Technology
Technology and capabilities
We choose boring, well-supported technology on purpose. Every tool below is open or widely used, well documented and maintainable by any competent team — because you will own the system long after the project ends.
What we build with
The core stack for ERP, CRM, portals, integrations and automations.
Web and mobile applications
TypeScript end to end: React front-ends, Node back-ends, Cloudflare Workers at the edge, PostgreSQL, SQLite or Cloudflare D1 for data and R2 for files. Responsive interfaces that work on the phones and low-cost laptops your staff already have.
Integration
REST and GraphQL APIs, webhooks and n8n workflows connect accounting, payment gateways, e-commerce platforms, messaging and legacy databases — so data is entered once.
AI with human approval
LLM APIs for document extraction, drafting, classification and summaries — always with a review step and an audit trail before anything is written back to your system.
Infrastructure
Cloudflare for hosting, DNS and edge security; managed cloud databases; automated, encrypted backups with tested restores; environments separated for development, testing and production.
Data and BI
KPI dashboards, scheduled reports by email or WhatsApp, and clean exports for your accountant, Steuerberater or auditor.
Engineering practices
Code review, automated tests, written documentation and monitoring on every project — the unglamorous habits that keep a system reliable for years.
Why this stack
Maintainability first. TypeScript across front-end, back-end and automation means one language, one type system and fewer surprises when someone else takes over. PostgreSQL and SQLite are stable, portable and understood by every developer.
Cost that fits an SME. Cloudflare Workers, D1 and R2 keep hosting predictable and low, and scale without a large operations team. For clients who prefer their own cloud or on-premise servers, the same code runs there.
No lock-in. Everything is delivered in your accounts and repositories, with documentation. If you later choose to work with someone else, they will find familiar tools, not a proprietary platform.
AI kept in its place. We use language models where they save real time — reading documents, drafting replies, categorising records — and pair them with human approval and logging. They do not make decisions on their own.
Not sure which parts of the stack you need?
In a free 30-minute call we look at your current tools and tell you honestly what to build, what to connect and what to leave alone.