reddit · SideProject · 11-01 04:50 50 分 ← 返回列表

I built an app that turns Macs into one local and private AI server

Hey all, I'm Alejandro, a software engineer based in Mexico ( kinduff on the web )! I've been experimenting with local models on Apple silicon for a while now. It's interesting how much these small machines can handle, especially considering how little we actually use their computing power throughout the day. That got me thinking about all the Macs sitting around in offices while their owners are in meetings, having lunch, or simply done for the day. Could we use some of t

五维评分

付费意愿

需求真实度

个人可落地

竞争空间

新鲜度

AI 分析

痛点：办公室和家里大量 Apple silicon Mac 在开会、午休、下班后处于闲置状态，算力被浪费；同时团队想用本地私有 AI 推理却缺算力，也缺少把这些闲置 Mac 汇聚成一个 OpenAI 兼容本地 API 的轻量方案。

目标用户：拥有多台 Apple silicon Mac、对数据隐私敏感的中小团队、开发者和企业内网 IT 负责人

证据：Siesta is a paid app, licensed per Mac, with a 30-day refund policy. / You can use it through an OpenAI-compatible API, CLI, or MCP. Everything stays on your own network. / Could we use some of that idle time to run models for everyone else?

变现：按 Mac 席位售卖永久授权或年订阅；企业内网批量部署授权加技术支持；后续把调度与监控面板做成订阅制 SaaS

MVP：两周内只做 macOS 菜单栏 App：检测空闲状态（锁屏或无键鼠输入 N 分钟）时用 MLX 自动加载小模型，在本机暴露 OpenAI 兼容的 /v1/chat/completions 接口，用 Bonjour 做局域网自动发现，一台当 server、其余当 worker，用户回来即让出资源。

竞品：Ollama、LM Studio、llama.cpp、Exo、Petals、Apple MLX

摘要：墨西哥独立开发者 Alejandro 发布 Siesta 1.1.0，让办公室里的 Mac 在闲置时段共享算力运行本地模型，一台做 server、其余跑 helper，用户回到电脑前会自动让出资源，对外提供 OpenAI 兼容 API、CLI 和 MCP，全部流量留在局域网内。该产品按 Mac 收费并有 30 天退款政策，属于已验证的付费形态，但帖中未给出任何收入或用户付费数据，需求真实性主要靠作者个人判断，且该赛道已有 Ollama、LM Studio、Exo 等成熟竞品。