i built wellposed as a plugin/skill which works with any agent (https://github.com/suraj-phanindra/wellposed) to ensure your agent understands how to choose the right kind of jev request and format it correctly not just for syntax but for completeness and correctness. it is well-documented in the "jagged-ness" docs (https://docs.typesafe.ai/model-jaggedness/jev-1.13) that typesafe include on their docs page that the absence of essential options can cause jev to choose the wrong option with high confidence (it cannot choose what it cannot see in the request) - so the responsibility to ensure whatever the intent behind your jev request is - it is captured correctly with the right states and options for jev to pick from falls on the user. wellposed should ideally make your agent better at converting NL intent into jev requests. please try and give me feedback. appreciate it!
https://x.com/dWeaths/status/2102415625301717065 - Here's my use case of jev, being able to accurately detect nouns (with adverbs, adjectives etc.) in realtime as the user is typing it, genuinely feels like it's running locally with how fast it comes back. I've sent over 1000 requests to Jev and its cost me $0.01 (probably rounded up!).
I had an idea for a 'write without space, basicallylikethis' and Jev (given how cheap it is) just being asked after every key press where inserting spaces (or making typo correction) would make sense.
Honestly just getting it integrated into mobile phone swipe keyboards would be a godsend. If I type "We need to get going " and then swipe the word "now", I really do not think "mower" should be the word it chooses. Present a set of swipe-based likely words and the preceding text message to Jev, or similar model, and pick its highest prob word.
The demo is here: https://levmiseri.com/nospace
And so, I typed "Thisoneiscoolashell", and got "This one is cool a shell" hahaha
Great idea, and it's really quick
https://github.com/midplane/clean-twitter
Demo here: https://x.com/RaahelSaidWhat/status/2102162969656475973