It would be nice to have something like a score to indicate how powerful it is, determined by the price, to see which one is kind of the best.

Neat: when clicking on the name, I would like to be redirected to Amazon. The link on the far right was hard to find. :)

Why are you posting here when you could be crushing it™ delivering 100x value to your shareholders? You've already failed.

-4 points, the joke worked well

Start to earn money guys! You can do that!

P.S. donated.

Looks super interesting

thank you for checking it out! what sorts of experiences have you had with agents so far?

Whatever everyone thinks, YC is playing to win. And they make winners

>>if you don’t make mistakes it means you are not trying hard enough. << Elon Musk

Big picture, people who never heard of PearAI or continue dev just did. I did. I tried it. It’s as good as cursor

And guess what? whatever side you are on, YC and those startups won either way. More developers heard about each one of them and more tried it.

Counterpoint: PearAI was launched as a trivial copy of another product and framed by its creators as the greatest thing since sliced bread.

It's obvious that these people are less interested in making something useful than they are in making something that will generate enough hype to get another round of funding.

Okey, I'll say it. It will not work because of network bottlneckes. You need to be sending gigabytes of Data.

so by definition you need (1) good internet 20mb/s+ and (2) good devices.

This thing will not go any further than cool demo on twitter. Please prove me wrong.

Try it out - don't trust me!

The way this works is that each device holds a partition of the model (for now a continuous set of layers). E.g. let's say you have 3 devices and the model is 32 layers. Device 1 could hold layers 1-10, device 2 holds 11-20 and device 3 holds 21-32. Each device executes the layers it's responsible for and passes on the output of its last layer (the activations) to the next device.

The activations are ~8KB for Llama-3-8B and ~32KB for Llama-3-70B (it's linear in the number of parameters in that layer and Llama-3-70B has more layers). Generally the larger the model gets (in terms of parameters), the more layers it ends up having, so we end up with sub-linear scaling so I expect Llama-3-405B to have activations on the order of ~100KB.

This is totally acceptable to send over a local network. The main issue you run into is latency, not bandwidth. Since LLMs are autoregressive (tokens are generated serially), additional latency limits throughput. However, over a local network latency is generally very low (<5ms in my experience). And if not, it's still useful depending on the use-case since you can get a lot of throughput with pipeline parallelism (overlapping requests): https://pytorch.org/docs/stable/pipeline.html

I can see myself using this in car while driving to work. interesting. How expensive is this btw? and will you open source?

Voice is pretty expensive unfortunately. We pay around $10 cents per minute for the voice interaction. LLM calls not included.

Can you just use the browser text-to-speech API?

I had an extremely negative experience creating apps in Swift.

I can't see any logical reason why people would not use Electron/Tauri.

Swift doesn't allow you to have full control of what you want to do.

I built a whole macOS [ fixkey dot ai] app in Swift, and it was a very painful development experience.

Were you new to Swift? It for sure has its issues, but lack of control wouldn’t make it on my list.

Love the key binding to select & fix the current line!

I dont have typos that often, so their 15 free fixes every month works perfectly good for me

Hey! I am the @karmedge who wrote this post on Twitter

I wonder what do you guys think about tool which will be able to automate actions on computer? Like clicking smth, automatically opening the window for me after I type in some request?

Something like hyperwriter assistant but for Mac. Very similar to open interpret but with accessibility option

I wonder what else is there in the space?

Something what I'd really like, would be:

1. You have an application open, let's use for this example Apple Notes 2. You ask it to summarize the "Recipe xyz" 3. It will search for the recipe 4. Open the note 5. "Copy" the text 6. Summarize it for you

This is a very simple example and maybe not "cool" enough. But would be great to give it any kind of interface and then do some more "advanced" (for an LLM) tasks.

I think this could be possible with GPT4-Vision, but it would be very costly, for things which you could do yourself with a few clicks and key presses.