AI Presence Map · Public research prototype

Explore an AI presence map, cell by cell.

This prototype shows how a defined panel of fictional buyer questions could be organized by stage and constraint. It does not call an AI service or report on a real company.

Fictional specimen ExampleCo Payroll · example.com

United States · English · constructed dataset

12 situations 2 illustrated AI surfaces 3 constructed runs per cell

Public prototype: the map interactions work. ExampleCo Payroll, the answer text, counts, source labels, and truth records are constructed. No real AI collection took place.

Constructed presence7 / 12

Fictional situations with at least one mention

Present in all runs1

Mentioned in 3 of 3 constructed runs

Variable presence6

Mentioned in 1 or 2 of 3 constructed runs

Potential truth issues1

Illustrative claims marked for human review

Buyer situation map

Constructed mentions by stage and constraint

2 or 3 of 31 of 30 of 3Truth conflict

Counts belong only to this fictional panel. They are not live observations or estimates of buyer prompts or market share.

Repeated presence2 of 3 runs

COMPARE · EU PAYROLL

Which payroll platforms are suitable for distributed European teams?

Constructed answer state

ExampleCo Payroll entered the comparison set in two runs. The answer preferred AtlasPay when tax support across multiple countries was emphasized.

Named competitors

AtlasPayNorthstar HRPersonio

Illustrative surfaced-source labels

G2 comparison page
Company product page
Remote work guide

Illustrative covered claim checkSupported

The constructed description matches the fictional approved product record.

Open illustrative answer excerpt

“ExampleCo Payroll and AtlasPay both support distributed payroll operations. AtlasPay provides broader country coverage, while ExampleCo Payroll is commonly evaluated by teams focused on the EU.”

Constructed excerpt · no AI collection occurred · example configuration: United States · English

Illustrative Intervention Card

Connect one constructed cell to a bounded decision.

The map alone does not recommend action. A reviewed card records the requirement, proposed change controlled by the client, uncertainty, owner, cost, completion condition, and matched remeasurement.

Constructed issue: A constructed answer claims coverage in all 27 EU countries; the fictional approved record lists 12.
Frozen buyer requirement: Accurate direct coverage information for an EU payroll decision across five countries.
Proposed change controlled by the client: Clarify the country coverage page and dated comparison material.
Hypothesis and alternative: Clearer source material may reduce the conflict. The answer may instead rely on another source or remain unchanged.
Owner · cost · completion: Product Marketing · fictional $5,000 estimate · approved page published and archived.
Matched remeasurement: Repeat the same cell after 30 days; compare claim status and count without assigning causality.

Illustrative dispositionMonitor

Do not fund a broader program from this constructed observation.

Research questions

What this prototype is designed to test.

These are product hypotheses. No live scan, saved workspace, subscription, agency plan, checkout, or API exists.

Hypothesis 01

Can a bounded map make the evidence legible?

Prototype

Stages and constraints remain visible
AI surfaces stay separated
Counts retain their denominator
Each cell opens to an illustrative excerpt
Absence is recorded without a fit conclusion

Comprehension test only

Not offered

Hypothesis 02

Would a saved panel improve a real decision?

Unvalidated

A panel and truth record approved by the customer
Matched collections under declared conditions
Complete evidence records for every run
Gained, lost, and materially changed states
A decision owner for every intervention

Requires paid evidence before productization

Hypothesis 03

Would repeated use justify software?

Gated

Retention has not been tested
Willingness to pay has not been tested
No live collection system exists
No agency or API product exists
Expansion would follow observed customer pull

Research direction · no current build commitment

Product research led by the founder

Built and reviewed by Nicolò

Nicolò Brignoni

Founder · Deep Ocean

I am exploring whether an AI presence map can make a defined set of buyer questions easier to inspect by stage, material requirement, AI surface, and dated answer record.

The screen above is a product hypothesis. No live scan, saved workspace, or subscription exists. The immediate test is whether teams understand the map and want to inspect a real version against a pending decision.

Visible panelStages and constraints stay explicit. Evidence requirementA live version would need a complete record for every run. Bounded claimsConstructed counts remain tied to this specimen.

Follow the product research on LinkedIn →

Measurement requirements

A live version would require a record for every run.

This interface is a fictional specimen. The requirements below describe what real collection would need to retain.

Defined panel

Stages and constraints stay visible. Any small live panel would remain labelled exploratory.

Separate surfaces

ChatGPT, Gemini, Perplexity, and answers generated through an API are never merged into one score.

Records for every run

Each count would need the complete answer, timestamp, visible model and surface, locale and account state, collection settings, and surfaced links for every run.

Bounded interpretation

A future collection would not reproduce buyer memory, establish causality, or predict traffic and revenue.