Features
Capabilities, Observability and Roadmap
What the platform does today, what it lets you see while it runs, and what is planned but not yet built.
Benefits
Why Use a Distributed AI Cluster?
Reuse Existing Hardware
Use available computers instead of requiring a single high-end machine.
Flexible Capacity
Additional nodes can contribute additional compute resources.
Centralized Control
Manage the cluster through one control plane.
Hardware Awareness
Track CPU, GPU, RAM and VRAM characteristics of participating nodes.
Fault Awareness
Detect unhealthy or disconnected nodes.
Developer-Friendly API
Applications can consume the AI service through an API rather than managing individual machines.
Benefits depend on workload characteristics, network performance, hardware configuration and runtime behaviour.
Use Cases
What Can It Be Used For?
Personal AI Infrastructure
Run AI workloads across your own computers.
AI Development
Create a private inference environment for experimentation.
LLM Applications
Provide an API backend for AI applications.
Distributed AI Research
Experiment with distributed inference and heterogeneous compute.
Local Compute Pools
Coordinate idle machines within a trusted network.
Developer Labs
Build and test distributed AI infrastructure concepts.
Security
Security & Trust Model
Dashboard │ │ Authenticated API ▼ Controller │ │ Authenticated/Secure Node Protocol ▼ Node Agents
All tokens, keys and URLs shown across this website are placeholders. Real secrets must never be committed or published.
Observability
Know What Your Cluster Is Doing
Node health
Hardware
CPU / RAM / GPU / VRAM
Jobs
QUEUED / RUNNING / COMPLETED / FAILED / CANCELLED
Events
Node joined · Node disconnected · Job started · Job completed · Node drained · Recovery triggered
event stream
Reference
Node States
| State | Meaning |
|---|---|
| ONLINE | Node connected and healthy |
| READY | Node available for work |
| BUSY | Node currently executing work |
| DRAINING | Node finishing work and accepting no new work |
| OFFLINE | Node disconnected |
| UNHEALTHY | Node failed health checks |
Stack
Technology Stack
Node Layer
Controller
Inference
Frontend
Deployment
Benchmarks
Benchmark Your Own Cluster
No benchmark numbers are published here. These metric cards are placeholders that can consume real measurements from your own cluster.
Time to First Token
-- ms
Total Latency
-- ms
Tokens / Second
-- tok/s
Active Nodes
--
Total VRAM
-- GB
throughput chart · awaiting data
Benchmark results depend on hardware, network, model, quantization, workload and cluster size.
Roadmap
Roadmap
Completed
- ✓ Node Agent
- ✓ Hardware detection
- ✓ Node registration
- ✓ Heartbeats
- ✓ Telemetry
- ✓ Cluster Controller
- ✓ Scheduling
- ✓ Distributed inference integration
- ✓ API layer
- ✓ Dashboard integration
- ✓ Fault / recovery testing
Future (planned, not completed)
- ○ More operating systems
- ○ Improved scheduling
- ○ More inference runtimes
- ○ Advanced observability
- ○ Automated model distribution
- ○ Better network optimization
- ○ Production-grade authentication
- ○ Expanded benchmarking
- ○ Additional model formats
Build Your Own AI Compute Cluster
Connect your machines. Deploy the Node Agent. Start the controller. Build a distributed AI environment around the hardware you already have.