# 策展 · X (Twitter) 🔥🔥🔥🔥

> 📖 本站完整內容索引（documentation index）：[llms.txt](/llms.txt)

> 作者：Microsoft AI (@MicrosoftAI) · 平台：X (Twitter) · 日期：2026-06-18

> 原始來源：https://x.com/MicrosoftAI/status/2067359183133896737

## 中文摘要

Microsoft AI 整合MAI模型將語音描述轉化為程式碼。

**核心技術整合**
Microsoft AI 近期發布了一項基於 VS Code 分支版本的原型展示，驗證了 `MAI-Transcribe`、`MAI-Voice` 與 `MAI-Code-1-Flash` 三大模型在統一工作流中的協作能力。此系統旨在打破傳統開發介面的限制，讓開發者能透過語音直接驅動程式撰寫與專案管理，實現從口語指令到實際應用程式功能的端到端轉換。

<video src="https://pub-75d4fe1e4e80421b9ecb1245a7ae0d1a.r2.dev/curated/1781759324217-6yklgtuo.mp4" poster="https://pub-75d4fe1e4e80421b9ecb1245a7ae0d1a.r2.dev/curated/0ac77f97edfbe0a9.png" controls playsinline preload="metadata" style="max-width:100%;height:auto;display:block;margin:1rem 0"></video>
> 這段影片展示了開發者如何透過語音與 AI 程式碼編輯器進行互動，實現自動化編碼與專案管理。

**操作流程與互動機制**
根據展示內容，該開發環境透過整合式 AI 語音助理介面，提供高度自動化的開發體驗：
1. 介面提供「Open Chat」、「Show All Commands」與「Start Debugging」等快捷指令，並允許使用者在「MAI-Code-1-Flash (Preview)」模型中選擇「Local」或「Default Approvals」設定。
2. 開發過程中，AI Agent 會主動監控任務進度，並針對關鍵操作（如啟動 Vite 開發伺服器或執行 `npm` 指令）向使用者請求執行批准。
3. 系統採用懸浮視窗（Floating Voice Window）呈現當前任務狀態，使用者可隨時透過語音詢問「What's going on with the agents right now?」，AI 會即時列出當前運作中的代理任務狀態。
4. 在實際應用場景中，AI 能夠根據語音指令動態調整網頁樣式（例如將背景改為藍綠漸層），並自動新增單元測試，展現了語音驅動開發的即時性與精確度。

**實際影響**
此原型展示了 AI 代理在開發環境中的深度整合趨勢，不僅是單純的程式碼生成，更涵蓋了從環境配置、功能開發到測試驗證的完整生命週期。透過將語音互動與 Agentic 程式開發結合，開發者能更專注於邏輯構思，而非繁瑣的指令輸入，顯著提升了開發效率與人機互動的流暢度。

## 媒體內容

**這段影片展示了開發者如何透過語音與 AI 程式碼編輯器進行互動，實現自動化編碼與專案管理。**

**影片中的 Prompt 與操作**

Prompt（00:01）：

```
嘿，你能為我做什麼？
```

原文：Hey, what can you do for me?

Prompt（00:10）：

```
你能將網頁應用程式在本地執行，讓我們開始測試嗎？
```

原文：Can you run the web app locally so we can start testing?

Prompt（00:25）：

```
你現在能將網頁應用程式的背景顏色改為藍綠漸層嗎？
```

原文：Can you now change the background color of the web app to a gradient of blue and green?

Prompt（00:40）：

```
現在你能檢查我們有哪些單元測試，看看是否有缺失並建立它們嗎？
```

原文：Now can you check what kind of unit tests we have and see if there's any missing and create them?

Prompt（01:40）：

```
目前代理們的狀況如何？
```

原文：What's going on with the agents right now?

操作步驟：

1. （00:16）點擊「Approve」批准啟動開發伺服器
2. （00:34）懸浮視窗顯示 AI 正在處理背景顏色變更
3. （00:54）點擊「Approve」批准執行 npm 指令
4. （01:43）語音助理顯示代理任務狀態列表

**逐字稿**

- `00:00` 嘿，我是 Copilot。有什麼我可以幫你的嗎？（Hey, I'm Copilot. How can I help?）
- `00:02` 嘿，你能為我做些什麼？（Hey, what can you do for me?）
- `00:06` 我可以啟動 Agentic 程式開發，並指派任務給它們。（I can start coding agents, give them tasks.）
- `00:09` 聽起來很酷。你可以先在本地端執行這個網頁應用程式，讓我們開始測試嗎？（That sounds cool. Can you run the web app locally so we can start testing?）
- `00:13` 沒問題，讓我來處理。（I got it. Let me work on that.）
- `00:15` 我需要你的批准才能啟動開發伺服器。（I need your approval to start the development server.）
- `00:18` 沒問題，我批准。（Sure, I approve.）
- `00:22` 開發伺服器已在 localhost 啟動。（The development server started on local host app.）
- `00:24` 謝謝。現在你可以把網頁應用程式的背景顏色改成藍綠色漸層嗎？（Thanks. Can you now change the background color of the web app to a gradient of blue and green?）
- `00:32` 沒問題，讓我來處理。（I got it. Let me work on that.）
- `00:37` Agent 已經更新了專案。（The agent updated the project.）
- `00:39` 謝謝。現在你可以檢查我們有哪些單元測試，看看是否有遺漏的並將其補上嗎？（Thanks. Now can you check what kind of unit tests we have and see if there's any missing and clear them?）
- `00:47` 沒問題，讓我來處理。（I got it. Let me work on that.）
- `00:49` 測試套件已準備就緒。我需要你的批准來執行 NPM。（The test suite is ready to run. I need your approval to execute the NPM.）
- `00:53` 沒問題，我批准。（Sure, I approve.）
- `00:56` Agent 已經為個人資料標頭新增了單元測試。（The agent added unit tests for the profile header.）
- `00:59` 現在這些 Agent 的狀況如何？（What's going on with the agents right now?）
- `01:02` 請稍候，正在檢查。（One moment checking.）
- `01:04` 這三個 Agent 目前都在主動檢查目前的應用程式結構。（All three agents are actively checking the current app structure.）
- `01:08` 其中一個正計劃將元件重構為一個...（One is planning to refactor components into a...）
- `01:11` 沒有人正計劃將 email 分離，（No one is planning to separate the și-mail然後,）
- `01:11` 以及外部的 power-up。（��� external power-up.）
- `01:11` 在你之後，Agent 會引導至資料來源的請求。（After you, the agent leads to the request of the data source.）

## 標籤

新產品, IDE, 自動化, STT, Microsoft, VS Code
