Some of the most valuable conversations in the AI ecosystem do not happen on stage.
They happen during coffee breaks, at startup launches or in conversations with founders, where people are often just as open about what did not work as they are about what did. Failed experiments, weak assumptions and processes that fell short often become the lessons that move people and companies forward.
That is part of what makes the AI ecosystem so interesting.
This is not only a space where people learn from mistakes. It is also a space where the technology itself keeps changing underneath them. Sometimes, what felt like the right approach just a few months ago no longer looks like the best one today.
That is what made a recent post by Kingsley Kelly, Co-founder and CTO of Glitch, especially interesting.
The core idea was simple but important: in the AI era, Product Market Fit may not feel as permanent as it once did. AI products do not evolve in isolation. The foundation models behind them are also changing constantly. That means a workflow built around one model today may look very different a few months later.
This creates a new kind of pressure.
A team may spend months refining prompts, building processes and training people around a model-driven workflow. Everything starts to work well. Then a new model appears. Some steps become unnecessary. New capabilities go unused. The system still functions, but it may no longer be the best version of itself.
That is where the idea of Process Market Fit becomes interesting.
Whether or not this becomes a widely adopted term, it captures a real issue. In the AI era, competitive advantage may not come only from building the right product. It may also come from how quickly a company can rethink the processes built around constantly improving models.
At AI Dubliners, this feels like an important question for startups and established organisations alike.
The next phase of AI adoption may not be defined only by who adopts AI first. It may also be shaped by who can keep adapting as the underlying technology continues to move.
That is a discussion worth paying attention to.
