MCP and A2A — The Protocol Layer of the Agent Ecosystem
The 2026-07-28 MCP revision makes the protocol stateless and hardens OAuth; A2A 1.0 brings signed Agent Cards, multi-tenancy, and three protocol bindings…
AI engineering for production · Made in Germany
We engineer AI systems to perform reliably in production.
From agents and retrieval to copilots, we take responsibility for architecture, implementation and operation. Evaluations, guardrails and human review are part of the system design — not an afterthought.
Short iterations · rigorous standards
Integration via MCP & A2A
Human review built into the design
EU AI Act & GDPR from the outset
fig. 01 — agent-system architecture Blue: structure · cyan: data · gold: human
Capabilities
AI engineering across the full system.
Our work begins where models meet real data, processes and compliance requirements — and the resulting system must be measurable, controllable and operable.
Cooperating peer agents
Stateful agents coordinate over A2A. Simulated counterparts make complete journeys testable before every real system is available.
View detailsTraceable RAG & retrieval
Hybrid search, reranking and graph methods tie answers back to reliable sources — rather than merely sounding plausible.
View detailsProven solution building blocks
ServiceNow agent, Dynamics mapping kit, Antlet.OS, FounderOS and Copilot company agents give projects a proven starting point.
View detailsAI for specific workflows
Document processing, process assistants and knowledge access are designed around clear tasks and verifiable outcomes.
View detailsFrameworks chosen for the task
Microsoft Agent Framework, LangGraph, Semantic Kernel or Copilot Studio: the choice follows your environment, integration and operational needs — not the latest trend.
View detailsOpen engineering notes
Since 2023, we have published the technical decisions, measurements, sources and limitations behind our work on AI systems.
Read the notesFrom our engineering work
What is changing, what holds up and where the limits are.
Subagents vs. Peer Agents — Two Working Modes of Multi-Agent Systems
Subagent and peer-agent architectures are both described as multi-agent systems, yet differ in state, lifecycle, cost, and trust model. This article…
Simulation-First Agent Engineering
End-to-end agent journeys ultimately trigger state changes in government systems, ERPs, and partner APIs — systems you cannot freely test against. This…
From a concrete use case to a realistic production plan.
In a free initial consultation, we assess your use case: what data is available, which risks need to be managed and what a realistic path to production looks like.