Hacker Newsnew | past | comments | ask | show | jobs | submit | aunch's commentslogin

Anshul from the Codeium team here! As our FAQs clearly state, only Chat currently uses OpenAI APIs - our flagship Autocomplete capability (think free Github Copilot) uses proprietary in-house trained models and no APIs. We are upfront about this and because we want to guarantee users’ code security preferences, if you have disabled code snippet telemetry, we actually do not allow you to use Chat for this exact reason - we cannot guarantee what OpenAI does with your data, and we never want to accidentally expose this.

We are working on building our own models for Chat to remove this dependency, so you can use Chat while still guaranteeing privacy.

Oh and none of this affects Autocomplete, so downloading Codeium and disabling code snippet telemetry will still give users tons of value :)


So your users are or aren't sending code fragments to your servers?

It doesn't matter if the model is OpenAI's or yours. If inputs are source code and it runs on a server outside the company, it's a fireable breach of security protocol to use it.


From the Codeium (https://www.codeium.com) team here!

Exciting to see more teams take on the code search problem - modern pure regex definitely is a suboptimal solution.

Just to objectively highlight some of the similarities and differences between Bloop and Codeium Search (Codeium also provides Autocomplete) to a user:

- Both are free for individuals

- Both use a mix of AI and regex

- Bloop has both free local and paid cloud offerings, Codeium only works locally (but therefore doesn't work as well on super large codebases)

- Codeium uses custom proprietary embedding models so no dependency on OpenAI and no code is sent to a third party AI


Posting about your company when relevant is one thing, advertising it another’s launch thread is another, and it’s pretty gauche… especially when, in this very thread, one of your testimonials is saying he doesn’t actually use your product: https://news.ycombinator.com/item?id=35236557


sorry! on a previous post, someone encouraged me to just post when Codeium was relevant, didn't realize the norms around launch threads.

and wrt the other comment, it is totally accurate - changing developer behavior for these search products is tough! we welcome any ideas :)


I disagree, I am very happy that parent posted this. I want to know about other tools in this space, and the comment contains a nice summary of pros and cons, some of which are extremely relevant to me!

Bloop looks awesome too, don't get me wrong, and I'll check it out.


I don't see the issue with this. It happens on most of these Launch threads and is a common occurrence. I don't think there are rules that prohibit it either.


There's no formal rule against it but it's often done in bad taste. When people overpromote their thing in someone else's launch thread, we tend to scold them. Someone else's launch is not a great place for competitive promotion—each product or project deserves its day in the sun. On the other hand, users like to discuss alternatives and comparables, and that seems healthy.

I'd say the sweet spot is somewhere between just leaving the competitor's launch thread alone, or, if you must, then (1) mention it once and stop there; and (2) when you have any relationship with the alternative thing, disclose it.

That's the sort of thinking we apply in practice but wouldn't make a formal rule out of, partly because it's always evolving, but mainly because we don't want the list of rules to be too long. If we tried to codify all such things we'd end up with a bureaucratic list of hundreds of rules—ugh!


Why do I care if they have custom proprietary embedding model or not? Why is OpenAI any less trust worthy than Codeium? I imagine people care about quality of results more, and I'm doubtful Codeium's proprietary embedding is better than GPT4.

Also most of your comment history is self promotion for Codeium which is not a great look.


that's fair - the point was less about the model and more about the fact that your code is not sent anywhere because it all happens locally

ill start commenting on other things - i usually just upvote on things that i find interesting but dont know enough to comment on


Just to be clear, since you say it's local, when you say no code is sent to a third party AI that includes Codeium? For search and indexing, code is not sent over the network at all?


yes. indexing and search happens on your machine, nothing sent over network


Codeium already has this enterprise option, with self-hosting, fine-tuning, and even things like search on top of autocomplete and way more IDEs: https://codeium.com/

How are you different?


aren't you codeium?


Yes, thus the question :)


Don't try to just blithely "smile" your self promotion away. Your question is a valid one, but it's pretty obviously morally dubious to mention Codeium without disclosing that you're involved with them.


this is really cool and well done. i will start using it in conjunction with Codeium's plugin (free Copilot alternative): https://github.com/Exafunction/codeium.vim

for transparency: I'm from the Codeium team, and we are big fans of getting this AI gen tech to all developers on all IDEs for free - we've also open sourced an emacs plugin: https://github.com/Exafunction/codeium.el


How are you supposed to use two plugins providing the same functionality? Are you gonna compare them and then uninstall one of them? Do they differ?

