Cherry Studio is my daily go-to, I hope Onyx deskktop can be a great alternative for personal users who just want a dedicated app to access any LLMs with full power of MCP and various tools
Great article! I always feel that the choice of embedding model is quite important, but it's seldom mentioned. Most tutorials about RAG just tell you to use a common model like OpenAI's text embedding, making it seem as though it's okay to use anything else. But even though I'm somewhat aware of this, I lack the knowledge and methods to determine which model is best suited for my scenario. Can you give some suggestions on how to evaluate that? Besides, I'm wondering what you think about some open-source embedding models like embeddinggemma-300m or e5-large.
I'm not against AI-generated userscripts, but I wish to review and edit the scripts more easily. Currently, the extension hides the code too much and doesn't provide a decent editing experience.
By saying "under your control," I believe it means you are not locked in by cloud providers. Since it's easy to switch and back up between different blob storages, if one is down, your data remains accessible and manageable.
Reminds of this one [1] that I read half a year ago, which I used to develop my first agent. But what fly wrotes is definitely easier to understand, how I wish it was written a year earlier.
Probably because they have to deal with customer feedbacks and bugs on existing system. The stress would be huge, making a day like a week long, I guarantee
This kinda resonates with me. I've been using Python for over a decade and the only async method I trusted was gevent, and since I moved to Go after 2016, I never want use Python async in production level project, even I came back to see how it goes almost every year.
But since Python's LLM ecosystem is so well, I really appreciate the courage it takes to migrate to Node when writing a RAG system. I've tried similar things recently, working on a document analyzing project using React Router as the full-stack framework, while put some ETL related work on the Python side, use inngest to bridge Node and Python services. In this way, I got the benefit of Node for LLM chat, while stil able to facilitate Python's SOTA ETL libraries.
Very thoughtful and inspiring suggestion, I think we are on the same road, which really makes it much clear to me. Looking forward seeing the Pipelex solution for these kind of problems.
I'm using the React Router v7 in framework mode, fetching data directly from the database in server-side components. So far, it works reasonably well as long as we avoid mixing server code with client components. However, although it offers the convenience of writing frontend and backend code together, the mental burden added by this approach makes me feel it is not worth it more and more often.
I noticed the author mentioned the stack they are currently using is NextJS + Hono, and according to the code in the blog, Hono is used as the API backend to provide data for NextJS through calls like `fetchUserInfo`. This is where I really got confusing, why is it designed like this? NextJS is already a full-stack framework, not to say you used RSC, you can directly get user data from the database in NextJS. If you decide to use Hono for HTTP API, which is fine because it's more light weight and adaptive, then why RSC rather than making the frontend a SPA? Even if you use NextJS to write SPA it would looks much reasonable to have both NextJS and Hono used together.
Recently I'm working on a project that let the users to analyze and rate the bid documents based on one tender document. The general logic is quite alike to the cv-job_offer example in Pipelex. The challenge I encountered is that both tender and bid documents are very large, it's impossible to load even a single one in LLM context, not to say both. Therefore I have to design a workflow to extract structured informations and evaluation criteria from the tender doc into predefined data models, and then dispatch multiple tasks to complete evaluation of these data models against each bid doc.
I'm wondering if this kind of scenario (essentially it's just the input document is too big) is possible to be handled in Pipelex. In my understanding DSL is good for it's high-level abstraction and easy to understand, but lacks flexibility and the power is restricted. How can the users of Pipelex iterate the pipelines to fulfill the complex need when the business logic became complex inevitably?
I love this concept! I have always believed that the old methodologies used in NLP and statistics can be better and faster than new LLM technologies like embeddings, depending on the scenario. Will the code be open-sourced someday? I'm thrilled to learn from it.
While I largely agree with the philosophy, the example provided is not very practical. The code snippet `getExpiredUsers(db.getUsers(), Date.now())` is unlikely to occur in real-life scenarios. No one would retrieve all users and then filter them within the program. Instead, it should be `db.getExpiredUsers(Date.now())`.
We should never be too extreme on anything, otherwise it would turn good into bad.
At first glance, it seems to be a satire of vibe coding, but after reading it a second time, I find that the author is more sarcastic towards those who think AI will replace programmers. Did I get it right?
Sorry if you feel my words mean. I'm not criticizing the joke, I just think it could make a better title, even for a joke, by adding quotes around "database" or calling it a "storage service," since it does allow data to be read. A good joke is both entertaining and difficult to deny.
This kinda like a PoC for ZK Proof used in digital devices, however, I don't think a Raspberry Pi in a 3D printed case should be made a real product, it lacks actualy use cases. Honestly, I like this concept, but I think it should belong to a personal art exhibition or DIY competition…
Business logics in PostgresSQL? Impressive, never think of using it this way but it looks actually pretty good! Last time I saw similar concept is at https://spacetimedb.com/ , maybe one day PostgresSQL can do the same, since it's already powerful enough to support web apps this way.
I use a speech-to-text tool every day for vibe coding. I have been using GPT-4o-transcribe for a while, and I feel that this could be a game-changer. I will start using it now to find out how it really performs compared to OpenAI's models.
When I made the script httpstat [1] 9 years ago, I had this exact thought - if I want to show the statistics of a HTTP request, why not just use curl, why bother working out the figures myself? And since then, the more I use curl, the more I find it robust, sophesticated and irreplaciable. It's the only thing and everything I need.