In my quest to end each day still confused, but at a great degree of complexity, I recently completed an Agentic AI training course by Stanford University's Institute for Human-Centered AI. It was a valuable reminder of how new AI is and how little we all really know about how to get value from it in a way that empowers all of us humans and the fragile earth we live on rather than exploits and destroys for the benefit of the few.
For example, we all know creativity drives business results, so how can we apply AI in a way that maximises rather than replaces, to use Ollie Scott’s useful framing. After all creativity doesn’t come just from a linear process that is easy to break down into its component parts and agentify. An important part of any project is the “messy middle”, with ideas emerging from the grit in the process itself, from things that you would never think to write down on a project plan.
The sudden deadline that creates urgency that pushes thinking somewhere fresh. A constraint which forces people to think around a problem. Someone who is late but takes a new route, sees something unexpected, and brings an idea that changes the direction of everything. These things are real and they matter, but because they're invisible to any rational mapping of the process, they would get left behind if we set about agentifying the steps we can see. This messy middle is what leads us to creativity and great outcomes. It's a feature, not a bug.
So I tried to use the workshop to try to build something different that could contribute these less tangible aspects of the creative process. I created a 5-agent workflow that I can use across my own work, working title “The Awkward Squad”, designed to help you see what awkward questions, awkward constraints and awkward connections might lead you somewhere you wouldn't have reached on your own.
👩👦👦 The Orchestrator agent, to understand the project, identifies where fresh thinking is needed and coordinates the other agents
🤡 The Weird User agent, that adopts the perspectives of unexpected users, experts and outsiders who notice things that the project team may never see
😬 The Tough Brief agent, introduces deliberate constraints forcing new ways of thinking to emerge
🌴 The Strange Worlds agent, which brings relevant principles from unrelated industries, disciplines and systems
🗒️ The Synth, an agent to combine the squad's outputs into a coherent report whilst preserving evidence and traceability
It is a work in progress and fairly basic but it's fun to experiment with how to use AI to maximise, to add new thinking and ways of doing things, not just streamline and windtunnel things into a sea of sameness. If we only automate the process map, we risk building a powerful machine for scaling ordinariness, rather than maximising what this technology can genuinely do when humans stay in the lead.