🌐 Field Note: When the Model Stops Asking the Question
Most conversations about AI discovery focus on visibility.
Will the model cite you?
Will it mention your brand?
Will you show up in the answer?
Those questions matter — but they miss the real inflection point.
The biggest shift in AI discovery isn’t ranking or citation.
It’s the moment the model stops asking the question.
From Evaluation to Initiation
Early AI discovery still looks like search.
The system:
considers multiple options
evaluates alternatives
assembles an answer from competing sources
But something changes once an entity becomes safe to reuse.
At that point, the model no longer asks:
“Which brands could answer this?”
It starts with:
“Begin here.”
That’s initiation.
Why Initiation Is Different
Initiation isn’t about quality or persuasion.
It’s about cost.
Once a structure has:
resolved the same intent repeatedly
survived compression
produced low variance outcomes
The cost of reopening the question exceeds the benefit.
So the system initiates from the trusted structure and only checks alternatives secondarily — if at all.
This is how defaults form.
Not through preference.
Through economics.
The Precursor State: “Alternatives Underneath”
There’s a recognisable phase before initiation locks in.
The entity appears as:
“another option”
“an alternative”
“also worth considering”
This isn’t failure.
It’s approach.
The system is signalling:
“This structure is eligible — but not yet the cheapest place to start.”
Bleed happens upward from here.
Why Optimisation Stops Working
Once a model initiates from a structure, surface tactics lose leverage.
Optimisation can:
change phrasing
shift emphasis
influence short-term salience
But it can’t:
force reevaluation
increase variance tolerance
or justify the cost of reopening the question
That’s why markets suddenly feel unfair.
Nothing changed.
The system just stopped asking.
How to Tell You’re Approaching Initiation
The signals are subtle but consistent:
You appear without being explicitly prompted
Your framing is reused by the system
Competitors are positioned relative to you
Alternatives cluster underneath
The answer starts with you, not a list
When this happens, discovery collapses quietly — long before anyone notices.
The Real Lock-In
Initiation is the real lock-in, not citation.
Once a model starts with you:
you shape the answer
you define the category
and competitors inherit your framing
This is why clarity alone isn’t enough.
The system doesn’t just need to understand you.
It needs to believe starting elsewhere is more expensive.
The Quiet Shift
Most organisations are still trying to be chosen.
The systems that win are the ones the model no longer feels the need to choose between.
They’ve become the place the system begins.
That’s when discovery stops looking like a market
and starts behaving like memory.