hansenexus
Pricing
Intro callSovereignty check
hansenexus
  • Services
  • Who it's for
  • Pricing
  • How we work
  • Training
  • Blog
Intro callClient portal
Sovereignty check
hansenexus
Talk to us directly, Mon to Fri, 9 am to 6 pm040 57307846[email protected]
Already a client? Go to the client portal

Company

  • About
  • Contact
  • FAQ
  • Developer subscription

Inside

  • Stack
  • Tools
  • Quality
  • Lab
  • Design system

Legal

  • Imprint
  • Privacy
  • Terms
  • Cancel subscription
© 2026 hansenexus GmbH, Hamburg
Deutsch

How we work

AI agents deliver.People are accountable.

I work with a team of AI agents. They write code, test and document in parallel. Every change runs through automated checks, and before anything reaches your systems, I approve it.

Our rulebookalways applies
  • Every task in its own workspace.
  • Nothing goes live unchecked.
  • No change to your systems without approval.
  • Credentials never in plain text.
  • Watched around the clock.
  • Decisions are recorded.

The process

Five steps. Two of them are mine.

  1. Step 1

    Clarify the task

    We talk about what you need. I write down what done means before an agent starts.

    Human
  2. Step 2

    Build in parallel

    Development, test and documentation agents work at the same time, each in its own workspace.

    AI agents
  3. Step 3

    Check automatically

    Function, security and privacy are checked. Red means: back to the agent.

    AI agents
  4. Step 4

    Approve

    I review the result and approve it. Without that yes, nothing reaches your systems.

    Human
  5. Step 5

    Live and watched

    A backup before every update, monitoring around the clock, one click to roll back.

    Agent and human

Evidence

How you can tell the rules apply.

  • Every task in its own workspace.

    Agents work in parallel without getting in each other's way.

    Every task gets its own copy of the code and its own ticket.

  • Nothing goes live unchecked.

    Automated tests and checks run before every change. Red means no release.

    Every feature is described in the behaviour catalogue and checked on every change.

    How we check
  • No change to your systems without approval.

    A person always approves changes to production systems.

    Technically locked: commands against client systems stop and wait for my yes.

  • Credentials never in plain text.

    Secrets live in the vault and are only injected at the moment they are used.

    The vault is 1Password, openly named in our list of services outside the EU.

    See the stack
  • Watched around the clock.

    A monitoring agent checks availability and certificates and reports anomalies.

    A deep health check every 15 minutes, restarts detected every 5 minutes.

  • Decisions are recorded.

    Every architectural decision is documented and traceable, for you and for any successor.

    With a date, the reasoning and what it replaced.

What it means for you

More done per hour.

Faster
Several tasks run at the same time instead of one after another.
Cheaper per result
You pay for results, not for typing.
Safer
Fixed rules instead of gut feeling.
Traceable
You can see what happened and why.

And your data?

You decide what an AI sees.

  • Source code

    For development we use leading models, mainly Claude from the US. Which code they see is agreed with you in advance.

  • Personal and business data

    We only process it with AI on our own hardware in Germany or with your explicit approval.

  • Stricter on request

    Then we work exclusively with local models or models operated in the EU.

All services outside the EU

For engineering teams

This way of working for your team.

For startups and agencies we set up the same environment in your company: agents, guardrails, approval rules.

Fixed price after a preliminary call

Book preliminary callJust the training
  1. 1Analysis of your development workflow
  2. 2Setting up agent tooling, guardrails and approval rules
  3. 3Protecting credentials and secrets
  4. 4Agentic coding training for your team
  5. 5Four weeks of support in day-to-day work

See it applied to your case.

In the intro call I show you how one of your tasks would run through this process.

Book intro callSovereignty check