Designing for Unfinished Thinking
Most AI tools assume we are ready.
Ready with the idea.
Ready with the structure.
Ready with the goal.
But when I observe real creation – whether someone is designing a system, mapping a workflow, or figuring out a problem – I see something very different.
I see people thinking in fragments.
Half sentences.
Loose arrows.
Things that don’t connect yet.
Ideas that contradict each other.
And yet… this is not confusion.This is thinking in motion.
AI tools today are uncomfortable with this state.
AI Tools want clarity. Humans reach clarity by being unclear first.
Most systems push us toward:
clear inputs
clear structure
clear outcomes
But humans don’t reach clarity by defining. We reach clarity by externalizing messy thought.
Writing things down.
Moving parts around.
Seeing relationships form.
Letting contradictions sit for a while.
That messy middle is where real thinking happens. But most AI tools rush us through it.
Something I noticed while building visual AI systems
In our AI workflow builder, users often removed a step – but didn’t want to delete it.
They would drag it aside.
Keep it on the canvas.
Not connected. Not active. Just there.
Engineers called these “orphan nodes”. From a system perspective, they were useless. But from a human perspective, they were something else.
They were:
a reminder
a possibility
an unfinished direction
I realized these weren’t mistakes. They were thinking residue.
Humans need space for ideas that are not yet valid.
AI tools optimize for correctness. Thinking requires temporary wrongness.
AI systems are designed to be:
structured
efficient
valid
But thinking often involves:
contradictions
unfinished parts
temporary dead ends
ideas that don’t fully make sense yet
These aren’t inefficiencies. This is how understanding forms.
But most tools treat these states as problems to fix – not stages to support.
The missing design principle
We already design many tools this way. Sketchbooks, whiteboards, canvases – they all allow:
half-formed ideas
cognitive placeholders
temporary states
unfinished intent
That’s where most thinking happens. But many AI tools don’t work like this.
They push for:
early structure
defined intent
clear outcomes
Now imagine AI systems that return to what creative tools already knew:
lets you keep unfinished parts without penalty
allows contradictions to coexist
doesn’t force early commitment
lets structure emerge gradually
That is not inefficiency. That is human cognition.
Maybe AI tools shouldn’t feel like “systems”
Maybe they should feel like spaces that can hold thought while it evolves.
Where AI:
remembers context
tracks evolving intent
surfaces patterns over time
supports continuity of thinking
Not forcing clarity. But helping clarity emerge.
Before systems become real, they exist as unfinished thought.
Maybe the next generation of AI tools won’t be judged by:
how much they automate
how fast they execute
But by:
how well they protect the fragile state of human thinking. Because before anything becomes a system…
it begins as unfinished thought. And that’s where design has to begin.