We're a small engineering studio from Bosnia and Herzegovina. We build AI systems, Python applications and backends that hold up in production, and now AI agents that work with the systems you already run.
You don't need to rebuild to use AI. We put a secure, agent-ready layer on top of the API you already have, then build the agents that use it.
We turn your existing API into an MCP server that people sign in to with their own account. They connect the AI assistant they already use, like Claude or ChatGPT, and work with your system through it, with their usual permissions and no new interface to build.
A chat assistant inside your product or internal tools that answers from your own data and gets tasks done, with a person approving the steps that matter.
For work too big for one prompt, we build long-running multi-agent flows in LangGraph. Specialised agents pass tasks to each other, check each other's output, and deliver what the user asked for. Every step is traced in Langfuse, so you can see how each result was produced.
An AI document-understanding system that supports humanitarian demining operations in Bosnia and Herzegovina.
Minefield records hold critical handwritten data that has to be digitised. Manual transcription was slow, error-prone, and held up life-saving work.
YOLO object detection combined with custom TensorFlow models for handwritten text recognition, served by a FastAPI backend that processes documents in real time.
We cover the whole technical picture, so you work with one team instead of four.
Custom intelligent systems that change how your business operates.
Production-grade applications built for scale and reliability.
Scalable architectures that power enterprise applications.
Modern infrastructure that scales as you grow.
Real systems for real companies, measured by what changed after launch.
Multi-channel booking with integrated payments for a global travel company.
Consolidate several legacy booking systems and integrate payment providers including Klarna, at high transaction volumes across regions.
A unified FastAPI backend on microservices, with payment orchestration and real-time inventory across every sales channel.
Event processing for cyber intelligence and research data analysis.
Millions of events a day from many sources, needing real-time analysis, sentiment detection and intelligence extraction with minimal latency.
Kafka streaming with PySpark processing, custom NLP models for sentiment, and Elasticsearch for fast querying and visualisation.
A full platform overhaul for a financial services company moving to cloud-native.
A 15-year-old monolith handling sensitive financial data had to be modernised with strict compliance and no downtime.
A strangler-fig migration to microservices, a full API layer, containerised workloads and solid CI/CD pipelines.
We've worked with international clients for years, and we work the same way on every project.
We collaborate with cross-cultural teams across time zones and deliver for clients in several countries.
Our code runs in production, with proper testing, monitoring and documentation.
From ML models to cloud infrastructure, we handle the complete technical picture.
We mentor your team and share what we know, so your organisation keeps the expertise.
Whether you need a data pipeline, an AI solution or a complete backend, tell us what you're working on.