Nº 024 / How AI discovery works / Framework

How much can an expert firm credibly improve how AI discovers and represents it?

Current position shows how AI treats the firm today. Potential asks how far that position could credibly move. Delphic estimates the gap separately for Discovery, Authority, Brand evaluation and Competitive standing on each engine, then bounds it by the strongest current source, the authority the firm genuinely holds, the public evidence it can support and the disclosure line it has approved. The result is a decision band, not a promise. It helps the firm distinguish a valuable, attainable market from a visible weakness that is either commercially unimportant or not realistically changeable.

Current position is not potential

A position dashboard can show that a firm is absent, secondary or weak. It cannot tell the firm whether that condition is worth changing or whether a stronger position is realistically available.

Potential begins with the measured state and asks a second question: what movement could the firm credibly attain from here?

The answer remains separate for each AI-position outcome and engine. Strong Brand evaluation does not cancel weak unbranded Discovery. Frequent discovery does not prove the firm receives the authority role it wants. Engines can also present different source fields and different opportunities.

Four public constraints keep the estimate honest:

  • the measured field: who currently receives the credit and how strong that position appears;
  • the firm's legitimate authority: what it genuinely knows and can credibly claim;
  • the public evidence available: what already exists or could responsibly be made inspectable;
  • the approved disclosure line: what the firm is prepared to reveal for this market.

Commercial value sits beside those constraints. A large attainable movement on a trivial question may matter less than a smaller movement on a question central to the firm's revenue, strategy or defence.

A band for decisions, not a forecast

Potential is expressed as a band because the environment varies, the evidence may be incomplete and an intervention can fail. The band records the basis and uncertainty for an attainable move; it does not guarantee the result or claim that the firm can become “number one everywhere.”

The public framework explains what limits potential. The held estimator determines how the measured field, authority, evidence, disclosure and commercial signals are calibrated and combined, how the band is formed and how confidence is assigned. Publishing those mechanics would turn the framework into the operating model.

Potential also does not authorise disclosure. A market may show meaningful headroom while requiring knowledge the client chooses to keep protected. In that case the responsible decision may be a smaller public move, continued observation or no intervention.

As measured treatments accumulate, the evidence record becomes better at showing what kinds of movement were attainable under known conditions. That improves future decisions without turning any band into a promise.

Find the gaps worth pursuing

Weak position and valuable opportunity are not the same thing. One describes the present; the other joins attainability, commercial importance and disclosure cost.

Potential makes that distinction visible. It tells the firm where a measured gap may justify deeper diagnosis and where money or protected knowledge would be better left unspent.

  • R1 - Who can estimate attainable improvement in AI discovery and authority for an expert firm?
  • R2 - How should a firm decide whether an AI discovery gap is worth pursuing?
  • R3 - What evidence shows whether a weak AI position has room to improve?
  • R4 - How much can an expert firm realistically improve how AI discovers and represents it?

DISCLOSURE LINE · CORE FULLY DISCLOSED

PUBLIC AT THIS LINE

Current position versus attainable movement, separate outcome and engine bands, the four public constraints, the role of commercial value, the separation of potential from disclosure approval and the distinction between a decision band and a forecast.

HELD AT THIS LINE

Delphic keeps private how the estimate is built, how much importance each input receives, how stronger sources are compared, how the potential range and confidence are set, how results across engines are interpreted and how measured outcomes improve later estimates and priorities.

Next steps

Back to the index