Doko
Agentic keyboard for getting things done from any text field
Start new thread

Doko - Agentic keyboard for getting things done from any text field

by•
Type what you need. Hold the Doko Bar. Doko understands your intent and hands back the right result, link, or action right where you're typing. Use it for restaurant recs, calendar invites, doc links, weather updates, and custom workflows, without ever leaving your conversation.

Add a comment

Replies

Best
Maker
📌
The problem nobody fixed Since touchscreens replaced physical keys, the keyboard has barely changed. It still just types letters. Every other part of your phone got smarter — the keyboard didn't. Why now AI can finally understand intent, not just words. And the keyboard is the one place every intent starts — before you send a text, search for something, or open another app. So we asked: what if the keyboard didn't just carry your words, but finished the job? That's Doko. Type your intent, hold the Doko Bar, and it happens right there — the result found, the invite sent, the doc pulled — without switching apps or copy-pasting. Build your own keyboard Every key can become a Skill. Connect your apps, wire actions to any key, no code required. Build it in plain language, then share it with the community so others can use your Skill too. The vision We're not trying to make typing faster. We're turning your keyboard from a typing tool into an invisible assistant that lives in every text field — always ready to act, not just type. Join the movement early Doko isn't live yet — we're building in public and opening up a waitlist first. Join now to get early access, founding-member perks, and a say in which Skills we build first.

How does Doko actually figure out intent from a casual sentence without asking a bunch of follow up questions, and is there a cap on how many custom workflows you can set up on the free tier?

Maker

 Good question. Short answer: we don't make the user spell everything out. The model reads the sentence naturally and we feed it context automatically, like location or which app you're typing in, so it usually just gets the intent right without asking follow ups. If it guesses wrong, you just rephrase and try again, which is still faster than switching apps yourself.

On the free tier, we haven't locked in an exact number yet since we're pre launch. Plan is to let people build and use a handful of skills for free, enough to actually feel the value, then have a paid tier for people who want more. We'll set the real number once we see how early testers actually use it.