An AI-assisted circuit design tool for new makers, electronics hobbyists and non-major students. Don't draw — read hardware. 面向新手创客、电子爱好者与非科班学生的 AI 辅助电路设计工具。 别画了，来读读硬件。
What it is 是什么
Fast Hardware breaks the high-barrier, operation-heavy ceiling of traditional EDA tools. Pioneering the "hardware reader" positioning, it uses LLMs as a cognitive translator — turning abstract electrical parameters and netlists into natural-language explanations and visual analogies. Built on Electron, it runs offline, loads Skill modules on demand, and ships with a Human-in-the-Loop design flow. More than a tool — a maker ecosystem of learn, design, share, iterate. Fast Hardware 打破了传统 EDA 软件「高门槛、重操作」的局限，首创「硬件阅读器」定位， 利用 AI 大模型充当认知翻译官，将抽象的电气参数、网表逻辑转化为自然语言解释与可视化类比。 Electron 桌面架构，支持离线可用、按需加载 Skill 模块与 Human-in-the-Loop 协同设计流。 不仅是设计工具，更是一个集「学习、设计、分享、迭代」于一体的硬件创客生态平台。
With STEAM education and the maker movement going mainstream, hardware design is moving from "for engineers" to "for everyone." But while AI penetration in software / content already exceeds 60%, hardware design is still under 5% — a massive tool gap. 随着 STEAM 教育与创客运动普及，硬件设计从「工程师专属」走向「大众创作」。 但 AI 在软件/内容领域的渗透率已超 60%，硬件设计领域仍不足 5%，存在巨大的「工具代差」。
Altium, LCEDA and friends focus on engineering compliance. Learning curves span weeks. Non-major users hit dense schematics, netlists and datasheets and drown in cognitive overload — "I can't read this, I'll wire it wrong, I'll burn the board." Altium、立创 EDA 等专业工具聚焦工程合规性，学习周期长达数周。 非科班用户面对密密麻麻的原理图、网表和 Datasheet 时，往往产生 「看不懂、怕接错、怕烧板」的认知超载与恐惧感。
The rise of LLMs makes "cognitive assistance" possible. Through natural language interaction, smart recommendation and automated generation, AI can convert hardware's "operation barrier" into a much lower "understanding barrier" — letting makers shift from drawing to reading hardware. LLM 的爆发为「认知辅助」提供了可能。通过自然语言交互、智能推荐与自动化生成， AI 有能力将硬件设计的「操作门槛」转化为「理解门槛」， 让创客从「画图」转向「读懂硬件」。
Who it's for 为谁而做
| Persona用户类型 | Profile特征 | Core need核心诉求 | Scenario场景 |
|---|---|---|---|
| University makers大学生创客 | EE / automation / CS, 18–25, theory-strong but practice-light 电子 / 自动化 / 计算机专业，18–25 岁，有基础理论但缺乏实战经验 | Validate ideas fast, ship coursework / contest builds 快速验证方案、完成课程设计 / 电赛 | Coursework, contests, capstones 课程设计、电子竞赛、毕业设计 |
| DIY enthusiastsDIY 爱好者 | Self-taught, idea-rich, hands-on 非科班但有硬件创意，自学能力强，喜欢动手 | Lower the entry barrier, see real builds quickly 降低入门门槛、快速看到实物效果 | Personal projects, smart home mods 个人项目、兴趣制作、智能家居改造 |
| K-12 STEAM teachers中小学科创教师 | IT / general-tech teachers needing classroom-friendly tools 信息科技 / 通用技术老师，需要适合教学的简易工具 | Reduce student frustration, standardized teaching cases 降低学生挫败感、提供标准化教学案例 | School curriculum, clubs, science contests 校本课程、社团活动、科创比赛辅导 |
| Maker spaces / labs创客空间 / 实验室 | Universities, training centers, incubators running batch training 高校 / 培训机构 / 众创空间，需要批量培训与方案复用 | Lower teaching cost, build internal knowledge assets 降低教学成本、建立内部知识资产库 | Trainee onboarding, project incubation, talks 学员培训、项目孵化、技术分享 |
