Software Engineering
We build products that can be maintained, extended and handed to another team long after the first release.
What the technology gives you
The product belongs to you, code and data included. You decide what changes and when, regardless of any vendor’s plans. It connects to your other systems, so no one enters data twice. The interface follows your people’s daily tasks, and they learn it quickly. It meets security and accessibility requirements from the first version, so audits do not force rework. We design and build new products, modernise old systems and support them after launch.
What’s included
- Architecture design
- Backend and user interface
- Testing and release
- AI features in the product
What it solves
We work around the system in spreadsheets
We build the product around your roles and permissions, so all the work happens inside it.
We find data errors too late
The system checks data as it is entered, so errors show up before they affect daily work.
Manual testing holds up every release
Automated tests check every change, so a release no longer waits a week for manual checks.
We fear changing what already works
Tests pin down how a module behaves today, so the team can change it safely.
Only one person understands the code
A second engineer reviews every change, so no module depends on a single person.
The team grows, but releases slow down
Clear module boundaries let teams work in parallel without getting in each other’s way.
When you need it, and when you don’t
You need it
- the product you need is not on the market
- an old system holds back business change
- external requests keep growing and the current way of handling them can’t cope
You don’t
- an off‑the‑shelf SaaS already does the job: a subscription and setup are enough
- you need a prototype in two weeks: it is faster to assemble from ready‑made blocks on UnityBase
How we do it
- 01
We guard the architecture from day one
An architect reviews every large task, and each sprint leaves time to tidy the code. That keeps the pace from dropping in year two.
- 02
We start with one modular system
We split out a separate service only when it must be released or scaled on its own. Otherwise running many services eats the time new features need.
- 03
AI drafts, engineers decide
AI drafts code and tests, and they pass the same checks as an engineer’s code. If an AI feature in the product goes wrong, it is switched off and rules or people take over.
Project stages
- 3–4 weeks
Discovery
Processes, risks and recorded decisions before the first line of code.
- 3–4 weeks
Architecture
Modules, data exchange, reliability targets and a recovery plan.
- 3–4 months
First version
One scenario with live users tests the architecture and assumptions.
- 6–18 months
Expansion
Features added in stages, the most valuable first.
- ongoing
Support
Component updates and regular security checks.
Platforms and standards
Java · .NET · Python · Node.js · Go · UnityBase
React · Vue · Angular · Next.js · TypeScript
Docker · Kubernetes · GitHub Actions · GitLab CI · Playwright · JUnit
ISO/IEC 27001:2022 · NIST SSDF · OWASP ASVS 5.0 · WCAG 2.2 AA · SBOM · Cyber Resilience Act
Questions
How long does a custom product take?
Production readiness usually comes 6–12 months from the start. A product promised in two months is most often a template or rework you will pay for later.
How do you measure a development team?
With the DORA metrics: how often and how fast changes ship, how many fail and how quickly service is restored. Targets are set for each service.
Does AI see our code and secrets?
Only the code the role may access. Secrets and personal data never reach the AI.
Further reading
What do you want to build?
Thirty minutes with an architect to go through the idea and its technical risks.