You build the agents. We run them.
A dedicated server in the EU where your AI agents and automations run around the clock. Our engineers handle scheduling, retries, alerts, spend limits, deployments and backups, so you know what ran, what failed and what it cost. From €119 per month, excl. VAT.
One monthly price per server: you know the cost before the month starts. No per-run platform fees
Monitoring on every plan. Response times depend on your plan
One server for your agents only, in an EU data centre
The agent works. Running it is the hard part.
Building an AI agent now takes days. Keeping it running every night, every weekend and every month is a different job, and in most companies nobody has it.
The agent that sorts incoming orders runs on a developer's laptop. It stops when the laptop goes to sleep, and nobody notices until a customer asks where their confirmation is.
An automation fails at two in the morning because an API changed. There is no alert and no log anyone can read, so on Monday morning someone has to work out what was processed and what was not.
An agent gets stuck in a loop and calls the AI provider thousands of times. You find out from the invoice at the end of the month. On another day the AI provider has an outage, and every agent that depends on it simply stops.
Usage-priced platforms solve part of this, but you pay for every run, often with a limit on how long a task may take, and your data and code sit on their infrastructure. We take a different route: one server for your company, with operations included.
What we run for you
If your developer or partner has built it, it can most likely run here. For example:
- AI agents written in Python or Node.js, whichever framework your developer chose
- Anything that runs in a container
- Workflow tools for automations between your systems
- Queues and scheduled jobs that feed your agents
- PostgreSQL with pgvector for agent memory and your documents
What operations include
Schedules and retries
Agents start on time, every time. When a run fails, it is retried automatically, and if it keeps failing, we are alerted.
Alerts that reach a person
Failed runs, stuck jobs and a full disk raise an alert with our team, not in an inbox nobody reads. Monitoring runs 24/7 on every plan; on Start and Team we respond during business hours, on Individual around the clock.
Run history
Every run is recorded: when it started, what it did and how it ended. When someone asks what the agent did last Tuesday, you have the answer.
Spend limits
Your agents get a limit for AI usage, on Team and Individual a separate budget for each agent. When it gets close, you get a warning; when it is reached, the agent stops instead of running up the bill.
Keys kept safe
API keys for AI providers and your systems are stored encrypted on your server, not in a script or a shared document.
Deployments and rollback
A new version of an agent goes live in a controlled way. If it behaves worse than the old one, we roll it back.
Backups with restore tests
Configuration, data and run history are backed up to separate storage, and we regularly test that a backup can be restored.
More than one AI provider
Your agents use the AI providers you choose, for example OpenAI, Anthropic, Google or Mistral, with your own accounts and API keys.
Every AI provider has outages, including the largest. When yours does, runs are retried and queued instead of lost, you get an alert, and where your agent supports it, work can switch to a pre-agreed European provider until the main one is back.
Data and access
- Your agents, data, logs and run history are on your dedicated server in an EU data centre
- A data processing agreement (DPA) under the GDPR for the server and our operations
- API keys stored encrypted on your server
- Masking of personal data before it goes to an AI provider, where your use case allows it, to reduce what leaves your server
- Access for your developers and partners with personal keys that you can revoke at any time
The server runs in a Hetzner data centre in Falkenstein, Germany, or in Helsinki, Finland, on request. It is used by your company only.
Who does what
We do not build agents. A clear split keeps both sides honest.
You or your developer
What the agent does and decides: its logic, prompts, approval steps and the choice of AI provider. You are responsible for the agent's results.
Our team
Everything that keeps it running: the server, schedules, retries, alerts, spend limits, deployments, backups and the run history that shows you what happened.
Plans and prices
You pay per server, not per run or per AI request. Set-up is free.
| Plan | For | Server | What's included | From, per month |
|---|---|---|---|---|
| Start | A few agents and automations | 6 cores (12 threads) · 64 GB RAM · 512 GB SSD | One environment. Schedules, retries, alerts, run history, spend limits, key management, backups and 24/7 monitoring. Support during business hours, reply within one business day. | €119 |
| Team | More agents, several people working on them | 8 cores (16 threads) · 64 GB RAM · 512 GB SSD | Production and test environments. Everything in Start, plus deployments from Git with rollback and a budget per agent. Support during business hours, reply within one business day. | €219 |
| Individual | Larger volumes and connections to company systems | From 16 cores (32 threads) · 128 GB RAM · 2 TB SSD | Scope by agreement: a more powerful server or several servers, a VPN into your company network, an optional standby server, log retention by agreement, 24/7 support. | €549 |
Dedicated server with mirrored disks, or equivalent. Prices per server per month, excl. VAT. Set-up is free, the minimum term is 3 months. AI usage is paid by you directly to your AI provider.
How it starts
Send us a description
What your agents do, how often they run, where they run today and which AI providers they use. A few lines are enough. No commitment.
Plan and price
We reply within one business day with the plan that fits and a monthly price. If a usage-priced platform would be cheaper for you, we say so.
Set-up and move
We prepare the server, move your agents from the laptop or trial account, and set up schedules, alerts and spend limits with you.
Ongoing operations
We run the server, watch the runs and handle failures, updates and backups. Your developer keeps working on what the agents do.
Questions about AI Agent Hosting.
Do you build the agents?
No. We run them. The agents are built by your own developers or a partner of your choice, and they get access to the server to deploy and improve them. If you are still choosing a partner, we tell you what they need to deliver so that we can run it.
Which tools and frameworks can run?
Anything that runs in a container: agents in Python or Node.js with the framework of your choice, workflow tools for automations, queues, scheduled jobs and databases such as PostgreSQL with pgvector. Send us a short description and we confirm it before you sign anything.
What happens when the AI provider is down?
Runs are retried and queued instead of lost, and we get an alert. Where your agent supports it, work switches to a pre-agreed European provider until the main one is back. When the provider recovers, the queued work continues.
How do you keep AI costs under control?
Your agents get a spending limit, on Team and Individual a separate budget for each agent. You get a warning as it gets close, and when the limit is reached, the agent stops. You pay your AI provider directly, so the full cost appears on your own invoice, not hidden in ours.
Why a fixed price instead of paying per run?
Because you know the cost in advance: one price per server per month, however often your agents run or however long a task takes. In pure numbers it depends on volume. For a few runs a day, a platform that charges per use is usually cheaper. For agents that run around the clock or process a lot of work, a fixed server price often comes out better, and there is no limit on how long a task may run. Send us a description and we calculate both options for you.
Send us your agent. We tell you what it costs to run.
Describe what your agents do, how often they run and where they run today. We reply within one business day with a plan and a monthly price. No commitment.