Turn enterprise knowledge into
an on-demand private AI
— confidential data never leaves
MAQ Alishan brings together enterprise knowledge scattered across documents, forms, SOPs, and databases into a private AI assistant you can query through conversation. RAG knowledge base × knowledge graph × permission auditing — all running on your own AI server, powered by NVIDIA RTX PRO 6000. No token anxiety no matter how much you use, and no traffic limits.
Named After Taiwan's Alishan
Alishan is one of Taiwan's most iconic mountain landmarks: during the Japanese colonial era, a forest railway built in 1912 to transport thousand-year-old cypress timber wound its way up the mountain in switchbacks, carrying Alishan from a logging hub to worldwide renown. A century later, it's known for its sea of clouds, sunrises, high-mountain tea gardens, and slow-ripened coffee beans. We named this enterprise knowledge appliance Alishan — because an organization's knowledge, like the mountain's resources, deserves to be carefully accumulated, allowed to settle, and passed down from one generation to the next. And because, like us, its roots are in Taiwan.
Alishan Coffee
High altitude and wide day-night temperature swings let the coffee cherries ripen slowly. One of Taiwan's leading specialty coffee regions, it has earned strong results in international competitions in recent years.
Alishan High-Mountain Tea
Tea gardens wrapped in clouds and mist year-round produce Taiwan's signature high-mountain oolong — rich, lingering, and unmistakable to connoisseurs.
The Century-Old Forest Railway
The Alishan Forest Railway, opened in 1912, transformed from a timber-hauling line into a world-renowned mountain scenic railway, still carrying travelers up to watch the sunrise today.
Adopt Generative AI to Drive Growth and Innovation
Cloud AI services are convenient, but they can't deliver the two things enterprises care about most — data security and long-term cost. MAQ Alishan brings the entire generative AI knowledge stack onto your own AI server: confidential data never leaves, hardware is a one-time purchase with unlimited use, and the knowledge base grows with your business.
Access Through Conversation
Ask in plain language and get answers on company policy, product specs, SOPs, and historical records — no more digging through documents or asking coworkers.
On-Premises Deployment, Data Stays In-House
The model and knowledge base run entirely on your own server. Sensitive data never touches the cloud, meeting internal control and compliance requirements.
Faster Knowledge Transfer
New hires get complete guidance, veterans get interrupted less, and organizational knowledge is no longer locked in a few people's heads — handoffs and training done right the first time.
Highly Extensible
The knowledge base grows with your business needs — adding departments, documents, or assistant roles never means starting over.
RAG + Knowledge Graph
Combines Retrieval-Augmented Generation (RAG) with a knowledge graph, so answers are structured and sourced — not invented by the model.
Rigorous Access Auditing
Built-in role-based access control and operation auditing — who can ask what, who has seen what, fully logged and traceable.
MAQ Alishan Product Architecture
A three-layer architecture — from knowledge base management to prompt design to AI assistant deployment — each layer fully controllable by your own team.
Accountable Knowledge Base Management
Document upload, chunking, vectorization, version control, and source attribution — knowledge that's owned, updatable, and retirable.
User-Friendly Prompt Management
Design and adjust prompts through a template-based interface — no coding required, so business units can maintain their own Q&A style.
Flexible AI Assistant Configuration
Create dedicated AI assistants for HR, IT, sales, and other departments, each wired to its own knowledge base and permissions, for targeted service.
Use Cases
One Alishan appliance, loaded with different knowledge bases and assistant roles, serving the entire organization.
HR Management
Employees self-serve on leave, attendance, benefits, and policies, plus onboarding guidance for new hires; HR can run resume matching against open positions.
IT Help Desk
Common troubleshooting, system how-tos, and account/permission request workflows — let AI field first-line tickets.
Sales Support
Instantly look up product specs, quotes, competitor comparisons, and talking points; meeting notes are auto-summarized to shorten the sales cycle.
Legal & Compliance
Ask about internal policies, contract clauses, or regulatory points anytime, with sources cited, reducing the risk of human misinterpretation.
Meeting & Document Summaries
One-click summaries of long documents and meeting transcripts, capturing key points and action items, with knowledge stored for future reference.
R&D & Knowledge Retrieval
Cross-database search across technical documents, patents, and past project experience, so engineers avoid reinventing the wheel.
Enterprise AI Appliance Server
Hardware, model, and knowledge base platform, all in one — no assembling GPUs, tuning CUDA, or wiring up vLLM yourself. Power it on and it's a private AI host ready to run a 120B model. The numbers below are measured on real hardware, not spec-sheet theoreticals.
NVIDIA RTX PRO 6000 Blackwell
96GB of VRAM · Runs a full 120B model on a single card
Server Platform with IPMI Support
Multi-core processor · Liquid cooling · Remote management
Pre-Loaded Alishan Software Stack
gpt-oss-120B · RAG knowledge base · Assistant & permissions admin panel
// Test environment: Ollama + gpt-oss-120B, single NVIDIA RTX PRO 6000 96GB, ~3ms direct local network latency.
Buy Once, Use Without Limits
Cloud APIs charge by usage — more locations and more users means your monthly bill grows linearly. A self-hosted server is a fixed cost, amortized over usage, with marginal cost approaching zero, and your data stays entirely in-house.
Cloud API Subscription
- Billed per token — the more you use, the more it costs
- Traffic caps and rate limits apply
- Sensitive data must go to a third-party cloud
- Cumulative cost keeps rising over time
MAQ Alishan Self-Hosted Server
- No more token anxiety — cost stays the same no matter how much you use
- No traffic caps, no rate limits
- Confidential data never leaves, meeting internal control and compliance needs
- Fixed cost amortized over time, marginal cost approaching zero
Turn Enterprise Knowledge Into an On-Demand Private AI
From hardware configuration and model deployment to knowledge base setup and permission integration, MAQ builds your dedicated AI knowledge appliance end to end — confidential data never leaves, and token anxiety is a thing of the past.