Three cognitive loads 三大认知负荷痛点
Intrinsic
Voltage, current, signal flow — abstract concepts that exceed working memory. Without grounded analogies, beginners can't form an intuitive model. 电压 / 电流 / 信号流向等概念抽象，工作记忆容量有限。 缺乏生活化类比，新手难以建立直观认知。
Extraneous
Legacy EDA UI is dense, steps are redundant, cognitive guidance is missing. "Afraid to burn the board, afraid to wire it wrong, afraid of datasheets" — fear silently consumes working memory. 传统 EDA 界面复杂、步骤冗余、缺乏认知引导。 「怕烧板子、怕接错线、怕看不懂 Datasheet」的恐惧感占用大量工作记忆。
Germane
Community content lives as static files or dry code — hard to reuse structurally. Great designs lack a Fork mechanism, blocking schema transfer and second-round iteration. 社区内容多为静态文件或枯燥代码，难以结构化复用。 优秀方案缺乏 Fork 机制，无法促进知识图式迁移与二次迭代。
Two real-world stories 两个真实场景
Wang, a junior, gets handed a "smart desk lamp" coursework brief and freezes in front of Altium. With Fast Hardware, he types his idea in plain language; AI picks parts, generates a schematic, highlights pin functions. Stuck on capacitor charging? AI draws a "water bucket" analogy. Three days later he's not just done — he publishes the build to the Maker Marketplace and racks up 50+ likes. 大三学生小王接到「智能台灯」课程设计，面对 Altium 无从下手。 使用 Fast Hardware，他输入自然语言需求，AI 自动推荐元件、生成原理图并高亮引脚功能。 遇到不懂的电容充放电，AI 用「水桶蓄水」类比解释。 3 天后，他不仅完成了电路搭建，还一键发布方案到创客集市，获得了 50+ 点赞。
Middle-school IT teacher Li wants to launch an IoT club, but skill levels are all over the map. She uses Fast Hardware's modular Skills, loads only the "canvas-wiring" skill, and the UI auto-simplifies. Students Fork her base project, ask the built-in agent questions, and ship personalized builds. Class completion rate climbs from 40% to 80%. 中学信息老师李老师想开设 IoT 社团课，但学生基础参差不齐。 她利用 Fast Hardware 的 Skill 模块化功能，按需加载「画布连线」skill，界面自动简化。 学生通过 Fork 机制复制她的基础方案，向内置 agent 询问，即可实现个性化项目。 课堂任务完成率从 40% 升至 80%。
From operating to understanding 从「操作」到「理解」
AI as translator: abstract electrical parameters become natural-language explanations and visual analogies. Lowers intrinsic load. AI 充当翻译官，将抽象电气参数转化为自然语言解释与可视化类比，降低内在负荷。
Skill modules load on demand to prevent feature overload. Advanced features unlock with proficiency, matching the user's schema-building pace. Skill 模块化按需加载，避免功能过载；随熟练度逐步解锁高级功能， 匹配图式构建节奏。
User confirms → AI executes → feedback refines. Keeps the user in control, eliminates uncertainty anxiety. 用户确认 → AI 执行 → 反馈优化，保持控制感，消除不确定性焦虑。
Fork projects, diff versions, credit contributors. Drives knowledge transfer and community-level co-evolution. 支持方案 Fork、版本对比、贡献者署名，促进知识迁移与社区协同进化。
Five core highlights 五大核心亮点
"Hardware reader" positioning — not industrial PCB precision, but cognitive understanding and educational support. 「硬件阅读器」定位 —— 不追求工业级 PCB 精度，聚焦认知理解与教育辅助。
Skill modularity controls cognitive load. Arduino, IoT, sensor packs load on demand — focused attention, no feature overload. Skill 模块化与认知负荷控制。Arduino / IoT / 传感器场景插件按需加载， 避免功能过载，专注力集中。
