🚀 Introducing Doks – The AI-Powered RAG Creation Tool! 🤖📄
Context is key in AI, and that’s exactly what Doks is built for!
With Doks, you can:
✅ Create multiple contexts for structured AI interactions
📂 Upload and manage documents within each context
💬 Engage in intelligent, document-aware AI chats
🤓 You can choose the OpenAI model you want to use for chat
🧠 Within a context, you can pick specific documents for a chat
📑 You can define how Doks should behave by giving your own context instructions
Built using LlamaIndex, OpenAI, and Pinecone, Doks leverages the best in AI and vector search to deliver powerful, context-aware chatbot experiences.
Whether you're a researcher, business professional, or developer, Doks helps you build AI-driven conversations tailored to your knowledge base.
@masump We are storing document id in the metadata of the indexes in Pinecone. So when you remove a document we just delete indexes with the metadata with the deleted document's id. And we you add a document we just add new indexes.
The tool is powerful but incorporating version control for document updates could really help future proof it for heavy users who frequently change their data sources.
The idea of intelligent document based AI chats is super exciting. Does the platform support multi document querying like being able to ask a single question across several uploaded files at once?
@tyler_wright1 Currently, encryption isn't implemented yet, but it's a top priority and we're actively working on adding encryption for both storage and processing.
How does your system manage large document uploads? Does it automatically break them down into chunks or does it require some pre-processing on the user's part?
@olivia_johnston1 No pre-processing is needed on the user’s part! Users can directly upload PDF or TXT files up to 500MB, and we handle the chunking automatically in the backend.
@sophia_watson3 Good question! We haven't tested this scenario yet, but it's definitely something we'll explore as we continue improving Doks. Thanks for highlighting it!
Is your system tailored for specific industries like legal, healthcare or finance or is it built to be versatile across various domains right from the start?
@samuel_morris Doks is built to be versatile across any domain, not tailored to a specific industry. You can upload any file, but it currently works best with text-based data. If a PDF has a lot of images, it may not extract information from them accurately.
@shreyans_assistiv Thanks! Currently, if documents are changed, you won’t be able to continue old chats, but you can still view all previous conversations. You'll need to start a new chat for updated documents.
Replies
@jithinlalk25 How does Doks handle updating or syncing documents after they’re modified?
@masump We are storing document id in the metadata of the indexes in Pinecone. So when you remove a document we just delete indexes with the metadata with the deleted document's id. And we you add a document we just add new indexes.
The tool is powerful but incorporating version control for document updates could really help future proof it for heavy users who frequently change their data sources.
@max_lincoln Good catch, thanks for pointing it out! It's on our list to implement soon.
Is there a limit on the number or size of documents that users can upload when creating their RAG contexts?
@manday_m_kramer Currently, for each context, users can upload up to 5 files (PDF or TXT), with each file up to 500MB.
The idea of intelligent document based AI chats is super exciting. Does the platform support multi document querying like being able to ask a single question across several uploaded files at once?
@william_rivera3 Yes! Within a context, it has access to all uploaded files, so you can ask a question across multiple documents at once.
When it comes to sensitive uploads how is document security managed?
Are the files encrypted both when stored and during processing?
@tyler_wright1 Currently, encryption isn't implemented yet, but it's a top priority and we're actively working on adding encryption for both storage and processing.
Can the AI remember previous interactions with documents, or is each conversation treated as a brand new start?
@wyatt_dior For a particular chat, it remembers all previous messages and builds context. But starting a new chat is treated as a fresh conversation.
How does your system manage large document uploads? Does it automatically break them down into chunks or does it require some pre-processing on the user's part?
@olivia_johnston1 No pre-processing is needed on the user’s part! Users can directly upload PDF or TXT files up to 500MB, and we handle the chunking automatically in the backend.
In situations where documents contain conflicting information how does the AI determine which context to prioritize when generating responses?
@sophia_watson3 Good question! We haven't tested this scenario yet, but it's definitely something we'll explore as we continue improving Doks. Thanks for highlighting it!
Is your system tailored for specific industries like legal, healthcare or finance or is it built to be versatile across various domains right from the start?
@samuel_morris Doks is built to be versatile across any domain, not tailored to a specific industry. You can upload any file, but it currently works best with text-based data. If a PDF has a lot of images, it may not extract information from them accurately.
How quickly does the document indexing happen once a user uploads their files?
@kamil_riddle Indexing happens asynchronously after file upload and mainly depends on the amount of text data, not the file size or page count.
Brilliant concept can users organize their uploaded documents into different projects or workspaces within the platform?
@musa_fischer Thank you! Yes, users can create multiple contexts and upload files to each context separately, organizing their work as needed.
Can users export their chat history along with links back to specific documents?
@julian_baker4 Currently, we don't have this feature, but it's a great idea and definitely something we'll consider adding!
Love the control over context—super important. How do you handle versioning when documents change but older chats still need to stay consistent
@shreyans_assistiv Thanks! Currently, if documents are changed, you won’t be able to continue old chats, but you can still view all previous conversations. You'll need to start a new chat for updated documents.