’re placing a dot here. It’s not a closing chapter – we’ll be back to covering agentic systems very soon. But now it’s time for a very different topic – and you’re going to love it. Stay tuned!

Meanwhile, to refresh and revisit the journey, here’s the full lineup of episodes that brought the agentic workflow series to life →