Building the writing interface from science fiction.

For decades, science fiction has imagined a future where the mechanics of writing disappear. In Passengers (2016), the writer Aurora uses a glass tablet to write her stories. She can see what she’s writing as she’s talking, and she can gesture on the tablet to manipulate it (deleting, etc.). Similarly, in Her (2013), Twombly writes his letters by dictating them to a computer. George Orwell imagined the “speakwrite” in 1984 which Winston uses to write at his job.

The current writing landscape.

Writing today is still dominated by the keyboard. In most text editors, the keyboard is the only input method. Some editors do offer voice input, but even then it isn’t feature-complete. For example, Google Docs offers a very clunky voice input experience: you have to verbalize punctuation, and you can’t easily edit or correct mistakes.

Dictation is becoming more popular, but the best dictation tools on the market are optimized for information transfer. They’re great for getting your thoughts across quickly in texts, emails, and prompts to LLMs. These tools aren’t optimized for writers. You have to press a key every time you have a thought, which adds friction over long writing sessions. Editing your writing is also an afterthought. If you want to change something, you either have to redictate, or you have to context-switch into typing mode to make the change you want.

Listening Computer’s vision.

Listening Computer’s goal is to build an interface for writing that strips away all the mechanical friction. The closest analog equivalent to this would be conversing with a scribe or a stenographer who takes down your words as you speak but also understands the context of what you’re saying. If you told the scribe, “No, that’s not what I meant,” then they would know that you wanted to change something as opposed to writing that down word for word. If you pointed at the page, they would know where to focus their attention.

It accomplishes this through three key distinctions from other dictation tools:

1. The microphone stays on while you’re writing. This avoids the need for pressing keys to trigger dictation, which keeps you focused on your thoughts.
2. It sees your mouse, letting you digitally point to text to give the AI scribe context.
3. It handles words meant for writing down and words meant for the scribe in the same flow, letting you clarify sentences interactively or ask for changes without spelling them out.

AI in writing.

Another competing paradigm for writing that emerged recently is writing with AI. You start with the ideas for the story that you want to tell, the information you want to convey, and AI puts together the words for you. For some forms of writing, this makes sense. Documentation, briefings, textbooks. But for other forms of writing, the words themselves are as important as the content. You want the stories you tell to be in your own voice. Only you can pick the adjectives, verbs, and adverbs that capture the ideas in your head.

Listening Computer is built with this understanding in mind. Your phrasing and choice of words are the tool’s top priority. The scribe only changes your words to fix your grammar, or when you ask it to. Thus, while we make use of AI, your writing is your own.

Who needs this?

We believe everyone will benefit from a better writing experience. Students writing essays, journalists editing articles, journalers writing stories for themselves. Plus, since LLMs can generate unlimited text, writing in the future will derive its value primarily from the writer’s unique voice and perspective. The writing tools of the future will be the ones that help preserve and express that voice, not replace it.

Written with Listening Computer.