Measure · Foresight · Experimental

Foresight

What AI will say — if you move.

Foresight is the future-tense view of the same engine behind the Index and the Audit. Measurement tells you where AI puts you now. Clio tells you where it put you before. Foresight takes on the question every leader actually wants answered: if we did this, how would AI’s reading of us move?

Register interest → Why it’s possible ↓
Why it is possible — and why it is experimental

A deliberate action is not a guess

Because Pitot models perception as a function of the evidence AI is standing on, a deliberate action is a change to that evidence — whose effect can be modelled, and in time tested against what actually happens.

A mechanism is not yet a proof.

The mechanism
Foresight is built on measured history, validated against real outcomes, and deliberately unhurried. We will not put a number on a forecast we have not earned the right to make. We would rather show it to you late and right than early and wrong.
Why it matters
The organisations that learn to read — and move — how machines perceive them, before their competitors have thought to measure it, will hold an advantage that compounds quietly and is hard to reverse.
How a projection is built

Change the evidence, then read the instrument.

A move is not modelled as an intention. It is modelled as the trace it leaves.

The question is never "will this work". It is "what would the machine be standing on afterwards".

The sequence
A proposed move is translated into the evidence it would create, that evidence is introduced into the reading environment, and the same standardised instrument is run against it.
  • The move is specified as observable output — what would be published, where, by whom, and at what volume
  • A counterfactual evidence set is assembled to match that specification
  • ARES-C is run against both the current and the counterfactual environment
  • The difference between the two readings is the projection, always with a band
What it can weigh
Moves that leave a machine-readable trace: a repositioning, a change in how a category claim is stated, a correction campaign against a known misattribution, a partnership, a sustained editorial programme, an entry into a segment you are not currently summoned for.
What it cannot
Anything whose trace is unpredictable in shape — a crisis, a competitor’s unannounced move, a shift in how the models themselves are built. Foresight projects the effect of your evidence, not the behaviour of the world around it.
The evidence bar

What must be true before we publish a number.

  • Enough measured history. A projection is only as good as the series behind it. Foresight cannot be run credibly until the Index and Clio have laid down sufficient record for the sector in question.
  • Closed-loop validation. Projections are compared with the measured results. The hit rate is tracked, and it will be published — including where we were wrong.
  • Bands, never points. Any figure Foresight returns will carry an interval. A forecast expressed as a single number is a claim about certainty that no honest model of this can support.

Until those four conditions hold, Foresight stays in development and is described as experimental — because that is what it is. We would rather be the company that shipped it late than the company whose forecasts quietly stopped being mentioned.

Instrument ARES-C · Foresight configuration · Experimental
Experimental · in development

Late and right, not early and wrong.

Foresight is how we intend to put a figure on the move before you make it. Follow its progress, or tell us you would like to be in the first cohort.

Register interest → Back to Measure →