EPIC EDGE PLATFORM

How many consoles does it take to run your infrastructure?

Epic Edge Platform is the multi-provider, white-label Cloud Management Platform for governing technologically different environments through a single, consistent experience. SentinelForge AI lives inside it: the module ecosystem that builds, watches over and optimises your infrastructure. All within your own perimeter.

The model proposes. The automation engine executes.

ON-PREMISE · SOVEREIGN · OPEN-SOURCE · AI-NATIVE

Want to see the platform on your own stack?

We run a live demo on your real use case

Talk to the Epic Edge experts

Tell us about your cloud challenge—we'll get back to you within one business day with the best solution and, if you'd like, a live demo.

One platform, three layers.

The Cloud Management Platform is the single point of access: a white-label portal that users reach with their corporate credentials, showing them only what is theirs. Beneath the interface sits a multi-provider middleware that normalises each backend's APIs and exposes a uniform capability catalogue — adding a technology means extending the platform, not rewriting it.

SentinelForge AI lives inside the platform: an ecosystem of AI modules that does not merely display the infrastructure, but builds it, watches over it and optimises it. The modules never talk to the infrastructure directly. They produce a plan, which the middleware executes in a fully traced way through AWX and Ansible.

THE PLATFORM

Cloud Management Platform — white-label portal, SSO, catalogue, consumption

↓
THE INTELLIGENCE

SentinelForge AI — EdgeForge (Forge) · CephSentinel (Protect) · CloudSentinel (Optimize)
↓
THE FOUNDATION

Multi-provider middleware · On-premise LLM · AWX + Ansible · Playbooks in Git · Audit trail
↓
MANAGED DOMAINS

OpenStack · Proxmox VE · VMware · Ceph · Kubernetes · hyperscalers

From assessment to operation in four steps.

01 — Environment assessment

We analyse your real infrastructure — OpenStack, Proxmox VE, VMware, Ceph, Kubernetes — and identify where manual work costs you most. You don't start with the whole platform. You start there.

02 — Platform activation

We install the CMP inside your perimeter and connect your existing environments through the middleware. Access federates with the identity systems you already use (LDAP, Active Directory, SAML, OpenID Connect), and observability builds on the Prometheus and Grafana you already run — no proprietary agents inside your machines.

03 — A pilot on one module, fully supervised

We activate the AI module with the fastest return on your environment. At this stage the system observes, correlates and proposes, but nothing is executed without explicit approval. The scope of automation is agreed up front.

04 — Autonomy and day-to-day operation

Policies widen as operations prove themselves, configurable per environment and per customer. Epic Edge support covers the entire stack — OpenStack, Ceph, Kubernetes and the AI engine — not just the module.

The platform, and what lives inside it.

01 · The Cloud Management Platform — ROLLING OUT

The governance layer: a single white-label interface for managing OpenStack, Proxmox VE, VMware and, where needed, the hyperscalers. Administrators see available and committed capacity for every environment along with per-customer consumption; end customers see only their own resources and how much of their plan they have used. The page structure is identical across environments: the content changes, the way you read it does not. Full white-label customisation is currently being released.

02 · The multi-provider middleware

The layer that makes it possible to present different technologies in the same way: it translates and normalises each backend's APIs, exposes a uniform capability catalogue and hands execution to AWX and Ansible. This abstraction is what allows a provider to be added or replaced without rewriting either the assistants or the interface — and what means users never need to know which technology they are working on.

03 · EdgeForge AI — FORGE · AVAILABLE

The module that builds. Infrastructure provisioning in natural language on OpenStack and Proxmox VE: users describe what they need, the system designs the architecture, sizes the resources and estimates the monthly cost. Once approved, Terraform creates the resources and Ansible installs the software from validated, versioned templates. Live today, integrated into the OpenStack Skyline dashboard; VMware support is currently being integrated.

04 · CephSentinel AI — PROTECT · PRIMO PILOT

The module that watches over storage. Continuous monitoring and operational automation for Ceph fleets, including multi-cluster estates and remote clusters behind restrictive firewalls. The heart of it isn't the chat — it's the fleet view, a single screen answering "where do I need to look today". It correlates signals that individual thresholds would never connect: a network problem on a rack and OSD flapping on that same rack are one incident, not four alerts.

