Getting Started with On-Device AI: RAG using ObjectBox Vector Database and LangChain

Today, we're exploring on-device AI / Edge AI, where the hardware isn’t a limit and the cloud optional.

Vector databases have transformed how we handle unstructured data like images, videos, and texts, making it searchable and more usable. Until today, however, they needed powerful servers or a cloud infrastructure.

With on-device vector database, you can bring this capability directly onto a wide range of devices, from mobile phones to IoT and embedded systems. In this tutorial, we’ll explore how you can set up your local AI tech stack to build new on-device AI use cases.

We’ll guide you through creating a Retrieval Augmented Generation (RAG) app using ObjectBox and LangChain. RAG techniques allow us to augment a language model's knowledge base actively, ensuring your AI can access and reason with your data and the very latest information. With ObjectBox you can do that, without the data ever needing to leave the device.

Stay tuned as we explore how to integrate these technologies into your projects effectively, and don't forget to check out the GitHub link below for the code: 

[adityadavid]

Krish sir you are amazing. You are sharing content faster than I can consume it❤

[SoloPlax]

Brilliant! I genuinely appreciate your work. You are under-rated man. But we love your work

[emirasa]

Great work! Quick question, how do you initialize a previously persisted store?

[rishiraj]

Sir ji, itni mwhnat kaise kar lete ho?!!😢

Sote bhi ho ki nahi?😮

Janata ko bhi batao, itni high n stable energy ka rahasya.🙏😌

[mushinart]

Question for you mr. Naik ... How and from where did the {context} get it's value?

[suchitgupta]

Hi Krish,
Can you also cover how to work with transactional data, such as a user's to-do list stored in a traditional database, and use this information as RAG

[nagalakshminimmagadda]

Hi Krish, Could you please post a video on how to query PDFs on-premises without using API keys? We are trying to implement these models in our systems, but we are encountering errors related to rate limits, such as "You exceeded your current quota." It would be greatly appreciated if you could provide some information on how to avoid these errors.

[uvhlcqq]

great video. Can you do a tutorial on how to use a local, on device opensource llm with objectbox, using flutter (ios), so that the mobile app will be completely offline?

[raymondcruzin]

Can i use ollama and this and create a mobile rag app offline? Can u create a series?

[chandrashekharchaudhari]

Hello krish.

Mai ek data scientist interview me gaya tha waha mujhe ek text latter mila aur usko sentiment analysis karna tha. Mai starting me khus ho gaya ki ye to 20 minutes me kar dunga but meri khusi jayad der nahi tiki.
Jab mai data read kiya to usme unable data tha. Jub maine unse puch ki cluster se solve karne ko to to na bole diye.

Kya aisa ho sakta hai ki hum unable data ka sentiment analysis kar sakte hai.

[HDSV]

Chat Q n A with YouTube video transcript by uploading yt link + multilingual text to speech sir make this project video