AI agentdevelopment

You do not need to build AI yourself. AI agent development at Voltti means we set up an open-source Hermes agent for one task in your business, with integrations, tests and support.

Open-source Hermes. Independently supported by Voltti.

What VolttiDelivers

You do not want an AI project, you want a working agent. When Voltti develops an AI agent for you, you get four things: a configured agent, a tested task, integrations with minimal permissions and support after the handover.

01

A configured agent

Open-source Hermes, installed and configured on a system in your network. No new base model and no custom framework, but an existing agent with an active community.

02

A tested task

One specific recurring workflow the agent runs or prepares. We test with your own examples before your team starts using it.

03

Integrations and access

Connections to the software you already use, with minimal permissions per integration. Only what the task truly needs gets access.

04

Handover and support

Training for your team, documentation of the agreements and support from Voltti: updates, monitoring and backups. You stay the owner of your systems and your data.

How it works

  1. 01

    Introduction and intake

    A short call about the task you want to tackle: what input does the agent get, what should come out, which systems are involved? You do not need any technical preparation: we ask the questions.

  2. 02

    Proposal and quote

    You receive a proposal with scope, integrations, planning and a custom quote. Only after you approve it do we start building.

  3. 03

    Build and test

    We configure the agent, set up integrations and test the task with your own examples. Your team approves the result before it goes live.

  4. 04

    Deliver and guide

    We deliver, train your team and evaluate together after the first weeks. Then you decide whether to stop, continue or expand, with support from Voltti.

Google Gemini
Moonshot AI
Hermes Agent
Grok
DeepSeek
Mistral AI
OpenAI

Buying questionsAnswered honestly

In short: AI agent development at Voltti starts with a custom quote. A pilot usually takes 2 to 4 weeks, from the first conversation to the evaluation. You stay in control of your systems and data. Below are the details on costs, timing, control and limits.

01What determines the quote?

You pay for three things. First, a one-time installation and setup, which depends on your workflow, hardware and integrations. Second, ongoing support: the guidance and maintenance you agree on. Third, your own model usage with your own provider. Hardware is an option, not a requirement. Prices you find online come from agencies that build agents themselves. Budget mainly for running costs. According to research firm Gartner, sending a task to a reasoning AI agent costs at least five times as much as a basic chatbot interaction. Gartner expects the inference cost per agent workflow to rise more than fivefold through 2028. Your price is always a custom quote, discussed openly during the introduction. More about what an AI agent costs.

02How long does it take?

A pilot usually takes 2 to 4 weeks, from the first conversation to the evaluation. That is weeks, not months. We deliberately do not promise a fixed delivery date: setting up the right task properly matters more than speed. At the end, we evaluate together whether the agent does what you need. Read how a pilot runs from intake to delivery.

03Who has control and who owns it?

You do. The agent works within boundaries you approve: which systems, which permissions, which actions. Publishing, sending and invoicing never happen without a human reviewing them. The installation runs on a system in your environment and the code and repositories live wherever you want. That is how it works at our own company: every content change at Voltti starts as a ticket the agent picks up. The agent reads our Search Console data through read-only access. Then the agent updates the page in the CMS and writes a verification report in the ticket. A human reviews that report and corrects where needed.

04What are the limits?

An agent makes mistakes, just like a trainee. That is why we test with your examples, keep human control over risky steps and log what the agent did. Voltti does not promise flawless work or guaranteed savings. We do promise an honest evaluation at the end of the pilot. Not sure whether your task fits? Read about how to choose one useful workflow.

Voltti builds and guides agent setups for businesses in Belgium and the Netherlands. Founder Seppe Gadeyne sets up the installations himself; who else is behind Voltti is on about Voltti. Hesitating between building it yourself or having it built? The trade-off is in your own AI assistant: build or outsource. Also read how we handle security and privacy.

Questions and answers

01Is there a difference between having an agent built and developed?

No, not in practice. Both mean a specialist sets up the agent, so you do not have to do it yourself. Voltti offers AI agent development as a defined service with a clear scope. You get one task, the integrations it needs, tests with your own examples, a handover to your team and support afterward. Not a block of hours or a one-off script.

02Does Voltti build its own AI model or framework?

No, and that is a deliberate choice. We configure the existing open-source agent Hermes for your situation: integrations, knowledge, permissions and tests. The source code is public on GitHub. That is faster, cheaper and more maintainable than having a custom framework built. You can read how the agent works on the Hermes Agent page.

03Do we need to assign technical people?

No. For the intake, knowing your own process is enough: which task, which input, which desired output. We handle the technical side, installation and maintenance. In a short session, usually one afternoon, your team learns how to give the agent a task, check its output and see when something needs a correction. After that, your questions go to Voltti support.

04What happens to our data?

The agent runs on a system in your own environment and uses only the integrations you approve, with minimal permissions. Encrypted backups go to a separate location in your network, not to the node itself. Which provider you optionally use externally for model usage is your choice. More about security and privacy at Voltti.

05What if we already have an AI subscription?

Some subscriptions work and some do not: Hermes cannot use every AI subscription as its language model. A ChatGPT plan for Codex works, Claude Pro does not. Model usage runs through your own account, with a subscription or an API key. Our article on how much model usage costs explains what a subscription covers. In the intake, we look at what you already have and include it in the quote.

06Do you also work in the Netherlands?

Yes. Voltti works for businesses across Belgium and the Netherlands, remotely and on site. Dutch businesses get the same approach: a remote intake, the installation on a system in your own environment and support remotely or on site. See our approach for a pilot.

Read more

Which task do you want built?

Discuss your pilot