QUESTIONS

Questions we are asked.

Every answer names its source. Where it is uncomfortable, it is here anyway.

An answer without a source is a claim — that holds for our assistant and for this page alike.

This page carries the questions that actually reach us in conversations and through support — with an answer in a few sentences and the source it comes from. Where the answer is uncomfortable, it is here anyway.

DATA AND LOCATIONS

Where is the data, and who can reach it?

Where is our data processed?

In the cloud edition, inside the EU. On an Explicio Box everything stays in your own network — language model, image understanding, search and graph all run locally. Which location applies to your tenant is shown in the admin portal under privacy, and can be changed there.

Source: privacy notice, section on processing locations and residency

Is a model trained on our knowledge?

No. Your knowledge trains no model – neither one that other customers use nor one just for you. We select and measure the large language model. Only our small decision models (Explicio pair4) are trained by us, solely on data we generate ourselves. Not training on customer data is market standard, not a differentiator.

Source: AI policy section 6 · model register

Can your team see our content?

During an incident, yes — and we say so instead of claiming zero-knowledge. Access is tied to named people and logged; remote access to a box only reaches it after your own admin account has approved it.

Source: privacy notice, section remote maintenance

What happens if your own model infrastructure fails?

Then the platform declines the request with an error message: with European processing, nothing falls back to an external provider. If our own hardware is temporarily not enough, Explicio can add a rented server in a data centre in Germany on which Explicio runs the same model itself. On a box there is no fallback in any case.

Source: T&C 2.5 · privacy notice § 4

Who sees which content?

Each person sees only what they are authorised for — in search as in answers. Accounts of external staff, working students or temporary workers can be limited to their own knowledge. External agents have the same rights as a person, and the assistant never names passwords or other secrets.

Source: page for works councils and data protection

PRODUCT

What can Explicio do — and what can it not?

How is this different from searching our files?

Explicio asks back. When a contribution has a gap, the assistant asks a follow-up question in the same channel instead of filing something incomplete. Only then does the knowledge node exist — and every answer names the source it came from.

Source: product page, chapter on asking back

How do our existing documents get in?

Through bulk import: a ZIP or a multi-file upload, and on a box a folder on your network. A dry run measures volume and duration beforehand and states both before anything is ingested. For live repositories there are connectors to SharePoint, Google Drive and Confluence — read-only, with a fixed scope.

Source: product page · runbooks on bulk import and connectors

Does Explicio assess employees?

No, and that is a build decision rather than a setting: the assistant refuses assessments of people, there is no ranking of individuals anywhere, and per-member metrics are purpose-bound and carry no history. Using the platform that way breaches our acceptable use policy.

Source: acceptable use policy · AI system register (design laws 2 and 3)

What can Explicio explicitly not do?

It controls no machinery, releases no payments and triggers no HR measures. It does not replace professional judgement: answers are proposals with a source that a human evaluates. And it does not invent numbers — where one is missing, it says so.

Source: product page, chapter on limits · AI policy

How do we capture experience knowledge before someone retires?

As long as possible before the last working day and without dedicated appointments: the experienced colleague sends what she explains anyway as a voice message, photo or video via the messenger the team already uses. Explicio stores it with her name and the time, links it with manuals and maintenance reports and asks when something is missing. Whoever asks later gets the answer with exactly this source.

Source: guide “Before experience retires”

What happens when Explicio does not know something?

Then it says so instead of guessing: if nothing on a question is in the knowledge, Explicio answers that nothing is stored on it. If a contribution has a gap, Explicio asks — and remembers a refusal. Every answer names its source down to person and time, and if two statements contradict each other, the contradiction stays visible.

Source: facts page

BUYING AND OPERATING

What does it cost, and how do we get out again?

What are we billed for?

For the knowledge you bring in, not for seats. There is a subscription with an included allowance and a pay-as-you-go tariff without a base fee; a machine working through the interface pays the same units as a person.

Source: pricing page — every figure live from the price list

Can we get our data out again?

Yes. One admin account requests the full export, a second approves it; a JSON stream and a media archive are then produced with time-limited download links. The format is documented so the export stays readable without us.

Source: T&C 7.4 · export format documentation

Is there an availability commitment?

We currently commit to no fixed percentage. Instead we measure from two independent locations and publish the result together with its measurement basis — a number you can recompute is worth more than one that is promised.

Source: status page · T&C 2.4

What is the Explicio Box, and who owns it?

A device in your network that we operate as part of the Box Service: updates are signed and reversible, backups are encrypted with your own key, and remote support reaches the device only after you approve it. Ownership stays with us; the term is in the Box Service terms.

Source: box page · Box Service terms

WHERE THESE QUESTIONS COME FROMFrom conversations, support requests and the legal texts. The connection to our demand research is prepared but delivers nothing yet: of 991 measured search terms, none was phrased as a question on 17 September 2026. This page therefore grows with real questions, not invented ones.

YOUR QUESTION IS MISSING

Ask us.

We record the question — and when it comes a second time, it appears here.

Service statusTo the live status page →