One Edge Platform. Every site. From containers to Edge AI.
Edge should be easy. Deploy, run, and secure containerized applications and AI models across hundreds or thousands of distributed sites, centrally managed from the cloud, fully autonomous on-site. If you’re looking for a fully automated, small-footprint container management platform — without the complexity of Kubernetes — you’ve come to the right place.
One edge computing platform, two components
The Avassa Edge Platform gives you a single, consistent way to operate applications and AI at the edge. It’s made of two integrated components that work as one product.
Together, they extend the cloud-native tooling your teams already know all the way to the physical edge, without retrofitting tools that were never built for that purpose.
Numbers that speak for themselves
10,000+
edge sites tested for management from a single Control Tower.
80%
of custom software at the physical edge will run in containers by 2028 (Gartner, 2023).
Zero
Kubernetes clusters to build, patch or babysit.
The Avassa Edge Platform Architecture
From a single Control Tower, you onboard to any number of sites with mixed hardware (Intel, ARM, GPU). On each site, the Edge Enforcer runs the workloads, discovers local devices, and keeps everything alive — online or offline.
Communication is secure and resilient by design, using secure call-home and asynchronous updates that tolerate poor or intermittent links.
Everything you need to operate the edge. Built in.
No add-ons to assemble. The platform ships the edge-native services that other tools leave you to bolt on.
Application lifecycle
Containers + VMs, declarative deployments, rolling & canary updates.
Edge AI management
GPU/device discovery, model-serving endpoints, and model lifecycle.
Offline autonomy
Self-healing clusters that keep running through disconnection.
Observability
Real-time application and site health; fast time-to-restore.
Intrinsic security
Zero-trust, encryption, micro-segmentation, distributed secrets, key rotation.
OS upgrades
Remote, policy-driven upgrades across e.g. Debian/Ubuntu, Red Hat, Wind River, Yocto or a Linux of your choice.
Multi-tenancy
Isolated workloads; let third-party teams share infrastructure safely.
Open integration
OCI-compliant containers, full REST API + CLI, reuse your existing CI/CD, OTEL & webhooks.
Control Tower. Centralized, automated edge management.
The Control Tower manages your distributed applications, AI models, and VMs through one UI and a complete REST API. It’s where your platform and application teams plan, deploy, automate, and observe everything that runs at the edge.
Central and remote management of edge sites
- Group distributed hosts into sites and assign resources per site with full control.
- Proactively monitor applications, hosts and the health and status across large fleets.
- Run predictive maintenance and roll out new OS versions without taking applications down.
- GPU and leaf device discovery and management.
Full application lifecycle management
- Define applications as sets of containers and AI models, with their networking, configuration and resource needs.
- Target deployments declaratively by label: location, hardware, available GPU or attached device.
- Automate deployment, upgrade and roll back across thousands of sites, including rolling and canary releases.
- Manage secrets centrally and distribute them securely, on a need-to-know basis. Segment edge site networking according to ISA-95 or Purdue.
Edge Enforcer. Secure, autonomous edge orchestrators.
The Edge Enforcer is a single lightweight agent installed on each host. It delivers everything a site needs to run on its own, with a remarkably small footprint, and the Control Tower upgrades it automatically. This is what removes the burden of maintaining a complex software stack at every location.
Local autonomous edge site management
- Automatic call-home, site-local clustering and state replication.
- Built-in DNS, application networking and a local container registry so workloads can reschedule across hosts.
- Automatic and autonomous OS upgrades.
Autonomous application scheduling
- Schedules containers and VMs onto the right hosts on request from the Control Tower.
- Built-in fail-over, migration and self-healing, no cloud connection required.
- Discovers local GPUs, cameras and sensors and makes them available to applications.
- Site-local pub/sub bus and secrets vault, maintained even while offline.
Zero-trust security, by default
Edge hosts can sit in exposed locations and untrusted networks. Avassa assumes the worst and protects every layer.
Locked-down hosts
At boot, a host needs an external key to unseal its data; untrusted hosts stay locked, traffic blocked by default.
Distributed secrets
Automatic provisioning and removal of credentials per site.
Protected data & traffic
Encryption at rest, micro-segmented application networking (VXLAN), communications over WireGuard.
Policy & compliance
Access control, logging, site quarantine and automated key rotation.
A practical Edge AI platform:
from model to inference at scale
AI models rarely run alone. They live alongside protocol adaptors, analytics, and APIs at the edge. The Avassa Edge Platform deploys the whole bundle, automatically places models and inference engines on hosts with the right GPU, and configures the serving endpoints. That way, trained models become production deployments across every site. Combine it with the MLOps tooling you already use to unlock Edge AI for real-time analytics at scale.
Explore the SolutionWhy we didn’t build on Kubernetes
Kubernetes was built for the cloud where abundant compute, stable networks, and central management can be assumed. The edge is different: many small sites, unreliable links, no local operations team. Retrofitting Kubernetes would mean clusters, add-ons, and specialists at every edge site.
The Avassa Edge Platform takes the other path: one lightweight, all-in-one platform that self-updates, runs autonomously, and lets your teams focus on applications instead of infrastructure — while still giving you Kubernetes-level automation.
Avassa in action
Avassa has provided us with a platform that makes it possible to manage and evolve these applications efficiently across distributed environments, supporting both our development work and future scaling ambitions.
Learn more about Germán Sacramento
Operating applications at the distributed edge
Want the technical detail? Download the Avassa Edge Platform data sheet and take it with you.
Get the data sheetLooking for our platform documentation?
Deep-dive into how the platform works, from architecture to the full application specification.
Frequently asked questions
What is an edge computing platform?
It’s a software solution for running and managing application workloads and AI at distributed sites close to where data is produced, rather than in a central cloud. It allows you to manage and operate the full lifecycle of edge workloads from a central component with full automation.
How does a distributed edge platform work?
A central control plane (Control Tower) holds desired state, policies and application specifications; a local runtime at each site (Edge Enforcer) receives that state and enforces it independently. Deployments are declarative, and sites are matched by labels, so one definition can automate deployments to hundreds of locations.
How is it different from cloud-based container orchestration?
Purpose-built for the edge: integrated and zero-touch, lightweight footprint, canary rollouts, and full autonomy in disconnected environments — no retrofitting Kubernetes.
Does Avassa require Kubernetes at every edge site?
No. The Edge Enforcer is self-contained; scheduling, networking, secrets and storage included, so there’s no cluster per site and no Kubernetes needed locally. It coexists with Kubernetes where you already run it.
How many edge nodes does it support?
Tens of thousands per Control Tower; scale horizontally with multiple Control Towers into the hundreds of thousands.
Can Avassa operate when an edge site loses central connectivity?
Yes. Each site holds its own state, images, secrets, and config, and keeps scheduling, restarting, failing over, and serving traffic on its own. Control Tower reconciles automatically when the link returns.
How does Avassa handle edge management and orchestration across many sites?
Declarative application specifications with label-based site matching, zero-touch onboarding, staged and canary rollouts with automatic rollback, and fleet-wide plus per-site logs, metrics, and alerts in a single pane.
Can the platform run containers, virtual machines and Edge AI workloads?
Yes: containers natively, VMs alongside them for legacy or vendor-supplied software, and GPU-accelerated inference workloads, all through the same specifications, rollout and observability model.
See the Avassa Edge Platform for yourself
Book a demo at a time that suits you, or start a free trial.
Request free trialInterested in learning more about our technology?
Send us a message and we’ll let you know how we can help.