Today I am launching D.U.M.B.
to build opensource AI native cli first media tools
With the first product being DumbEditor :
Native in cli video editor !
Try it at : https://t.co/OjKuceK9ai / https://t.co/RfXENF9MOT
Still a bit early but pretty cool!
A lot of folks were trying Jev from @typesafeai as reranker
I benchmarked Jev against Nemotron (openrouter) as a reranker,
Jev is 2-10 times faster !!! but .. nemotron did had an edge over accuracy! + being OpenSource !
check out full benchmark at : https://t.co/Yc5f4TGWwe
Jev City @ https://t.co/cevO5MlRWv
Fully Live jev city with 15 Jev agents !
Uses a minimalist set of words and emojis for communication + set of actions so that they can interact !
Will expand rules and other features to see how this society develops !
@typesafeai
Sakura / Cherry Blossom theme plugin for youtube :🌸
Simple plug into chrome via chrome extension and covert any website into any style you prefer
WDYT?
Sakura / Cherry Blossom theme plugin for youtube :🌸
Simple plug into chrome via chrome extension and covert any website into any style you prefer
WDYT?
Within 1-2 sec :
Process any DOM into interactable elements using : @typesafeai JEV
No agent -> vision -> decision loop
Recursive DOM Decompilation (fancy name)
check it out at : https://t.co/MOcppsCmLZ
@typesafeai The classic approach of Agent -> screenshot -> vision -> decision is too slow
Splitting the DOM directly into the interactables : brings it close to the real time experiance !!
@yuntiandeng Natural Language Programming !
This is really cool !
I kindoff wanted to see something like this since gap between programming languages is being closed day by day by AI!
Been exploring MCPs + agents, and built this smart todo app – https://t.co/zfyaDJamQD 📝
It’s an LLM with function calling that handles CRUD – just ask it to create/update todos.
Made a quick video about it (👇).
BTW👋 I’m a web + GenAI dev, will be posting here now! 🚀