It is the second phase, and it needs access, because this is the point at which inference becomes validation.
Inside Out is not a different product. It is the same engine interrogating a far richer data set. This is why Inside Out reliably surfaces black swans the public record could only hint at. You test the outside read against your own data, your context, and the things that never reach the public record, and the engine reads that larger set exactly as it read the first, across the same lenses and the same repeated passes.
What comes back is a validated picture: what holds under scrutiny, where the real constraint sits rather than where it appeared to, and what to change first.
The value here comes from corroboration, from findings interrogated and pressure-tested with the people who run the business, until what survives is validated rather than merely inferred. It is the difference between a credible hypothesis and a decision you can act on with confidence.
The output is a validation, graded and prioritised, that tells you which findings to act on now and which to hold, where the binding constraint actually is, and the change that will move the business first. As at Outside In, you keep the interactive report and the agent behind it, now holding the fuller inside picture, so it goes on answering long after the work is done.
Inside Out can run entirely within your own environment, so nothing sensitive leaves your control.
Or it can run through a secure, access-controlled data environment we provide, in the same way investors and lawyers exchange sensitive material during a transaction, where information is held under strict access controls, used only for the purpose agreed, and returned or destroyed when the work is done.
Either way, the choice is yours, and it is made before any inside data changes hands. See more at Security & Trust.