Using AI safely in your business comes down to selected access per employee, minimal permissions per connection, encrypted daily backups to a separate location in your network, and human approval for higher-risk actions. That keeps your business data inside boundaries you control.
Open-source Hermes. Independently supported by Voltti.
We configure access, backups, and review gates as part of the setup. Each needs to match the work the agent is allowed to do.
Give access to the employees you select. Grant each connection only the permissions it needs for the agreed task.
Encrypted backups to a separate location in your network. Periodic restore tests check that the environment can be recovered.
Configure approval gates for higher-risk actions and retain logs of the work so people can review what happened.
Private access through the business network and VPN.
Access for selected employees, with the ability to revoke it.
Daily encrypted backups and periodic restore tests.
Selected systems with only the permissions the workflow needs.
The EU AI Act and the GDPR ask companies to manage risks, restrict which data an AI system may see, and be able to explain what the system does. The AI Act entered into force on August 1, 2024 and has applied since August 2, 2026, with some exceptions. It sorts AI systems by risk. According to the European Commission, the vast majority of AI systems currently used in the EU fall into the lowest category, with minimal or no risk. Since February 2, 2025, every business that uses AI must also make sure its employees know enough to work with it responsibly. For a small or medium-sized business, security and privacy come down to ordinary questions: who gets access, which data goes to a model, and what happens when something breaks. The four measures above line up with exactly those questions.
Under the GDPR, your business is usually the controller: you decide why and how personal data is processed. A party that only processes personal data on your instructions, such as a cloud service, is a processor, and you record those arrangements in a processing agreement. The Dutch Data Protection Authority explains the difference. We go over which party has which role with you for each setup.
Selected access and minimal permissions are data minimization in practice: the agent only sees what the task requires. That is also the core of privacy. For each connection, you decide which data the agent may read and which data may go to an external model provider.
Logging makes transparency concrete, because you can show afterward what the system did. It also lets you demonstrate that people review the work when someone asks how a result came about.
Daily encrypted backups to a separate location in your network limit the damage when something goes wrong. Because restore tests run periodically, you know in advance that the environment can be brought back.
Our safety checklist for AI and business data walks through seven points to review before installation. For the regulatory side, the European Commission’s overview of the AI rulebook and the official GDPR site are the authoritative sources.
Handled by the Voltti team, serving businesses in Belgium and the Netherlands.
| Measure | What it means | How Voltti sets it up |
|---|---|---|
Network access | The agent stays reachable inside your own network | Private access through the business network and VPN, not the open internet |
User access | You decide who gets in | Access per employee, revocable at any time |
Backup and recovery | Daily encrypted, tested periodically | Backups to a separate location in your network, with restore tests |
Integration boundaries | Every connection gets minimal permissions | Only the systems that add value, with only the rights the task needs |
The employees you select. Each employee gets their own access, which you can revoke at any time. Voltti does not have permanent access to your daily work; remote support is provided when you request it. We agree on the access needed for that support. Use our practical safety checklist for AI and business data as a reference.
No. Voltti does not monitor your use of Hermes or read your prompts for its own purposes. Our role is installing Hermes Agent and providing support afterward. If support does require access, we agree on that access with you in advance and use it only for the task you asked us to handle. Your data stays yours.
You choose the provider and what data may be sent to it. We advise on the setup, but you remain responsible for the provider agreements and how the service fits your processing obligations. Review its retention and data-use terms before sending business data. If certain data should not leave your network at all, a local model can handle those tasks. Also read about local and hybrid AI setups for businesses.
We put 4 practical safeguards in place: selected access, encrypted backups, limited permissions, and logging. These measures do not make every use case automatically GDPR-compliant. We discuss the processing, responsibilities, and required agreements for your situation. Under the AI Act, your employees also need enough knowledge to work with AI responsibly. Our checklist for safe AI use summarizes the practical measures.
The support setup includes daily encrypted backups to a separate location in your network, never on the node itself, and periodic restore tests. If an incident occurs, Voltti helps with recovery and analysis within the agreed support scope; the logs show which actions the agent took. Read how we handle backups and recovery for an AI environment, or see our approach from pilot to support.
See how Sisu combines Hermes, Home Assistant, and local device APIs on a Raspberry Pi 5 for practical control through chat.
See when Hermes on your own computer is enough and when a managed Linux computer is a better fit for your team and recurring tasks.
Hardware fails. Learn how Voltti handles backups and recovery tests, and what you can do to prepare your AI environment for a hardware failure.
What computer do you need for local AI? Compare models, RAM, unified memory, NVMe offload, and hardware prices. Start with your tasks and model choice.
In a hybrid AI setup, the environment runs locally while language processing can use an external provider. See exactly what stays with you and what leaves.
Seven checks for safer AI use in your business: selected access, minimum permissions, backups, and human oversight. Review your setup before connecting data.
Local AI runs on your hardware; cloud AI runs through a provider. A hybrid setup suits most businesses. Compare the options and make an informed choice.
