Looks promising. I'm building an AI video editor and multi tool calls take >30s using Gemini. This would be a a game changer if Jev can take that down to single digits at p95.
You can stand out by being authentic. Your face on the video, etc. Even writing it yourself genuinely instead of using an LLM. ex the article felt authentic.
But how you get authentic and engaging for bank statement converter? People search for it to solve particular problem in the first place, not to hear your life story.
And regarding face on video, I just got stuned with video of 106 years old giving life advices and all people but one in comment section didn't notice it is well generated.
-> But how you get authentic and engaging for bank statement converter? People search for it to solve particular problem in the first place, not to hear your life story.
How does it work? Im building a video editor and right now it has access to nearly 100 tools. Would be good to learn the techniques you used to make tool discovery more efficient.
The readme has some examples for what it does. It doesn’t list the entire schema (noisy). Instead it uses shorthand. Perhaps a sufficiently smart agent can do this.
It’s mostly a web app with a native swift companion app to access local resources like TTS, file storage and video rendering.
I arrived at the same conclusion as you; every app should have AI chat. Just give it full access to the app like a real user. I found that users prefer talking to the AI when editing vs clicking around.
> Donkey Vision — a screen-understanding API that turns any screenshot into structured, clickable UI, so you can build computer-use tools of your own.
Not faulting your (necessary given today's reality) work, but sad to think that the Amiga still remains ahead of it's time some 36 years later (since ARexx was bundle with the OS), or closer to 40 years if you count when v1 was released separately.
ARexx allowed any application that supported it (and most productivity apps worth their salt did) to communicate with each other, and to be scripted - essentially a standardised API for applications.
We're slowly getting there (at least for SaaS apps) with MCP, but screen scraping shouldn't ever have been necessary.
Ifound Palmiers interface dense, cluttered, and confusing - just take the panel top left (to the right of chat) for example. So much going on in one small space.
Donkey Cut offers more functionality in less space, and everything is easy to find and understand. It isn't without bugs, but the structure is there for it to be great
I feel the same way too. I also open sourced a video editor. It's free to use and you can buy credits whenever you want to access AI specific features.
