The machine stays available.
You keep the key.

One shared machine, simple uses

The machine hosts the functions chosen for your business: searching authorized documents, preparing a record from a case or running a validated task. Users reach it from their usual workstation, through an interface suited to their role. Using the service does not give access to server administration or to all of the company’s cases.

The network, the accounts and the availability hours are defined according to your business. Access can stay within your premises or be opened remotely to authorized people, through a protected connection. Any processing entrusted to a cloud service, if there is one, is identified together with the data concerned. “Local” does not automatically mean “no data leaves”.

What this could be used for in your business

Find a procedure from the shop floor

A maintenance team consults, from a workstation on the shop floor, the manuals and procedures it is entitled to see. The AI can offer an answer with its sources. A missing document or an uncertain answer stays visible; the system must not invent the missing instruction. The team keeps the approval of any work on the physical machine.

Prepare a case without moving files everywhere

An SME wants to reconcile a purchase order, a customer record and a tracking sheet. The service prepares a draft within a defined data scope; the person in charge checks the information before the export or sending. Other users do not gain access to the customer file as a result.

Resolve an incident with a specialist

A search stops responding. You open a support request and describe the symptom. If a remote intervention is appropriate, an authorized session makes it possible to examine the service concerned. A video call can accompany the diagnosis: you see the steps, ask your questions and follow the useful instructions.

These situations are scoping examples, not references from client deployments or features guaranteed for every installation.

The support access belongs to you

The arrangement to be designed separates your users’ everyday access from a specialist’s exceptional access. For support, the scope, the duration and the permitted actions are agreed before the connection. The authorization remains revocable; the session must end at the scheduled time.

• An explicit request: you trigger the intervention and authorize the session concerned.
• A limited scope: the rights cover the components needed for the diagnosis, not your whole company.
• Traceable actions: the mechanism chosen must make it possible to log the connections and the relevant operations, with a defined retention period.
• A verifiable close-out: temporary access is removed and the changes made are summarized for your team.

A sensitive action — deleting data, changing rights or restarting a process that sends information — requires suitable approval. Being present on a video call replaces neither this authorization nor the technical controls.

What we test before going into service

An interface that opens is not enough to validate operation. The project scope provides for observable tests, on test data or in an isolated environment where necessary.

• Restart: bring the service back after a planned shutdown, then check a representative task.
• Access: allow a workstation, refuse an account outside the scope and check the separation between users.
• Support: open a temporary session, revoke it while it is in use and check that it expires.
• Degraded connection: observe what remains usable and what stops when the network or a dependency is missing.
• Restore: recover backed-up data and check that the application can actually use it.

The results, limits and recovery steps are documented for the people who will run the operation.

Available does not mean infallible

Power supply, network, storage and software dependencies remain possible points of failure. Service hours, backups, alerts and support response times must be agreed; this guide does not promise permanent availability. A disk replacement, a physical connection or certain network failures may still require someone on site.

The Cristallo OS direction

Cristallo OS is the working name of this direction. It is under development; it is not a product ready to install.

The idea is to bring the machine’s management, its services and support together in an operations layer that the client can understand. For a current project, Saphan starts from your tools and a specific need: the components, responsibilities and features that can actually be delivered are defined in the proposal, without imposing a global migration on you.