惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

Engineering at Meta
Engineering at Meta
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
小众软件
小众软件
博客园_首页
T
Tailwind CSS Blog
美团技术团队
博客园 - 叶小钗
Microsoft Security Blog
Microsoft Security Blog
有赞技术团队
有赞技术团队
Apple Machine Learning Research
Apple Machine Learning Research
大猫的无限游戏
大猫的无限游戏
Microsoft Azure Blog
Microsoft Azure Blog
H
Hackread – Cybersecurity News, Data Breaches, AI and More
I
InfoQ
MongoDB | Blog
MongoDB | Blog
The Cloudflare Blog
J
Java Code Geeks
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
博客园 - 聂微东
酷 壳 – CoolShell
酷 壳 – CoolShell
Blog — PlanetScale
Blog — PlanetScale
IT之家
IT之家
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
Y
Y Combinator Blog

DEV Community

Authentication Security Deep Dive: From Brute Force to Salted Hashing (With Java Examples) Why AI Systems Don’t Fail — They Drift Spilling beans for how i learn for exam😁"Reinforcement Learning Cheat Sheet" I Replaced Chrome with Safari for AI Browser Automation. Here's What Broke (and What Finally Worked) How Python Borrows Other People's Work The $40 Architecture: Processing 1 Billion API Requests with 99.99% Uptime Vibe Coding: A Workflow Guide (From Zero to SaaS) Most webhook security guides protect the wrong side. The scary part is delivery. Headless CMS for TanStack Start: Build a Blog with Cosmic EU Age Verification App "Hacked in 2 Minutes" — What Actually Happened Comfy Cloud’s delete function does not actually remove files Running AI Models on GPU Cloud Servers: A Beginner Guide Event-driven media intelligence with AWS Step Functions and Bedrock I scored 500 AI prompts across 8 quality dimensions — here's what broke How to Call Google Gemini API from Next.js (Free Tier, No Backend Needed) The Portal Protocol: Reclaiming Human Connection in the Age of AI How to Fix Your Team's Scattered Knowledge Problem With a Self-Hosted Forum Intro to tc Cloud Functors: A Graph-First Mental Model for the Modern Cloud Designing Multi-Tenant Backends With Both Ownership and Team Access I Built a Neumorphic CSS Library with 77+ Components — Here's What I Learned PostgreSQL Performance Optimization: Why Connection Pooling Is Critical at Scale Cómo construí un SaaS multi-rubro para gestionar expensas en Argentina con FastAPI + Vue 3 🚀 I Built an Ethical Hacking Scanner Tool – Open Source Project I Replaced /usage and /context in Claude Code With a Single Statusline A Pythonic Way to Handle Emails (IMAP/SMTP) with Auto-Discovery and AI-Ready Design I Collected 8.9 Million Polymarket Price Points — Here's What I Found About How Markets Really Move EcoTrack AI — Carbon Footprint Tracker & Dashboard Everyone's Using AI. No One Agrees How. 5 self-hosted ebook managers worth trying in 2026 Building Your First AI Agent with LangChain: From Chatbot to Autonomous Assistant
AI Button UX — Where to Put It, How to Label It, What to ...
hiyoyo · 2026-05-05 · via DEV Community

If this is useful, a ❤️ helps others find it.

All tests run on an 8-year-old MacBook Air.

The AI feature is only as good as the UI around it. A powerful diagnosis that's hard to trigger, slow to show, or confusing to read doesn't get used.

Here's what I learned from iterating on HiyokoLogcat's AI button.


Where to put the trigger

Inline with the content, not in a toolbar.

My first version had an "AI Diagnose" button in the top toolbar. Users had to: spot the error, note the line number, click the toolbar button, somehow tell it which line.

The fix: put a small icon directly on each error line. Click the icon → diagnose that line. Zero ambiguity.

{logLines.map((line, idx) => (


    {line.message}
    {line.level === 'E' && (
       handleDiagnose(idx)}
        title="AI診断"
      >
        🐥

    )}


))}

Enter fullscreen mode Exit fullscreen mode

Show the button only on error lines. No button on debug/info lines — that would be noise.


How to label it

Don't say "AI." Users don't care about the implementation.

Bad Better
"AI Analyze" "Diagnose"
"GPT Explain" "What's wrong?"
"Neural diagnosis" "Fix this"

One word. Action-oriented. No tech jargon.


What to show while waiting

Three states, three UIs:

type DiagnosisState = 'idle' | 'loading' | 'done' | 'error';

function DiagnosisOverlay({ state, result }: Props) {
  if (state === 'idle') return null;

  return (


      {state === 'loading' && (


          🐥
          診断中...


      )}
      {state === 'done' && (

{result}

      )}
      {state === 'error' && (


          もう一度お試しください
          再試行


      )}


  );
}

Enter fullscreen mode Exit fullscreen mode

Loading: Show something immediately. A spinner or pulsing text. 3 seconds of silence feels like a crash.

Done: Show the result. No animation needed — the content itself is the payoff.

Error: Short message + retry button. Never a raw error code.


Dismiss behavior

Three ways to close, all feel natural:

  • Press ESC
  • Click outside the overlay
  • Click the X button
useEffect(() => {
  const onKey = (e: KeyboardEvent) => {
    if (e.key === 'Escape') onClose();
  };
  window.addEventListener('keydown', onKey);
  return () => window.removeEventListener('keydown', onKey);
}, []);

Enter fullscreen mode Exit fullscreen mode

Don't make users hunt for a close button. ESC is the universal "get out of this."


The one thing that matters most

Speed of feedback. If clicking the button produces no visible response for even half a second, users will click it again, or assume it didn't work.

Show something within 100ms of the click. The AI response can take 3 seconds. The UI response cannot.

const handleDiagnose = async (idx: number) => {
  setDiagnosisState('loading');  // ← immediate, < 1ms
  const result = await invoke('diagnose', { idx });  // ← takes 3s
  setDiagnosisState('done');
  setResult(result);
};

Enter fullscreen mode Exit fullscreen mode


Hiyoko PDF Vault → https://hiyokoko.gumroad.com/l/HiyokoPDFVault
X → @hiyoyok