i made a calorie tracking app with jev. it's INSTANT 🤯

at this point we're just trying to demonstrate how fast it is and not how accurate it is. But it's day 2, so...

tbf llms are terrible by default for calorie estimation so the only thing to compare is speed and price lol

fair enough 😂

Is this a joke? This is ridiculous and not an AI usecase. This has been solved deterministically like 13 years ago. AI cal tracking means visual understanding of what’s on the plate, chatbots for additional insights, etc.

Perfect case of "to use a sledgehammer to crack a nut", already solved deterministically

This is cool, but in real world nobody is going to use that - it needs to be frictionless to the max. So using an app for that mogs . Btw, everyone uses jev now or will be using by tomorrow - so no moat.

This is cool!

Gonna do demo on @AgentConf ? 😁

@AgentConf i already have too many slides here :D

@AgentConf I can see now 🤣

@parkthomp

instant calorie tracking sounds like a cool feature. what made you choose jev for this?

Desperately trying to find time to dive into Jev!

hi mate, what db are you using for these calorie estimates? or is it just jeff predicting it?

instant != accurate

>not sure how accurate it is — that’s v nice for a pre-check and instant estimate other model could later refine is needed btw i think that food capture is bigger problem, or even food planning than logging if plan is good - i stopped feel need to track it all even 🤔

Brother gonna be the first one to spend 10$ of Jev tokens in a day with this use case

I can't take it seriously after it predicted F to be the 4th letter of "elephant"

There's no point?

jev is the ai we always needed

@raroque ?

the killer feature isn't instant. it's that the app never has to pretend prose is data.

this is the ideal interface for me (when i used to do this). i don't calorie count anymore though.

You don't sleep at all I presume.

So it’s a more matter of fact model? And not something that can do let’s say strategy type of prose?

this is cool af🔥do u have a link to download or maybe github?

that's pretty good. quick feedback is all we need.

@raroque

Man you are looking great! Can tell the big difference since that photo in the hospital. Well done 👍

how do you predict the macros, using scoring questions? I'm confused

Hmm very interesting…. Lemme try this… First use jev to separate the hiven sentence to separate the quantity, unit and food name, pass food name to the search function then passing to jev which selects the right item then it chooses the unit and boom… is that’s right?

do you use something like binary search for this?

@raroque check this out

Thank you for this demo. It unlocked something new for me that applies to a very specific niche.