No long strategic track. We pick one useful workflow that saves your team time right away, install the AI agent on a system at your premises and guide your team, one step at a time. From first conversation to evaluation usually takes 2 to 4 weeks.
A short call: which work costs your team the most time, and does Hermes fit? You talk, we ask questions. No preparation needed.
Together we choose one specific recurring task and agree on what you want to see working after four weeks.
We install Hermes on a system in your network and configure projects, knowledge and access. You do not need to install anything.
Your team starts using the new workflow. After the evaluation you decide: stop, continue or expand.
A pilot with Voltti runs in 4 steps and usually takes 2 to 4 weeks, from the first conversation to the evaluation. That period is short on purpose: you want to know quickly whether an agent actually makes your work easier, not after six months.
Pilots rarely fail on the technology; they fail on the setup. MIT's report The GenAI Divide (2025) points the same way: only about 5% of the AI pilots it studied delivered rapid revenue growth. According to the researchers, the cause was not the quality of the models but how they were fitted into daily work, as Fortune summarizes the research. That is why we stick to one task instead of ten: the agent proves itself on one recurring workflow, such as quote follow-up or a weekly report. We agree on one measurable outcome upfront, so 'useful' is something you can check after four weeks instead of a feeling. And the agent prepares work while a human approves it: publishing, sending and invoicing never happen without someone reviewing them. Not sure where to start? Read how to choose one useful workflow.
Then we tell you honestly. You are not locked into anything: no obligation to continue. Sometimes a task turns out to need a human touch after all, or the return is too small to automate. Stopping is then cheaper than continuing, and at least you know where you stand. If the pilot does prove useful, we discuss support: updates, monitoring, backups and recovery tests. More about how we protect data and access, on the security and privacy page.
Voltti guides pilots for businesses in Belgium and the Netherlands, remotely and on site. Curious what else an agent is good at? Browse the use cases, or read what an AI agent costs — including the three cost categories that also appear in your quote. More about Voltti and how we work, on the about page. Prefer to have Voltti set it all up for you? Read about AI agent development.
| Cost category | One-time or recurring | Depends on |
|---|---|---|
Installation and configuration | One-time | Your chosen workflow, hardware and any integrations |
Support | Recurring | How much guidance and maintenance you agree on |
Model usage | Recurring, based on usage | Your own provider and how much the agent works |
We handle most of the setup. Here is what you need and what Voltti provides.
A network, two or three team members and a recurring task that takes up time.
Hardware if needed, installation, configuration, onboarding and support.
Voltti handles updates, monitoring, backups, recovery tests and your team's questions.
Three clear cost categories: one-time installation, support and your own model usage. You receive a custom quote so you know what to expect.
A pilot usually takes 2 to 4 weeks from the first conversation to evaluation. The exact timing depends on the workflow you choose, the integrations it needs and how much time your team can set aside for onboarding and feedback. We agree on the measurable outcome upfront, so you know where you stand at the evaluation. Not sure where to start? Read how to choose one useful workflow.
A pilot is always a custom quote. The price depends on installation and configuration, any hardware, onboarding for your team and the integrations you need. Model usage runs through your own provider. We discuss the costs openly during the first conversation, with no obligation. Read more about the three cost categories of an AI agent.
If the pilot is useful, we discuss support: updates, monitoring, encrypted backups to a separate location on your network and recovery tests. After that, you can expand step by step to a next workflow. If the pilot is not useful, you simply stop: there is no obligation to continue. More about backups, monitoring and security within support.
Yes. Hermes is open source, so your team can handle updates and maintenance itself. You can also choose support from Voltti: we then take care of updates, monitoring and backups. That way your team can focus on its work rather than the infrastructure. Learn more about Hermes Agent and what the environment needs.
Yes. Voltti works with businesses throughout Belgium and the Netherlands. The first conversation, choosing the workflow and much of the onboarding can happen remotely. Installation happens on site or remotely, depending on your setup and the hardware you already have.
A copy-ready briefing in seven questions: task, sources, output, rights, boundaries, acceptance tests and ownership. Fill it in before your pilot.
What an AI agent can do in customer service: first-line support, triage, and follow-up, with a clear escalation path to your team.
Three cost categories, eight price factors, and a quote checklist: how to compare AI agent quotes on the same scope. Always a custom quote.
Your own AI assistant for business? Compare four DIY options, learn to install Hermes, and see when outsourcing with support makes sense.
AI agent or chatbot? Compare 7 differences, costs, and a decision table to work out which capabilities your business needs.
What is an AI agent? Get a jargon-free definition, see how it turns instructions into results, and learn what it can do for your business.
