Question 1

How is ExploreWorldAI classified under the EU AI Act?

Accepted Answer

The platform runs limited-risk systems under Regulation (EU) 2024/1689. The reason is simple: every result is decision support for a person, and no run produces a legal effect on its own. Matching, scores and prices are arithmetic with published weights, not a learned prediction. Our assessment of relevance: the transparency articles are the part that applies to the delivery. No use that the regulation lists as prohibited. No biometric identification, no emotion recognition, no social scoring. No profiling of individuals and no automatic decision about a person. Recruitment support is decision support for the employer, who remains responsible. What this means for you: You can take the platform into a purchase process with our reading written down, because the delivery is a document your own people act on.

Question 2

How do you tell users that they are dealing with a service?

Accepted Answer

You are always told when an answer comes from a service and not from a person. Written text is a draft until a person approves it, and every number can be expanded into the arithmetic behind it. The support answer presents itself as a service and hands over to a person on binding matters. Written text is marked as a draft, nothing is published on its own. Every number shows its calculation, line by line. A field that could not be read is marked missing, never filled in with a guess. Each result carries the source beside the field it came from. What this means for you: Nothing arrives without a trail, so you can show a customer or an auditor where a sentence or a figure came from.

Question 3

Where does the data come from?

Accepted Answer

The platform reads openly published pages, in the same way a search service reads them, plus what you type in yourself. Nothing is bought from a data broker and nothing is taken from behind a login. Openly published company pages and public search summaries. Public registers and official sources, listed per page in the source register. Information the customer supplies about their own business. No purchased personal records and no data broker. A fetched page is deleted within fifteen minutes; the record keeps counters, not names. What this means for you: You get a delivery your legal team can approve without a project, because the origin of each field is written next to it.

Question 4

Which parts use AI and which parts do not?

Accepted Answer

Two different kinds of work run side by side. Reading, matching, scoring and pricing are fixed rules in our own code with published weights. Generated text and document reading use a general-purpose provider, and that part never sets a score or a price. Scores, matches and prices: fixed rules, published weights, the same input gives the same result. Written suggestions and support answers: a general-purpose provider, used to write, never to decide. Document reading returns fields that a person confirms. Provider, version, purpose and cost are recorded for every call. Customer data does not train any external system. What this means for you: You can predict what the platform will do, and repeat a result later with the same input.

Question 5

How do you keep the output from being wrong?

Accepted Answer

The design starts from a simple rule: a wrong answer costs more than no answer. When the reading is not certain enough, the platform abstains and says so instead of producing a number. Closed word lists and fixed rules, so a result cannot drift with a rewrite. A final check runs on every answer before it is delivered. Spending ceilings and rate limits per customer stop runaway use. Known limitations are published beside the result, not hidden. Every applied change is written to a record that can be reviewed afterwards. What this means for you: The failure mode is a blank field with a reason, which is something your team can work with.

Question 6

Who makes the decision?

Accepted Answer

A person does, in every case that matters. The platform prepares the ground, ranks the alternatives and shows the arithmetic; the decision to hire, to lend, to invest or to publish stays with a named human. No hire, no credit and no capital decision is made by the platform. Written text needs a human approval before it is published. Binding questions are handed over to a person, never answered by the service. A result can be challenged and reviewed by a person on request. The customer can switch off any module they do not want. What this means for you: Responsibility stays where your governance already places it, so nothing new has to be delegated to a supplier.

Question 7

How do I report a problem with a result?

Accepted Answer

Write to hello@exploreworldai.com and describe the address, the result and what looks wrong. We confirm receipt within two working days, investigate, and answer with what we found and what we changed. An incident is a wrong result, an unavailable service, or a reading that should not have happened. Receipt confirmed within two working days, written answer within thirty days. Every case is written to an internal record with time, cause and action. A page read in error is deleted immediately on request. A personal data matter can also be raised through the request form. What this means for you: There is one address, one clock and one written answer, so a complaint does not disappear into a support queue.

Question 8

Where is the platform operated?

Accepted Answer

The platform is built and operated in the EU by a Swedish company. Processing, storage and support all stay inside the union, and the delivery does not depend on a source outside it. Operated by Valkiv Ventures AB, Kungsgatan 8, 111 43 Stockholm, reg. no. 556995-1311, VAT SE556995131101. D-U-N-S 352720102, verifiable in the public register. Hosting and processing inside the EU. GDPR and the data processing agreement apply to every customer. No transfer of customer material to a third country as part of the delivery. What this means for you: The compliance conversation stays inside one legal frame, which shortens procurement.

AI Act statement for ExploreWorldAI

01 System classification

How is ExploreWorldAI classified under the EU AI Act?

02 Transparency

How do you tell users that they are dealing with a service?

03 Data sources

Where does the data come from?

04 System description

Which parts use AI and which parts do not?

05 Risk handling

How do you keep the output from being wrong?

06 Human oversight

Who makes the decision?

07 Incident reporting

How do I report a problem with a result?

08 EU compatibility

Where is the platform operated?

AI Act statement for ExploreWorldAI

Contact and responsibility

Binding texts

AI Act statement for ExploreWorldAI	What this means for you
01 System classification	You can take the platform into a purchase process with our reading written down, because the delivery is a document your own people act on.
02 Transparency	Nothing arrives without a trail, so you can show a customer or an auditor where a sentence or a figure came from.
03 Data sources	You get a delivery your legal team can approve without a project, because the origin of each field is written next to it.
04 System description	You can predict what the platform will do, and repeat a result later with the same input.
05 Risk handling	The failure mode is a blank field with a reason, which is something your team can work with.
06 Human oversight	Responsibility stays where your governance already places it, so nothing new has to be delegated to a supplier.
07 Incident reporting	There is one address, one clock and one written answer, so a complaint does not disappear into a support queue.
08 EU compatibility	The compliance conversation stays inside one legal frame, which shortens procurement.