Product · Technology
Our App Framework
Every agent (Operations, Customer-Facing, Knowledge) is built on the same App Framework. Azure-native, not Azure-locked.
Identity-governed everywhere: no static credentials, least privilege per operation, secrets in a project-scoped vault.
Vector storage, structured memory, session context, inside your environment, with right-to-erasure traceable end-to-end.
The framework runs each agent on a managed runtime, selects the best model per step, enforces HITL and approval workflows, and applies the six integration patterns per system.
Decision logging, throughput metrics, escalation rates, drift monitoring, structured for operators, IT and audit.
Deployable to on-premise, private or air-gapped environments. The platform layer is pre-built for Azure; delivery on Google Cloud or AWS is supported on request and scoped as additional engineering.
Two topologies
Every Machines Like Me deployment lives inside the customer’s environment. Two topology options cover the full range of enterprise security postures. The choice is a design decision at kickoff, not a product tier.
Deployed inside a client-owned Azure subscription, with in-region inference.
Recommended where the engagement meets the regulatory baseline (EU AI Act, GDPR) without additional residency constraints.
Deployed inside private infrastructure, on-premise or sovereign cloud.
Recommended for public sector, defense, healthcare and regulated financial services, where residency or sovereignty requirements are elevated.
Sovereignty & on-premise
Besides our cloud-based setup, we deploy on-premise and locally hosted, for public institutions, defense, and any organization that requires complete control over its knowledge and data.
The full platform runs on your own hardware, inside your own data center, locally hosted, locally operated.
No dependency on any single model provider. Open-source and European models available where sovereignty requires it.
Air-gapped and restricted-network operation supported, nothing leaves the environment, by design.
Public institutions, defense and regulated industries: deployments designed around sovereignty requirements, not adapted to them afterwards.
Reference architecture
The same components, the same boundaries, the same control planes, adapted at the edges to your identity system, integration surface and compliance posture.
A single authenticated entry point over HTTPS. Application services are not exposed directly.
Your identity system governs every call. Least privilege by default.
Memory and context inside your subscription, in your region.
Managed container runtime, isolated per agent and environment.
In-region model hosting, providers chosen per engagement.
Decision logs and metrics; retention agreed per project.
Adapters in source control, versioned like code.
HITL and thresholds enforced across every plane above.
The shared platform layer is real and growing, and not every component is unified yet. Reuse reduces the scope of a project; it does not reduce it to configuration alone.
Concrete evidence
The Technology Platform is agreed with your IT in the project specification before go-live, verifiable by audit after, in place from day one.
Get in touch
Book 30 minutes with our team. Walk us through your process. We'll show you where AI agents can make a difference.
