Managed AI agents on your terms
We build, run, and maintain custom AI agents for your company. On trusted servers, with self-hosted LLMs, GDPR-compliant, and fully on-premise if you want it.
Off-the-shelf AI tools send your data to foreign clouds
Most AI tools are thin wrappers on a US model. Your quotes, contracts, customer data, and internal processes leave the company and land on servers you do not control. For regulated industries and the German Mittelstand, that is not an option.
At the same time, repetitive work eats capacity every day: writing quotes, checking invoices, answering requests, compiling reports. A generic chatbot does not solve this, because it does not know your processes and nobody runs it for you.
Agents we build and run for you
We build AI agents tailored to your specific processes and connected to your systems. They run on trusted infrastructure: self-hosted LLMs on our EU servers, or directly inside your own network.
You do not get another tool to operate yourself, you get an outcome. We build, monitor, update, and improve the agents continuously. You keep control, we run the operation.
Three steps to a managed agent
From process to a production AI agent running on your infrastructure.
Understand the process
We analyze a concrete process that fits automation and define the goal, the data sources, and the points where a human should approve.
Build and integrate the agent
We build the agent, connect it to your systems such as ERP, CRM, email, or file storage, and test it on real cases from your operation.
Run and improve
The agent runs monitored on trusted infrastructure. We maintain, update, and improve it continuously. You can see what it does at any time.
An agent that fits your company
Tailored to your processes
Not a generic chatbot. Each agent is built for a concrete workflow in your company and speaks the language of your operation.
Integrated with your systems
Connected to ERP, CRM, email, document storage, and other tools. The agent works where your data and processes already live.
Fully managed
We build, monitor, update, and improve. No maintenance burden on your side: you get an outcome, not another project to run.
Transparent and traceable
You see what the agent does and why. Every action is logged and auditable, no black box.
Human stays in control
Critical steps go to a human for approval. The agent acts autonomously where it is safe and asks where it matters.
Gets better over time
Every agent learns your cases, terms, and exceptions. The value compounds over time instead of being obsoleted by the next model.
Agents we build
Concrete use cases from the daily reality of mid-market and industry. Your agent is tailored to your process.
Quoting and invoicing agent
Drafts quotes from inquiries, checks incoming invoices against orders, and prepares the approval. You decide, the agent does the groundwork.
Support and communication agent
Answers recurring requests, drafts replies, and escalates complex cases to the right person with the right context.
Reporting and monitoring agent
Collects data from your systems, produces regular reports, and flags anomalies before they become a problem.
AI agent pricing and capability levels
The level depends on how many systems the agent works across, how much judgement the task needs, and how independently it acts.
Assistant
Works in one system plus email. Follows a fixed sequence and prepares work for a person to approve.
Typical work- Drafts replies to recurring inquiries
- Transfers data from incoming documents into your system
- Produces a standard report from one source
Specialist
Works across two to four systems. Applies your rules and exceptions, acts within defined limits, and escalates edge cases.
Typical work- Drafts quotes from inquiries using price lists and history
- Checks incoming invoices against orders and prepares the approval
- Triages support requests and answers the routine ones
Expert
Works across five or more systems, or inside your own network. Handles open tasks that need judgement and coordinates other agents.
Typical work- Reconciles orders against invoices and investigates anomalies
- Monitors compliance across sites and raises findings
- Combines several sources into a recommendation you act on
Included in every monthly price
- Model and token costs
- Hosting, monitoring, and incident response
- Model upgrades as better models ship, at no change in price
- Tuning and corrections within the agent's defined scope
Setup and integration
- Setup is quoted per customer, based on how many systems we connect
- Systems that need custom or on-premise integration are priced separately, from 1,200 € per system
- We tell you which category your systems fall into before you sign
Adding more agents
- No setup fee for additional agents on systems we already connect
- From six agents, 20 percent off the monthly price
- Move an agent up a level when it outgrows its scope
12-month minimum term. Hosted in the EU. On-premise operation available. All prices exclude VAT.
Built for European data protection requirements
Self-hosted LLMs
The models run on our trusted EU servers or in your own data center. Your data does not train foreign models.
Data stays in the EU
Processing and storage in Germany or the EU. GDPR-compliant, with clear data processing agreements.
On-premise available
On request, the entire agent runs inside your own network. Nothing leaves your infrastructure.
Auditable and certified
Every action is logged. Role-based access, encryption, and an information security management system run to ISO 27001 are included.
Frequently asked questions
What is an AI agent?
An AI agent is a program that completes a task on its own instead of just answering questions. It understands a goal, accesses your systems, executes steps, and requests approval for critical decisions. We build such agents for concrete processes in your company and run them for you.
Does our data leave the company?
No, not if you do not want it to. We run self-hosted LLMs on trusted EU servers or directly inside your own network. On request, everything runs fully on-premise, so no data leaves your infrastructure.
Which language models do you use?
We prefer open, self-hosted models that we run on our own infrastructure. If a commercial model is clearly better for a given use case, we use it only with your explicit approval and with appropriate data protection agreements in place.
Can the agents run on-premise?
Yes. The entire agent including the model can run in your own data center or private cloud. This is especially relevant for regulated industries and sensitive data.
Which systems can this integrate with?
Common systems such as ERP, CRM, email, document storage, and internal tools, through their interfaces. If there is an API or a defined access path, an agent can work with it.
What does a managed AI agent cost?
An agent costs between 199 € and 899 € per month depending on its level. An Assistant is 199 €, a Specialist 499 €, and an Expert 899 € per agent per month. Model and token costs, hosting, monitoring, and model upgrades are included. On top of that there is a one-time setup fee that depends on how many systems we connect.
How quickly can we go live?
We work in days, not quarters. A first, clearly scoped agent can be in production within a few weeks. We deliberately start with one concrete process and build from there.
What happens when something goes wrong?
Critical steps always go to a human for approval. The agent is monitored, every action is logged, and we step in on anomalies. You keep control at all times and can stop the agent.
What does an additional agent cost?
There is no setup fee for additional agents that use systems we already connect. From six agents the monthly price drops by 20 percent. The minimum term is 12 months.
Related solutions
Managed AI agents complement your existing processes
FlowForward - free of charge for nonprofit organizations
Registered nonprofit organizations get these AI agents free of charge. Through our FlowForward initiative we build and run the agents at no cost, so nonprofits can focus on what truly matters: their mission.
Learn more about FlowForward →Ready for your first AI agent?
In 30 minutes we find a process together that fits a managed AI agent.
Book a demo