TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Show HN: Cognita – open-source RAG framework for modular applications

142 点作者 supreetgupta大约 1 年前
Hey HN, exciting news! Our RAG framework, Cognita (<a href="https:&#x2F;&#x2F;github.com&#x2F;truefoundry&#x2F;cognita">https:&#x2F;&#x2F;github.com&#x2F;truefoundry&#x2F;cognita</a>), born from collaborations with diverse enterprises, is now open-source. Currently, it offers seamless integrations with Qdrant and SingleStore.<p>In recent weeks, numerous engineers have explored Cognita, providing invaluable insights and feedback. We deeply appreciate your input and encourage ongoing dialogue (share your thoughts in the comments – let&#x27;s keep this ‘open source’).<p>While RAG is undoubtedly powerful, the process of building a functional application with it can feel overwhelming. From selecting the right AI models to organizing data effectively, there&#x27;s a lot to navigate. While tools like LangChain and LlamaIndex simplify prototyping, an accessible, ready-to-use open-source RAG template with modular support is still missing. That&#x27;s where Cognita comes in.<p>Key benefits of Cognita:<p>1. Central repository for parsers, loaders, embedders, and retrievers. 2. User-friendly UI empowers non-technical users to upload documents and engage in Q&amp;A. 3. Fully API-driven for seamless integration with other systems.<p>We invite you to explore Cognita and share your feedback as we refine and expand its capabilities. Interested in contributing? Join the journey at <a href="https:&#x2F;&#x2F;www.truefoundry.com&#x2F;cognita-launch" rel="nofollow">https:&#x2F;&#x2F;www.truefoundry.com&#x2F;cognita-launch</a>.

14 条评论

Jianghong94大约 1 年前
Congrats on the launch!<p>I find it relevant to what I want to do next and put in some time to understand the application vs other stuff e.g. Langchain. And if my understanding is correct, what this tries to do is:<p>For a lot of typical web services, there&#x27;re non-realtime batch-processing data processors, e.g. search engine&#x27;s crawler and indexer, or database&#x27;s OLAP system, Hadoop, spark, etc. Once their processing is done, they will output data in arelevant, easy-to-use form for real-time web services to consume, e.g. search engine&#x27;s index, or a list of e-commerce&#x27;s best selling items.<p>If we extend such analogy to today&#x27;s LLM RAG application and compare it with an out-of-the-box Langchain or LlamaIndex implementation, we&#x27;ll realize everything is in one process altogether. Of course, for demo purpose, they have to.<p>Cognita tries to fit in by splitting the process into real-time and not real-time parts, on top of existing LangChain and LlamaIndex, and comes with an API endpoint for each part and a web UI for user querying.<p>For my use case, I&#x27;m looking into setting up a very basic RAG-based internal doc QA app, to see if this helps with some of our notoriously bad wikis. So I&#x27;m likely going to use this UI and just shovel whatever simple LangChain or LlamaIndex implementation into it. I&#x27;m not that interested in the modular design. Honestly, I could see a couple of different ways each market segment approaches such a problem: for demo&#x2F;mainly static document&#x2F;low stake application, the need to periodically refresh vector-db is non-existent; for companies with enough engineering expertise, they&#x27;ll likely put the data processing part into existing data processing framework; for the rest segment, they probably can also get away with putting the whole offline data processing into a very long python script, setup cron and call it a day.<p>---<p>I haven&#x27;t look into RAG in a year or so, but my overall sensation is this: 1. the RAG layer (on top of vector-db) isn&#x27;t technically difficult, vs say OS development, database development, etc, after all, text manipulation has been around since 60s. 2, since the whole LLM generation is very sensitive to prompt, an early, too rigid abstraction likely do more harm than good.
评论 #40185388 未加载
评论 #40190844 未加载
magaton大约 1 年前
Hello, a very interesting project. Conratulations for putting everything together. I have expressed some thoughts in the discussion sections of Cognita github repo: <a href="https:&#x2F;&#x2F;github.com&#x2F;truefoundry&#x2F;cognita&#x2F;discussions&#x2F;146">https:&#x2F;&#x2F;github.com&#x2F;truefoundry&#x2F;cognita&#x2F;discussions&#x2F;146</a> It would be great if the maintainers could reply.
评论 #40190634 未加载
dmundhra1992大约 1 年前
Congratulations on the launch! Will give this a try!<p>We were looking for a solution that would help our team test out the LLMs &amp; prompts for repeatability and identifying edge cases.<p>The UI looks interesting, like a playground on top of the RAG framework, allowing the team to test out various prompts &#x2F; configurations to handle edge cases, without requiring a lot of tech bandwidth!
评论 #40190833 未加载
parentheses大约 1 年前
Looks like a great product. I&#x27;ll have to give it a try!<p>I like that the product seems to solve the RAG need only and not be an &quot;everything framework&quot; for LLMs. It makes for a richer seeming product for RAG while making other aspects of AI apps open for the user to choose their approach.
评论 #40184661 未加载
johnea大约 1 年前
Whatever you do, never say &quot;free software&quot;!!!<p>That &quot;freedom&quot; stuff is commonism...
评论 #40207370 未加载
ComputerGuru大约 1 年前
Does a &quot;web&quot; data source only scrape the individual page or linked pages as well? I&#x27;m <i>assuming</i> the former. What would be the least painful way to ingest a knowledgebase (say a wiki-like site) from the web?
评论 #40190623 未加载
TechSageWow大约 1 年前
This product appears to be promising. I&#x27;m intrigued to test it out. I appreciate that it focuses solely on addressing the RAG requirement and doesn&#x27;t attempt to be a one-size-fits-all solution for LLMs.
评论 #40190641 未加载
hiteshvyas11_大约 1 年前
Interesting, is there any feature roadmap for future reference ?
评论 #40211867 未加载
sagarpandey1大约 1 年前
Congratulations and good luck.Will give this a try!
评论 #40190853 未加载
esafak大约 1 年前
Many of the links are broken and lead to <a href="https:&#x2F;&#x2F;www.truefoundry.com&#x2F;cognita-launch#" rel="nofollow">https:&#x2F;&#x2F;www.truefoundry.com&#x2F;cognita-launch#</a><p>I tried on Firefox and Chrome.<p>I would make the GitHub link more prominent.<p>Congratulations and good luck.
评论 #40184038 未加载
namanyayg大约 1 年前
Congrats on the launch Supreet! Can you talk about how Cognita compares against competitors like RAGFlow?
评论 #40184680 未加载
vivek0203大约 1 年前
Congratulations on the launch. I am building GenAI application. Will explore it.
评论 #40207365 未加载
b2bsaas00大约 1 年前
What’s best practice to integrate this in a Ruby on Rails application?
评论 #40184363 未加载
评论 #40184532 未加载
评论 #40184312 未加载
adastra22大约 1 年前
What is RAG?
评论 #40183842 未加载