A small, specialised studio: two partners, a network of collaborators, and AI as a multiplier. Agent systems and custom enterprise software, from a testable prototype through to daily operation.
Four capabilities that hold together. An agent that does not integrate with your existing systems, cannot be evaluated and cannot be operated is not a project: it is a demo.
Agents that read your data, use your tools and see real processes through to the end. Orchestration, tool use, document retrieval, memory, escalation to a human.
Continuous evaluation on your own cases · Every decision traceable · Cost and latency under control
Enterprise-grade applications built as microservices, with documented APIs and a release cycle nobody dreads. It is the trade we have practised for twenty years.
.NET and dotnet core · REST and internal APIs · Custom portals and business systems
AI is worth exactly as much as the data it can reach. We connect agents to the ERP, CRM, business systems and document archives you already run, without asking you to start over — and where the data comes from an installation rather than from software, as in home automation, nothing changes.
Document pipelines · Vector stores · Home automation and IoT devices · Synchronisation and migration
Kubernetes clusters, monitoring and backups across remote datacentres. We run our own infrastructure in service of development and, where it matters, the models run on yours: the data never leaves your perimeter.
On-premise or cloud deployment · Self-hosted open models · GDPR compliance
Four phases, each leaving you something in hand and a point where you can stop. We move on only if the previous one produced a measurable result — not if it produced enthusiasm.
We go inside your processes and come out with the use cases ranked by value and by risk. Including the ones best not handed to an AI today.
A working agent on a single case, with its own test suite built from your real data. This is where you see whether it holds, before the budget is spent.
Systems integration, permissions, logging, spend limits, fallback plan. These are what decide whether a pilot becomes a system or stays a demo.
Quality monitored over time, models kept current, new use cases added. An agent is not delivered: it is kept alive.
We are small by choice and specialised by necessity. A light structure decides quickly, has no layers to cross and no consulting machine to justify: on an AI project, that is the whole advantage.
We do not sell everything to everyone. Agent systems, enterprise software, integration and infrastructure: four capabilities, handled by the same people from start to finish.
We use AI in our own daily work, not only in our clients'. It is why two people deliver like a team, and why they genuinely know the tools they are recommending to you.
Around the studio there is a network of professionals and firms we have worked with for years. On large projects the team is assembled to fit, and afterwards it goes light again: you pay for the complexity the work needs, not the complexity we happen to carry.
We are not tied to a single model vendor. We choose per use case, we measure, and we keep the exit door open.
The person who answers the phone is the same one who writes the code and keeps the infrastructure up. When a project grows we call in the collaborators we have worked with for years — we do not hire to fill an org chart.
Meet the studio
Describe the process you would like to automate. In half an hour we will tell you whether an agent makes sense — and if it does not, we will tell you that too.