05 · CloudSentinel AI — OPTIMIZE · IN DEVELOPMENT · 2027

The module that optimises. Predictive oversight of the physical infrastructure: anomaly detection, failure prediction hours or days in advance, and automated root cause analysis. It includes an energy efficiency component that consolidates workloads without service interruption and automatically produces consumption and emissions reporting per service, in a format designed to align with European sustainability frameworks.

06 · The shared foundation

The modules are not separate products built on different technologies: they all rest on the same base. An on-premise language engine (open-weight models via Ollama), deterministic execution handled separately by AWX and Ansible, playbooks versioned in Git and tied to the signal that generated them, a complete audit trail, and autonomy policies across three levels — notify only, execute after approval, or execute automatically for low-risk actions alone. Learn it once, it applies across the whole platform.

The model proposes.
The automation engine executes.

The AI assistants never interact with the infrastructure directly: they generate a structured plan, which is passed to a separate, deterministic execution engine. That decoupling is what makes audits, dry runs and revocation before execution possible — and what makes every action defensible under scrutiny: what was executed, on which environment, in response to which signal, with whose approval. In regulated contexts, that is a requirement, not a nice-to-have.

Datasheets and overviews in PDF — the real status of every component, not promises.

Product documentation

Epic Edge Platform ENG

EdgeForge AI ENG

SentinelForge AI ENG

CloudSentinel AI ENG

CephSentinel AI ENG

Why Epic Edge

How this differs from a conventional CMP

Established multi-cloud management platforms exist, but nearly all of them hand conversational automation and analysis to external services. The difference here isn't the number of features: it's that the artificial intelligence runs entirely on-premise — the model never leaves your organisation's perimeter — on a technology chain that is fully open source and reversible. That combination is what makes AI automation compatible with sovereignty requirements a commercial CMP, by its very construction, cannot satisfy.

The platform comes out of hands-on experience designing and running mission-critical OpenStack, Ceph and Kubernetes infrastructure. Not a laboratory exercise: an architecture designed by people who have lived with those constraints in production.

25+ years of operations on real production infrastructure — real data centres, real SLAs, real enterprise customers

EdgeForge AI is live on Skyline today: the platform isn't a roadmap, half of it is already running

On-premise LLM: no data, no telemetry and no credentials leave your network

Fully open source stack, with no restrictive proprietary licences and no exit costs

White-label: the service reaches the end customer under the operator's brand, not the technology vendor's

A backend that evolves independently: adding or retiring a technology doesn't change the user experience

Every action, manual or assisted, leaves a complete and audit-defensible trail

Who it is for

Cloud & Service provider — a single white-label front end to offer your own customers, without multiplying consoles and training
Telco & ISPs — multiple environments across multiple data centres, where no single overview exists today
Public sector — data sovereignty, reversibility and full traceability of infrastructure changes
Finance & Banking — every infrastructure change leaves an audit trail
Enterprises multi-datacenter — infrastructure teams undersized against internal demand
MSPs and storage teams — CephSentinel AI can be adopted on its own, across fleets belonging to different customers

Where control breaks down

Fragmentation

Every platform has its own console and its own conceptual model: training costs multiply with the number of technologies in the estate.

Slowness

Requirements gathering, sizing, provisioning and configuration take days, not minutes — and the outcome varies depending on who runs the deployment.

Dependency

The AI automation available on the market requires data, telemetry and credentials to leave your perimeter: incompatible with the sovereignty requirements of public sector, telco and finance.

The problem

Most organisations today do not run a single cloud. They run a private platform built on open source, a traditional virtualisation environment, distributed storage, and sometimes resources with international public operators. Each of these environments comes with its own console, its own language, its own access model and its own way of measuring consumption.

The result is fragmentation: training costs multiply, operational errors creep in, reporting becomes difficult and — above all — overall control is lost. On top of that, building environments, keeping them healthy and optimising their consumption are three separate activities, handled by separate tools, with the knowledge concentrated in a handful of people.