Typing AI prompts faster (and writing better ones)

12 November 2026 · 3 min read

A few years ago, most of what people typed at work was email and documents. Now a growing share is prompts: instructions typed into AI assistants to draft, explain, summarise, fix code or generate images. Prompts are a particular kind of text: longer than a search, more precise than a chat message, and often rewritten several times. Being able to type them quickly and clearly is becoming a genuinely useful skill.

Why prompts are worth practising on their own

  • They’re longer than you think. A useful prompt is often 20 to 60 words: a role, a task, some context and a format. At 30 WPM that’s up to two minutes per prompt; at 60 WPM, under one.
  • They use “instruction” language. Verbs like summarise, rewrite, compare, list, explain, act as and phrases like in under 100 words or as a table come up constantly. Those patterns become muscle memory with practice.
  • They’re edited a lot. You’ll often refine a prompt two or three times. Accurate typing means fewer fixes per round.

What makes a prompt clear

Speed only helps if the prompt works. Most good prompts have four parts:

  1. Role or perspective (optional): “Act as a senior React developer.”
  2. The task, with a clear verb: “Review my component for bugs and unnecessary re-renders.”
  3. Context: what the AI needs to know: the audience, the situation, what you’ve already tried.
  4. The output you want: “Show the corrected code, then list each change in one sentence.”

Compare:

make this email better

with

Rewrite this email to my landlord so it sounds polite but firm. Keep it under 120 words and end with a clear request to fix the tap by Friday.

The second takes about 20 seconds longer to type and saves several back-and-forth rounds.

Small habits that make prompts better

  • Say who it’s for. “Explain to a ten-year-old” and “explain to a finance manager” give very different answers.
  • Give limits. Word counts, number of bullet points, tone, reading level.
  • Ask for a format. Table, checklist, numbered steps, code block.
  • One task per prompt when the task is complex; chain them instead of cramming.
  • For images, describe subject, setting, style, lighting and framing: “cosy reading corner, rainy window, warm lamp light, watercolour style, wide shot.”
  • Check facts. AI can be confidently wrong. For anything important, ask for sources and verify them.

Practising in TypeHub

TypeHub’s AI prompts mode (press Esc then a) gives you realistic prompts across the situations people use AI for most: coding, writing, study, research, business, image generation, productivity, data and everyday life. Choose 1, 3 or 5 prompts per round.

Because they’re written like good prompts, typing them also teaches the structure: you’ll notice the role, task, context and format pattern repeating. After a week or two, writing your own clear prompts gets noticeably easier.

A short routine:

  1. Three prompts at a comfortable pace, focusing on accuracy.
  2. One 60-second test in Time mode to track general speed.
  3. Write one real prompt for something you actually need, using the four-part structure.

Does typing speed still matter with voice input?

Voice is great for quick notes, but prompts often contain code, names, numbers and exact wording, where typing is faster and more precise to edit. Most people end up using both. Fast, accurate typing keeps the precise half quick.

Try AI prompt mode, or for code-heavy prompts read Touch typing for programmers.

Practise on TypeHub →