HITL relieves fear. AI shows reasoning, parameters are editable, AI re-computes and previews impact. Every action is reversible. HITL 协同与恐惧感缓解。AI 显示推理依据，参数可改、AI 重算并提示影响， 所有操作可逆。
Marketplace + Fork-style iteration. One-click Fork to local sandbox, branch on edit, contributor lineage tracked — a "GitHub × RedNote" hybrid for hardware. 创客集市与 Fork 式迭代。一键 Fork 至本地沙箱，修改后生成新分支， 系统记录贡献者链条 —— 硬件领域的「GitHub + 小红书」混合生态。
Offline-capable, professional feel. 114 MB lightweight installer, autosave, state recovery, keyboard shortcuts — desktop-grade UX. 离线可用与专业体验。114MB 轻量安装包，核心功能离线可用， 自动保存、状态恢复、快捷键支持。
Four-layer architecture 四层架构
Tech stack 技术栈
Frontend
Electron 27.0.0 + HTML5 + CSS3 + JavaScript ES6. Electron 27.0.0 + HTML5 + CSS3 + JavaScript ES6。
AI integration
Default Qwen3.5-27B; pluggable across GLM / Qwen / DeepSeek and other models. 默认 Qwen3.5-27B，支持 GLM / Qwen / DeepSeek 等多模型切换。
Render engine
Canvas API for circuit visualization, marked.js for Markdown rendering. HTML5 Canvas API（可视化电路设计）+ marked.js（Markdown 渲染）。
Storage
JSON file system + persistent local config + cloud marketplace sync. JSON 文件系统 + 本地配置持久化 + 云端集市同步。
Cross-platform
Deeply optimized for Windows 10/11 (64-bit). macOS / Linux versions on the roadmap. Windows 10/11（64 位）深度优化，macOS / Linux 版本规划中。
Footprint
Lightweight installer; core flow works offline. 轻量安装包，核心功能离线可用。
Buy-once + value-add 买断制 + 增值服务
| Edition版本 | Price价格 | Includes功能权益 | Best for目标用户 |
|---|---|---|---|
| Personal Personal edition个人版 |
¥99 one-time, lifetime永久买断 |
AI scheme generation, smart part selection, multi-project management AI 方案生成、智能选型、多项目管理 | Individual makers / DIY enthusiasts 个人创客 / DIY 爱好者 |
| Education Education edition教育版 |
¥999 per school / year每校 / 年 |
50-seat batch license, dedicated support group, teaching case packs 批量授权（50 席）、售后群专属支持、教学案例包 | Universities, K-12, training institutions 高校 / 中小学 / 培训机构 |
Revenue extension 01延伸 01
One-click BOM ordering with 5–10% commission. BOM 一键下单，赚佣金（5–10%）。
Revenue extension 02延伸 02
Project-based custom part libraries and features for enterprises / institutions. 企业 / 机构定制元件库与功能（项目制）。
Revenue extension 03延伸 03
Vendor placements (LCSC, Taobao stores) with high targeting precision. 元器件厂商精准广告（立创 / 淘宝店）。
Roadmap 未来规划
v1.0 GA release. Maker Marketplace 1.0 goes live with 100+ open-source builds. 完成 v1.0 正式版发布，上线创客集市 1.0，积累 100+ 开源方案。
macOS / Linux releases. Lightweight web collaboration support. 推出 macOS / Linux 版本，支持 Web 端轻量协作。
Multimodal AI: snap a photo of a real part and generate a schematic. The "Hardware Copilot." 接入多模态 AI（支持拍照识别实物生成原理图），推出「硬件版 Copilot」。
Build the world's largest hardware UGC community — every idea, readable and reusable. 构建全球最大的硬件 UGC 社区，让每个创意都能被「阅读」和「复用」。
Download builds, browse the GitHub repo, or write to us about education licensing and custom builds. 下载构建版本、查看 GitHub 仓库，或就教育授权和定制开发与我们联系。