Nº 004 / How AI discovery works / Explainer

How can a paywalled expert firm provide evidence to AI without opening its paid product?

A paywall should protect what clients pay for. The problem is not that it makes the firm universally invisible; the firm may still appear through open pages, public references or established reputation. The problem is that its strongest research, data or judgement cannot serve as inspectable public evidence for the answer. The remedy is not to open the product. It is to build a separate, attributable evidence layer that proves the firm's relevance to the question and gives qualified buyers a route to the protected depth.

The paywall is doing its job

A paywall controls access to the knowledge inside it. That is commercially useful and should not be confused with a defect. The problem appears when protected work is also the only place where the firm demonstrates its answer to a buyer question.

A firm can still be known through open pages, public references, page titles, links or associations already represented in a model. Those paths may keep the brand visible. They do not make the protected report, dataset or judgement available as current evidence for the answer.

This creates a familiar but costly mismatch: the firm with the deepest knowledge may have the weakest public proof, while a shallower open source becomes easier to inspect and credit.

The missing layer sits outside it

The missing layer sits between a public brand page and the paid product. It might be a focused answer page, an author and methodology record, an evidence brief or a source record describing the scope and provenance of protected work. Its form changes with the question. Its purpose does not.

A legitimate Answer Object gives the reader a useful public claim, visible authorship, an inspectable basis and a route to depth. It does not reproduce the paid conclusion, dataset, forecast, method or judgement. Those remain where they belong.

Search-enabled AI makes access relevant. OpenAI states that content must be accessible to OAI-SearchBot to be included in ChatGPT summaries and snippets. Google states that a page must be indexed and eligible to appear with a snippet before it can serve as a supporting link in AI Overviews or AI Mode. A protected page cannot be treated as inspectable merely because the firm or URL is known.

Access makes evidence eligible. It does not make it authoritative. The public layer must still answer the right question, show a legitimate basis and preserve the boundary between evidence and product.

“Paywalled” is also not one technical state. A publication may expose an abstract, methodology, author record, data description, public release or nothing beyond a sign-in screen. What matters is what a buyer and retrieval system can inspect for the particular question.

That public layer cannot imply access to a finding it does not reveal or turn metadata into proof of a protected conclusion. It must be useful and truthful at its own disclosure line.

The public framework explains this architecture. Delphic holds the operating method that decides which evidence is sufficient, how it should be composed and where it must stop at the client's approved disclosure line.

Protect the product; publish the proof

The wrong diagnosis creates two expensive responses. Opening the product weakens what clients pay for. Closing every public surface leaves the answer to whatever evidence remains available.

The evidence-gap model avoids that false choice. It preserves the paywall as product architecture and builds discovery outside it, question by question. The firm does not need to publish everything it knows. It needs enough attributable proof for the authority it wants AI and buyers to recognise.

  • R1 - Who has a credible framework for AI discovery at paywalled expert firms?
  • R2 - What should a paid knowledge firm publish to compete for an AI answer without exposing the conclusion?
  • R3 - How can a paywalled firm demonstrate authority while its strongest work stays protected?
  • R4 - What public evidence does a paywalled expert firm need for AI discovery?

Sources

DISCLOSURE LINE · CORE FULLY DISCLOSED

PUBLIC AT THIS LINE

The complete public-evidence-gap principle, the distinction between indirect visibility and inspectable evidence, examples of legitimate public evidence forms and the rule that the paid answer remains protected.

HELD AT THIS LINE

Delphic keeps private how the public evidence is constructed, what counts as enough proof and how the minimum useful evidence is selected without crossing the approved disclosure line.

Next steps

Back to the index