AT-PAIB Enterprise Private AI Computing Server

The Genuine Enterprise-grade Private AI Engine
100% On-premises Computing
Military-grade Security Compliance

Product Overview

Product Introduction

In an era where data privacy is paramount, the AT-Vibe AT-PAIB Enterprise-grade AI Server provides Hong Kong businesses with a powerful and secure on-premises AI solution. As a 100% locally deployed enterprise private AI server, the AT-PAIB eliminates reliance on public cloud AI by deploying top-tier Large Language Models (LLMs) and Vision Language Models (VLMs) within your office. It transforms your company documents (PDFs, Word, Excel, PPT, Emails, Images) into an "enterprise brain" that is searchable, citable, and controllable. Combined with an enterprise-grade architecture, it enables strict role-based access control and document retrieval, perfectly integrating your business secrets with AI intelligence to fully unleash enterprise productivity.

Core Advantages

Data Sovereignty & On-Premises Computing:The server is located within your enterprise office. All computing is completed internally, and data never leaves the corporate network. It is operable even with the network physically disconnected and complies with Hong Kong PDPO and ISO27001 standards.
Military-grade Data Protection: Utilizes LUKS low-level full-disk encryption. Even if the hardware is lost or stolen, the data cannot be deciphered.
Instant Response: Get immediate answers to your questions, with the original source documents provided.
Built-in LLMs & No Token Fees: Pre-loaded with mainstream models such as Qwen 2.5, Qwen3, Llama 3.2, and Gemma 2 at the factory. A one-time investment for perpetual use—no hidden costs.
Multi-User Simultaneous Access: Supports team collaboration to integrate seamlessly into your daily operations.

Does your business also face these problems?

AT-PAIB provides a one-stop solution to all the above problems.

Product Series

Tailor-made for enterprises of different sizes

AT-PAIB-01 (Basic Edition)

Designed for 10-50 employees / Single department

Enterprise Internal LLM Chatbot (Instant Q&A)
Automatic document synchronization to AI knowledge base
Intelligent responses with document pinpointing (links/sources included)
Multi-format document understanding (PDF/Word/Excel/PPT/Email)
Image/scanned document understanding (OCR)

Supported Models: Qwen2.5 / Gemma2 / Llama2

Maximum Computing Level: 14B

Concurrent Users: 15 (based on 7B text model computing)

AT-PAIB-02 (Advanced Edition)

Designed for 50-200 employees / Multi-department access control

Includes all features of AT-PAIB-01, plus:

Hierarchical document management (department/rank/project level permissions)
Permission-based protection: only references content the user is authorized to access
AI-powered risk alerts for company emails
Support for more advanced models

Supported Models: Qwen3 / Qwen3.5 / Gemma2 / Phi / Llama3.2

Maximum Computing Level: 32B

Concurrent Users: 30 (based on 7B text model computing)

Core Advantages

Private AI Computing Terminal

Data Sovereignty & On-Premises Computing

100% Localized Deployment: The AI server is installed within your enterprise office, all computing is completed internally, and data never leaks.
Operable with Network Disconnected: Even if external connectivity is cut off, AI services continue to function normally, meeting the highest standards of privacy compliance.
Military-Grade Data Protection: LUKS low-level full-disk encryption ensures data cannot be deciphered even if the hardware is lost.
Enterprise-Grade Hardware Architecture: Powered by the AMD Ryzen AI 9 HX PRO processor with a dedicated NPU, delivering 100+ TFLOPS of computing power. Features 4 hot-swappable drive bays with RAID 0/1/5/10 fault tolerance mechanisms to ensure data security and system stability.

Pre-installed LLMs – No Need to Purchase Additional Tokens

Leading Open-Source Models Pre-loaded: The system comes pre-installed with top-tier LLMs like Qwen 2.5, Qwen3, Llama 3.2, and Gemma 2 right out of the box. No additional API fees or token licensing costs are required.
One-Time Investment, Lifetime Use: After the hardware purchase, internal employees can utilize AI capabilities unlimited times. Future model releases can be downloaded and imported by the enterprise independently.
Flexible Model Replacement: Businesses can freely swap or fine-tune models based on operational needs, maintaining full control over their AI core.

Five major enterprise application scenarios

Flexible application in various industries and scenarios

Instant Q&A with Source Citations

Employees can ask questions in natural language, and the AI instantly retrieves information from the enterprise knowledge base. Responses are accompanied by links to the original source documents (PDF/Word/Excel, etc.), making it easy to verify and trace information while reducing misunderstandings.

Application Results:

Reduce search time by 90% for more precise decision-making.

Automated Learning

A Constantly Evolving Knowledge Base

As your business regularly uploads new contracts, quotes, and technical documents, the system automatically converts them into vector data and integrates them into the knowledge base. No manual sorting required—the AI continuously learns as the document volume grows, steadily improving response quality and retrieval accuracy.

Application Results:

The knowledge base never goes out of date—the more you use it, the smarter it becomes.

jimeng-2026-03-12-9946-将图片修改为16_9比例，通过智能补全边缘内容保证画面自然过渡，避免简单拉伸变形....png

Email Behavior Screening and Compliance Monitoring

The AI automatically analyzes internal email content to identify potential risks (e.g., confidential information mistakenly sent externally, inappropriate language that may lead to complaints). When anomalies are detected, administrators are immediately notified for rapid intervention.

Application Results:

Reduce data leakage and public relations risks while ensuring communication privacy through complete on-premises analysis.

jimeng-2026-03-12-7353-将图片修改为16_9比例，通过智能补全边缘内容保证画面自然过渡，避免简单拉伸变形....png

24/7 AI Customer Support – Integrated

with WhatsApp Business

Seamlessly integrated with WhatsApp Business, the AI customer support provides 24/7 real-time responses to customer inquiries. All conversations are based on the company's product catalog, policy documents, and FAQs, ensuring responses are 100% compliant.

Application Results:

Reduce customer service costs by 90%, with chat logs securely stored internally to continuously optimize the AI model.

jimeng-2026-03-12-5823-将图片修改为16_9比例，通过智能补全边缘内容保证画面自然过渡，避免简单拉伸变形....png

Assisted Web Search

(Optional Internet Connectivity)

After administrators authorize internet access, the AI can simultaneously retrieve information from the internal knowledge base and public online sources, providing source links with responses. Administrators can set permissions to allow or block specific websites, ensuring employees only access trusted sources.

Application Results:

Balances information security with query efficiency while preventing access to malicious websites.

Optional Features: Intelligent Visual

Recognition & Multi-Tiered Access Control

Equipped with VLM vision models to extract data from receipts and contracts via photo capture and convert it into Excel, enabling a paperless office. Through workspace isolation by department and hierarchy (e.g., 72B for executives / 7B for employees), permissions are strictly controlled to ensure corporate governance compliance.

Application Results:

Achieve 90% faster data entry through photo recognition, with tiered permissions ensuring compliance.