“Tonight isn’t just about the people in front of the camera. In this room are some of the most important TV and film executives in the world. People from every background. But they all have one thing in common: They’re all terrified of Ronan Farrow.”
I developed my own selection and input handling. The app focuses on a hidden input and it extracts the events when you hit keys, and handles them in my editor engine. I haven't done extensive testing but it works pretty well for foreign languages, I've even added decent support for RTL ones like Arabic.
I haven't thought much about accessibility, it's not really a priority for the project right now. If the project really gets big I might revisit that. Google docs is canvas-based, so that was enough precedent for me.
That makes sense, the hidden input approach is clever. Google Docs definitely proved it's viable at scale. The RTL support is a nice touch, that's notoriously tricky to get right.
Great work pulling this off in 10 months. Good luck with the project!
sounds like you be more open to it if Revise worked as a native app (Tauri) with ability to run it against your own LLM endpoint/keys
this may be something I offer one day, with some kind of one-time license. I can't really make this work without a subscription right now, because it relies on servers I host and LLMs provided by 3rd parties which have metered billing.
Yeah Im aware of Lex. I think it falls short in a lot of ways. If you have ideas of how Revise could improve for writing, please email me your thoughts art@revise.io
It was pretty painful and frustrating, but now that it's relatively stable I'm really happy I took that path. I have full control!
I think the way people approach agentic coding depends on their career experience. I had my project acquired previously at the acquiring company hired a large team to work on it. This forced me to go from being a solo dev to a technical manager who's got dozens of people working on code. I treat codex/claude the same way as that; I am leading the ship and it's doing most of the heavy lifting. It's leverage and helps you get more done if you give the right guidance, constraints, and acceptance criteria.
Thank you! I have considered adding some settings like this, but don't want to encourage the use of this software for cheating.
For now, the catch-all solution that anyone can attempt to use is the custom prompt, under account settings. You can instruct it never to use emdash, avoid certain cliches, or simple things like that. But you have to write it yourself for now, there's no convenient presets or anything like that.
My line is I don’t want to market it as a tool that’s obviously for cheating/deception by building features for that. But beyond that, I don’t want to make it my business.
I've definitely had the models refuse to process some content, like a lockpicking guide PDF. It happens pretty rarely, but I'm not testing with especially edgy content.
The agent is model-agnostic though; I already have integrations with 3 providers. It could be extended to more free speech friendly models in the future, should there be good ones that can handle tool calls well enough.
Revise can answer any of those questions! You just have to ask.
You can also focus your questions by selecting a segment of your document, and then writing a prompt; the agent will see what you've selected and focus its efforts on that. You can even prompt with multiple selections attached at once.
I'm hoping to add more "proactive" AI to this eventually, like automatic comments raising the critiques along the lines of these questions you enumerated. Right now the agent has to be prompted first for it to do any real thinking.
reply