Dictation is the Way Forward

If you're working with AI tools and still typing everything out, you're leaving performance on the table.

I switched to dictation as my primary input method about six months ago. Started with AI prompts — figured I'd try it for those long explanations where I knew what I wanted but couldn't be bothered typing it all out. The bandwidth difference was immediate. Speaking lets you convey nuance, reasoning, and context far more naturally than typing ever will. The AI responses improved noticeably.

Now I dictate almost everything. Code comments, commit messages, documentation, even these blog posts.

Why it works

Typing forces you to compress your thoughts. You naturally edit as you go, cutting things down because typing is slow and tedious. With dictation, you just... talk. The ideas flow. You explain yourself properly. When working with AI agents, that extra context matters. A lot.

The keyboard is a bottleneck. We've just been living with it for so long we stopped noticing.

The tool

I started with Superwhisper, which was solid enough until one too many bugs made me go looking again. Now I use VoiceInk. Either works. Built-in Mac dictation doesn't... not for my accent anyway.

The shift

It feels odd at the start. You'll want to correct every incorrectly spelt word, every slightly wrong transcription. Then you realise the LLM doesn't care. It understands context. It knows what you meant even when the dictation garbled a word or two.

After a while, you stop worrying about precision and just focus on expressing yourself. The flow really works.