As far as I can tell, it seems like you just wanted to plug your own plugin, but I'm happy if I'm wrong. I just to understand how you can use two plugins that provide the same functionality and why you would do that.


they're complementary! codeium works like copilot, constantly autocompleting what you want to type next, which helps a lot when you already have a decent idea of what you need to get done and need to get it done faster. chatgpt (and chat interfaces in general) help you get some ideas on what to do when you are doing more open-ended exploration, and is explicitly invoked. as developers, we all code in these different "modes" and different tools can be built for each!


HN encurage you to mention your own products when relevant, so you don't have to wrap it in, just put it bluntly.


exciting! we built (no pun intended) something similar as a free functionality in Codeium (https://www.codeium.com/waitlist/codeium-search) - we just soft launched so anyone on VSCode can opt in to it

it's cool to see other attempts as well. natural language search done properly can definitely accelerate developers by a ton


Codeium has a self hosted enterprise solution that gives the enterprise the ability to fine tune on their repos within that self hosted instance!


if you want an actual enterprise solution with in-customer-tenant/on-prem hosting, check out Codeium (https://www.codeium.com/enterprise)

disclaimer: i'm from the Codeium team. but really, we will even ship you a physical box if that level of data security is important to you


I tried using the playground code completion to ask it to write a script for pyautocad that colors in a grid, it completed with a different library, pyautogui. Even after saying "import pyautocad" myself, the function it completed was pyautogui.

I'm sure it's prob still very useful for people who care about the privacy tradeoff, but I've had more success with ChatGPT


Thanks for the feedback! Yeah the playground is a bit limited as it is for demo purposes.

We are fans of ChatGPT and think that ChatGPT is pretty complementary to tools like Copilot and Codeium. ChatGPT is helpful for longer form exploratory questions from natural language while Codeium in its current form is great to accelerate your coding.


No worries haha, not at all easy nor cheap to build these things. I'm sure it's useful in most cases.


you’re going to need to ship that emacs extension if you want to keep advertising on HN :-)


haha we'll have it out polished in the next week or so :P


That plus lisp / clojure on the roadmap is exciting :) I'll definitely give it a try when it comes out. The thing that slows me down is that I actually don't like to get visual feedback while I'm typing. I'm curious if codeium has a good way to compromise there


with copilot.el, i flip it on only when i’m doing something repetitious where i know it’s likely to give me a useful suggestion. then i flip it off. works well for me


I looked around your web site and I thought about trying out your product, but one feeling never stopped nagging me: even though I'm not in a large organization, I need absolute assurance that the AI is trained only on our code and permissively licensed open source software (like MIT or BSD.) Also, whenever it uses permissively licensed code, I need a complete list of everything it based its work upon so I can declare the relevant licenses.

Without that, I can't even entertain the idea of using an AI code tool for anything but private projects that I don't share with anyone.


Exactly this. Also, even the "our code" case if not done carefully may copy code from one or more internal projects that had in turn copied with attribution from an Open Source project, and fail to propagate the attribution.


What model do you use? CodeGen?


our own!


Are there plans to support the full Visual Studio IDE?

Edit: Also, Notepad++ support would be awesome


VS is on the roadmap, Notepad++ isn't something currently on the roadmap, but we'd be totally open if someone wants to write an open source plugin for it, just like we did for Vim/Neovim (https://github.com/Exafunction/codeium.vim) and are planning on doing with Emacs!


To each their own! We were motivated to blog about it by others who have had issues like this: https://daverupert.com/2023/02/solved-the-case-of-the-bing-b...

And at the end of the day, we got a lot of great SEO tips here. Thanks HN!


I understand the sentiment and why it might come across that way, but we genuinely have gotten a lot more reasonable pointers in these comments than we were able to rack our brains about. And it seems like SEO attacks, Bing incompetence, and Bing actually going after some terms (like Chrome) actually have happened, and the main point here was that we genuinely don't get any feedback from Bing.


> but we genuinely have gotten a lot more reasonable pointers in these comments than we were able to rack our brains about

So the ends justifies the means?


oof that's rough. good to know what it took for you though, thanks!


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: