HN user
i make toons at toonworld.io
congrats on the launch. i'm curious how you think about on device AI vs providers that are in the cloud. where do you see the capabilities of models that can run on the phone, and are there inherent advantages to this?
nice but seriously: how do i turn on ad block? don't want to see ads on youtube
pulling a cursor on google chrome lol
there's a long tail of edge cases where a smaller model does something dumb with input/output, which causes a decent amount of pain to the user (e.g. nodemon, do we stop/do we start?). Coding agent developers can't make an opinion on every possible terminal command
maybe what we're asking for is that terminal commands keep in mind that agents are now the primary users of them, and have a feeling that new packages or commands could get wider adoption if they are better understood by a coding agent
- similar to SEO for LLMs growing in adoption - documentation sites offering .md file raw for LLMs to ingest easily
These things will make it so that your packages or libs adopted faster
most terminal commands are being run by AI. we need a standard way to tell terminal commands to be token efficient/non-interactive
I'll keep it simple, here's Firebender, a coding agent in android studio, in 30 seconds.
really excited for the NFC support
disabling history on YouTube disables short feeds and recommendations on the home page. It takes 10 seconds and has saved easily thousands of hours and mental sanity.
Also, if you ever want to revisit a video, just use chromes history, but you'll find also this rarely happens if ever.
That is correct, and is definitely a problem. basically there should be a button that toggles for no code data leaving the network (unless its the proxy you configured). Right now you have to be aware of what features do and don't rely on the proxy which is not a good UX. At the very least there should be modal popup asking for an override or cancel on features that require our custom models.
I can solidify this option with stronger guarantees.
Separately, we're working on Soc II at the moment and should have type 1 soon, and type 2 pending the observation period.
I know trusting my word is difficult bc I'm a random person on the internet, but we do NOT store you code data or use your code data to improve our product in any way (like training models).
Which suggests that there's always some cloud component? How usable is the plugin with fully local setup?
right now requests go directly to your proxy from the plugin if you have it configured (i.e. if you set up a clean VPC/VPN network environment with no outbound requests besides anthropic): both chat, cmdk, and agent will work. We are still working on the DevX for this, but need someone to work closely on this. Enterprise is also pushing us to make this more friendly.
But there are massive downsides, we use some custom models and hosting infrastructure to speed things up. For example, code edits will take much longer.
For fully local LLMs, we just need to setup a unified API client, but there aren't any good kotlin ones and I'm scrambling to write this from scratch. It is very annoying how there are different nuances in the anthropic/openai/etc. and all the "Open source" gateways are cloud hosted. I don't think people will want to "host" a gateway locally, the best experience is to just to put your keys/base url in settings which could be localhost:3000.
just read this - that is definitely a bug. email me at kevin@firebender.com please and I can fix this bug asap for you
only the AI agent, master of all 4 ecosystems, could stop them, but...
it works with flutter. we many active flutter developers in our discord too that will flame us if we introduce any bugs for them. its good because then we can fix this asap.
same is true for intellij and kotlin/java support
didn't expect you to comment here - really appreciate the kind words! I have a ton of respect for what y'all are doing and actually making a hardware product, AR glasses. Let me know if you need anymore old Android phones for you to brick, can bike over again and hand them over :)
this is fixed now. Added Firebender Corp (registered in Delaware), with addresses. Let me know if anything else is missing!
about: https://firebender.com/about
terms: https://firebender.com/tos
privacy: https://firebender.com/privacy
wow, thanks for taking the time to go through the website and call out what you're seeing. I'm going to fix both of the About Us and address issue in privacy policy. I'll respond again after I'm done.
for some reason, I keep thinking "hugh janus" when I see the name
going from tax api reverse engineering to making it easier to reverse engineer any API is smart pivot
we're working on this right now. Expect to see auto-complete, and cmd+K functionality this week!
makes sense! Excited for y'all
launched twice? gotta launch as much as possible in this competitive landscape
My primary usage of cursor is cmd+L and cmd+K, I'll give it a shot. Also, makes sense to make an open source alternative. How do you plan on monetizing?
you can try https://firebender.com. Its specifically designed for android studio, but still works well for intellij. I use it every day to develop the plugin itself.
Intellij community edition is open source, but not the enterprise version. However, CE is pretty feature rich with a ton of users and if you're targeting indie developers, most are on CE anyways. I'd say if you do Intellij, you have to be in for the long game to build up to enterprise, but its doable.
We're targeting android studio - https://firebender.com. I imagine its because Java Swing is actually extremely painful to work with and learn. Most developers interested in AI and the experience overlap with Sun Microsystem's Java Swing is not very large.
This is pretty cool! Every time I play bullet, maybe 20% of my moves are premoves, especially in the opening. I'm not very good, only ~1300 on chess.com, but I've noticed some players as black that will play C5 against D4 opening and it gets me everytime lol because I premove C4 as a bad habit
The sample paper descriptions could be more concise. Maybe Chain of density (COD) can help here https://arxiv.org/pdf/2309.04269.pdf ?
oh shit, accidentally posted wrong link. We are at https://firebender.ai
on a second note, why can't I edit the link right after posting like ~10 min after maybe?
We are building a product analytics tool that works to understand unstructured text data and provide insights on how users are engaging with your product, what features or experiences they love, and why some users are churning.
The problem we aim to solve is the lack of understanding of the user experience with new emerging LLM applications that are heavily chat based.