Platform Foundation

The Unified, Privacy-Audited AI Architecture

AliAI is the foundational intelligence backbone powering AliCode, AliWrite, AliChat, and AliDoc. Rather than messy fragmented API integrations, AliAI provides one centralized, privacy-conscious model orchestrator.

Local Model Auto-Discovery

Instant Integration with Ollama, ONNX & Localhost Endpoints

Have Ollama or LM Studio running on your computer? AliAI automatically detects active local endpoints at localhost:11434 or custom ports.

Your local models (Llama 3, Mistral, Qwen, DeepSeek) are immediately available in every AliDesk application with zero manual configuration.

AliAI — Local Discovery Panel
[DISCOVERY] Probing http://127.0.0.1:11434... CONNECTED
[MODELS DETECTED]
• llama3.3:70b-instruct-q4_K_M (Active)
• deepseek-coder-v2:16b
• mistral:7b-instruct-v0.3
STATUS: 100% Offline AI Enabled
Privacy Guardrails

Strict Notice Before Any Prompt Leaves Your Machine

Other software silently streams user context to third-party endpoints. AliAI enforces an uncompromising privacy boundary:

No document text, code repository snippet, or chat prompt ever leaves your machine without an explicit, visual user prompt. You inspect token estimates, destination URLs, and provider disclosures before transmission.

AliAI Privacy Boundary Modal
External Network Notice: You are about to send 412 tokens to Google Gemini Pro.
FAQ

AliAI Architecture Questions

No! API keys for OpenAI, Anthropic, or Google Gemini are stored strictly on your local computer in an encrypted vault partition (via AliVault). They are never uploaded or synced to Elevoire servers.