How we work
The road from an idea to a product you can use.
We do not tell an AI “build this” and hope. Every project walks the same road. Agents do the work at each stop. A person with experience checks it before we move on.
The idea
It starts with a problem: yours, or one we met ourselves. We write down who has it and what would be better if it were solved.
Agents: research the problem and what exists.
Person: decides if it is worth building.
Requirements
What it must do, for whom, and what it must never do. A clear list, with a test for every point, so “done” is not an opinion.
Agents: draft the list and find the gaps.
Person: cuts it to what matters.
Architecture
The parts of the system and how they talk to each other. Where the data lives, where the AI sits, what happens when something fails.
Agents: draw options and compare them.
Person: chooses, and writes down why.
Data model
The tables, the fields and who may see what. Planned before the first line of code, because a wrong data model is the most expensive mistake.
Agents: propose the model and the access rules.
Person: checks privacy and security.
Process and experience
How a person moves through it, step by step. We understand the real process first, then make the path short and hard to get wrong.
Agents: map the process and the screens.
Person: walks through it like a real user.
Interface design
What you see and touch. One shared design system across all our products, readable, and usable with a keyboard and a screen reader.
Agents: build the screens as real, clickable pages.
Person: decides the look and the words.
Build and test
The tests are written first, from the requirements. Then agents build until every test passes. A second agent checks the work of the first.
Agents: write code, tests and documentation.
Person: reviews every change before it is merged.
Ship and care
It goes live with documentation, a legal check and monitoring. Then we watch how people use it and improve it.
Agents: deploy, watch and report.
Person: says yes to every release.
Human in the loop
AI makes us faster. Experience keeps it right.
Nothing moves to the next stop without a person's yes. That person has 15+ years of building software for aviation and logistics, and knows what a mistake costs.
- Every stop ends with a review
- One agent builds, another one checks
- Tests decide what “done” means
- Decisions are written down, with the reason
Technology
What we build with.
The tools we use every day. Looking for one in particular?
AI models and agents
- Claude
- OpenAI
- Local models
- Copilot Studio
- MCP
Code and delivery
- GitHub
- TypeScript
- React
- Next.js
- Python
Cloud and hosting
- Cloudflare
- Azure
Microsoft
- Power Platform
- Power BI
- Dataverse
- SharePoint
No technology matches your search.
Brand icons come from Simple Icons. Products without an icon are shown with their initials.
See the road in a real product.
Every product page shows what we built and how.