Softwareentwicklung . KI . Compliance
Systeme,softwarethat hold up in daily use.
For companies without their own development team. From discovery through prototype to operations, which we keep running afterwards.
Services
Nine services.
Six of them without AI.
Most of what slows companies down can be solved with clear rules and clean interfaces. We use AI where it genuinely helps, and say so honestly when it does not.
Why we say it so openly
Many providers sell everything as AI right now.We would rather tell you when you do not need it.
Not because we cannot do AI. We build AI systems ourselves. But an AI system you do not need costs you more, decides less reliably and brings obligations that a simple rule does not.
Rules decide the same way every time
A rule gives the same result for the same case, can be traced and tested. AI works with probabilities and is sometimes wrong, which is fine where a person checks, and a risk where nobody does.
No usage costs per request
An automation built on rules runs without ongoing model costs. With AI you pay per request, and the price and model can change.
Fewer legal obligations
Without AI there is no labelling under the EU AI Act, no classification check and no question of where the model processes your data. That saves you effort before and after launch.
And where AI genuinely helps?
With free text, scans and requests no rule can capture. Then we use it, with fixed limits on what it may decide.
Expertise
What we can do
in detail.
Behind the four services are individual capabilities that we combine depending on the project.
Structured interviews with every department. The result is a data requirements catalogue: which attributes exist, who holds them, what is missing.
From the first conversation
to a running system.
Five steps with fixed hand-overs. At every point you know what we are working on and what comes next.
Discovery, not assumptions
Structured interviews with every department. At the end we know which data exists and what is missing.
1 to 2 weeksConcept and fixed price
Functional concept, data model and interface contract. Scope and price in writing before development starts.
1 weekPrototype instead of a spec
A clickable prototype replaces ten rounds of alignment. Requirements become concrete on a real interface.
1 to 2 weeksBuild
AI-assisted development as a way of working. Architecture and review stay with people.
4 to 12 weeksOperations
Monitoring, backups with tested restores, security updates and ongoing development.
ongoingSelected projects
One system live, three in development.
From client work and our own products. Where a client may not be named, the case is described anonymously. We publish figures only once they are measured.
Deutsche Telekom · live
TalentHub
Corporate, agency and independent sales partners in one system, each with their own view of the same data.
Case studyIT service provider, anonymised · in development
Service Hub
Several service lines that grew apart, no shared source. Building a data foundation that makes services reliably measurable.
Case studyOwn product · in development
Hironaut
Location analysis, ad creatives and reporting in one loop. Every campaign should improve the next one.
Case studyDevonaut · in development
THE CONCIERGE
WhatsApp, email and phone come together in one thread. The AI writes, fixed rules decide.
Case studyContact
What should run automatically for you?
Four short questions, then you choose how we reply: via WhatsApp or by email. It is the same kind of enquiry flow we build for our clients.
Enquiry to devonaut
Four questions, about a minute