Blessing and a curse.

Just finished listening to Lauren Tan's interview about how SpaceX uses AI agent workflows to improve productivity. It seems AI, and especially AI agents, are on everyone's mind right now. Every company is racing to outpace its competitors, creating what feels like a global-scale case of "shiny object syndrome."

To be fair, AI is absolutely transforming the world and evolving at an incredible pace. Personally, I believe its significance is parallel to the rise of the Internet in 90s. However, I also think much of what we're doing today with AI agents, orchestration frameworks, guardrails, verification, testing, and trust-building is ultimately TRANSIENT.

AI will eventually become self-sufficient to handle tasks and automate large portions of our daily lives with little to no human intervention. Humans are increasingly becoming the bottleneck in many workflows. AI agents can already out perform tasks 100x faster and more consistently than us, and that capability will only improve over time. It's both a blessing and a curse.

That leads me to wonder whether many of today's best practices around agent workflows are simply temporary scaffolding. Verification, testing, orchestration, and trust mechanisms are critical today because AI is still maturing. But if AI reaches a point where it can reliably manage these processes itself, much of the infrastructure we're building around it may eventually become unnecessary.

What I find most interesting is the democratizing effect of AI. Eventually, everyone may have access to the same powerful capabilities at their fingertips, much like how Google Search gave nearly everyone access to information and learning opportunities. In that sense, AI could flatten the playing field once again.

If that's true, it raises an important question: how much should organizations invest today in AI agents and related "shiny objects" if many of today's architectures and workflows could be temporary? The challenge is distinguishing between what will become foundational and what is merely a bridge to a future where AI can do far more on its own.