reddit · SideProject · 10-07 11:50 47 分 ← 返回列表

I built my own running coach app for macOS and the coach runs on your AI

I had Claude create a training plan for a half marathon for me, and I thought, “I might as well have it displayed as a nice GUI instead of just in the chat.” I know, there are already plenty of training apps out there, but that's just my personal use case, and maybe it'll help others, too. It's called Pace Lab: a native macOS app, vibecoded with Claude Code, free and open source. The main idea: your own coach, powered by your own AI. Pace Lab doesn't have its own AI backen

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：跑步者想用 AI 生成/调整训练计划，但又不想把健康数据交给第三方云服务，也想要一个原生的图形界面来查看和追踪训练。

目标用户：已有 AI 订阅（Claude Code/Codex）或本地模型（LM Studio）、注重隐私、使用 Garmin/Strava 的跑步爱好者与半马/全马训练者。

证据：The main idea: your own coach, powered by your own AI. / Pace Lab doesn't have its own AI backend. The coach runs on your Mac through whatever you already use. / No server, no account, no telemetry. Your data stays in a folder on your Mac. / It's built for my own training, so expect rough edges.

变现：免费开源核心 + 付费增值（云同步/多设备、进阶训练分析、结构化计划模板市场、团队/教练版）。也可做“自带 API Key”的付费桌面客户端或订阅制高级分析仪表盘。

MVP：做一个跨平台（或 Web）桌面壳：读取用户训练文件夹中的 Markdown 训练计划和 Garmin/Strava 数据，调用用户自己的 API Key（Claude/OpenAI/本地模型）生成周计划、逐次跑步复盘，并用 git 做版本回滚，两周内可完成核心闭环。

竞品：TrainingPeaks、Garmin Coach、Runna、Strava、Athletica.ai、Stryde

摘要：作者用 Claude Code 开发了原生 macOS 跑步教练应用 Pace Lab，核心卖点是“教练跑在你自己的 AI 上”，支持 Claude Code/Codex 或 LM Studio 本地模型，无服务器、无账号、无遥测，数据留在本机，可读取 Garmin/Strava 并同步到手表，免费开源。发帖仅为个人用例分享，无任何付费数据或他人需求附和，但“自带 AI + 数据私有 + 本地运行”这一模式本身是较新的可变现切入点。