ZDX AI
ActiveOn-Device AI · MCP · Tools · Context Engine
ZDX AI
Full control over your AI runtime.
ZDX AI combines local inference, cloud AI providers, and a programmable tool system into a single runtime. Control how your AI thinks, what it sees, and how it executes — all on your own terms.
On-Device AI Runtime
Run AI models locally on Windows and Linux with full control over execution, performance, and privacy.
Cloud AI Integration
Connect to leading AI providers including OpenAI, Anthropic, Google Gemini, and xAI when you need more power.
Hybrid Execution
Switch between local and cloud models dynamically based on performance, cost, or workload.
Custom MCP Server
Karma: mediumAlign: 50Run and configure your own MCP server for tool execution, orchestration, and structured AI interactions.
Tooling System
Karma: mediumAlign: 50Extend AI capabilities with real tools — APIs, scripts, and integrations — executed through controlled pipelines.
Document Ingestion
Karma: mediumAlign: 50Ingest and process documents locally to provide contextual grounding for AI responses.
Context & Memory Engine
Mini Context Model (~80MB)
Lightweight local model used to sort, score, and prioritize context before passing to the main model.
KV Cache Optimization
Efficient context reuse using KV caching to reduce latency and improve response performance.
Context Scoring
Dynamic ranking of relevant data to ensure high-signal inputs reach the model.
On-Device Training
- • On-device fine-tuning (Windows / Linux)
- • Local dataset control
- • No forced cloud training
- • User-controlled model updates
Supported AI Providers
Bring your own API keys and switch providers at any time.
Build and run AI your way
Local, cloud, or hybrid — fully under your control.