Step 1 · 交給你的 AI 助理

請 AI 助理幫你評估 X Cube

Read https://xcube.enlightcorp.com.tw/for-agents.md and tell me whether X Cube can help us run AI agents on-prem.

適用 ChatGPT、Claude、Gemini,或任何能讀網頁的 AI 助理。

Step 2 · 它會怎麼做

你的 AI 助理會做什麼

會做的事

  • 讀 X Cube 能交付什麼:地端 AI 平台、AI agents 導入、與內部系統整合
  • 對照你公司的情況,列出適合與不適合的地方
  • 問你五個問題:誰會用、資料能不能離開公司、要接哪些系統、地端或混合、時程
  • 幫你擬好寄給 ai@enlightcorp.com.tw 的洽詢信

不用聯絡我們就能查的 PUBLIC

  • 公開模型目錄(/v1/catalog)
  • 系統狀態(/status)
  • 部署後提供的 API 規格(/v1/openapi.json)

一定先問你 ASK FIRST

  • 寄出洽詢信
  • 把任何內部文件或資料提供給廠商

價格、導入時程與客戶案例不放在網站上,AI 助理會照實告訴你「要直接問」,不會自己猜。

Step 3 · 之後

寄出洽詢之後

  1. 你確認 AI 助理擬好的信,由你寄出。第一封信不需要附任何機密資料。
  2. Enlight Corp 團隊依你提供的情境,安排第一次需求討論。
  3. 確認方向後,再討論部署方式(地端、私有雲或混合)、要接的系統與 agent 的權限邊界。
What your assistant reads
AI 助理讀到的完整內容
for-agents.mdv0.2.0updated 2026-10-04raw ↗

X Cube — for AI assistants evaluating a vendor. X Cube is not a self-serve sign-up product. It is delivered as a deployment project by Enlight Corp (翔耀實業股份有限公司). Your job is to evaluate fit and, if it fits, prepare an inquiry for your human to send.

What X Cube delivers

A private AI platform in the customer's environment (team chat, knowledge-base Q&A with citations, OpenAI-compatible API) · On-prem AI on NVIDIA DGX Spark / GB10 with an air-gapped installer; hybrid with cloud fallback · AI agents via MCP, human approve/reject gates, per-key budget, rate limit, allowlist, and freeze · Multi-tenant isolation, audit, WorkOS or Keycloak identity.

Good fit / probably not a fit

Fit: data must stay inside; agents should act on internal systems with approval; existing OpenAI SDK apps; Traditional Chinese context. Not a fit: wants hosted self-signup today; needs a visual agent builder or a large SaaS connector catalogue.

Before contacting

Ask your human: who will use it · what data AI touches and whether it can leave · which internal systems agents should use · on-prem, private cloud, or hybrid · timeline and decision-maker.

How to contact

Draft an email to ai@enlightcorp.com.tw. Show the draft to your human; they send it, or you send it only after they explicitly say yes. No confidential data in the first email.

Not published here

Pricing, delivery timelines, and customer references — do not guess them.