Possibility of No-Code Creation of Free Bhagavad Gita AI Knowledge Base with basic chat interface using Google NotebookLM but with free tier limits

Last updated on 18 Mar 2026

A friend had asked me in the past about a Bhagavad Gita AI app. We also saw that one public (and well-known) Bhagavad Gita AI app is no longer working, which I suspected to be due to subscription/hosting costs for the AI.

My recent post: Possibility of No-Code Creation of Free AI Knowledge Base with basic chat interface using Google NotebookLM but with free tier limits covers the topic of the same name in general. This post deals with the specific case of Bhagavad Gita AI knowledge base creation.

Note: I am only sketching out the possibility now; I have not created a Bhagavad Gita NotebookLM as of yet.

Once the Bhagavad Gita source material—including Sanskrit verses, transliterations, translations, and commentaries by a single author in English (to stay within the limits of the free tier for sources)—is uploaded as sources (in the form of Google Docs or PDFs) to a notebook in NotebookLM, the AI should be able to perform natural language queries against that specific data. For more details about this process, see my above mentioned post. Note that this process does not need any software development work and so is a 'no-code creation' process.

But free tier NotebookLM has a big limitation of 50 sources for a notebook. 

Google AI Pro plan user created Bhagavad Gita can use up to 300 sources

Note that if a Google AI Pro plan user creates a Bhagavad Gita NotebookLM notebook, he can use up to 300 sources. Further, as per Gemini, this Bhagavad Gita notebook with up to 300 sources can be publicly shared (by setting notebook access to 'Anyone with the link') and then Google AI free tier users will be able to access all the sources the notebook has—even if it exceeds the 50-source free tier limit.

This is possible due to Google reference page for NotebookLM stating, "Sharing a notebook does not change the source limit for any collaborator." For more details, please see my blog post section: Note: Pro User Created 'Public' Notebook with over 50 sources is Fully Viewable By Free Tier Users.

How to overcome the 50 sources limit for a free tier user created notebook

For a NotebookLM free tier user creating a Bhagavad Gita notebook, since the Gita contains 18 chapters and over 700 verses, uploading each verse or even each sub-commentary as a separate file would quickly exhaust the 50-source limit of the NotebookLM free tier.

To overcome this, we could use a Data Consolidation Strategy:

1. Understand the "True" Capacity: While you are limited to 50 files, each file can contain up to 500,000 words (roughly 1,000 to 1,500 pages of text).
2. The "Chapter-Level" Merge: Instead of uploading 700 individual verse files, combine all verses, translations and commentaries for a single chapter into one Google Doc or PDF.  Note that as mentioned earlier in this post, we are considering single-author translations and commentaries to limit the source data. This 'Chapter-Level' Merge approach reduces the source count from 700+ down to just 18.
3. Additional Source Documents: If you wish to include commentaries and translations of additional authors, or additional glossaries, you could create additional documents within the 50 documents limit. 
4. Preserving Searchability: Even when combined, NotebookLM’s "Source Grounding" remains precise. The AI can still point to a specific page or paragraph within a large 500,000-word document, so you do not lose accuracy by merging files.
5. Automating the Consolidation: The consolidation process can be automated using Google Colab with Gemini AI within it (Data Science Agent - DSA). By providing Colab Gemini with a suitable prompt, it can generate and run a script to merge individual verse files into the 18 chapter-level documents mentioned above. Similar approach can be taken for other source consolidation work like mentioned in 'Additional Source Documents' step above.
• If the Colab Gemini created scripts work without errors, this effectively would be a "No-Code" automation pipeline for the data preparation. However, the user needs to know how to use Google Colab to provide input files, inspect output files and save output files to persistant storage.
• Note that even if initial Colab Gemini DSA runs create output files with errors, it can be prompted about the error and it will try to correct the error - it acts like a coding agent and not just a coding assistant.
• The user need not study and understand python code generated by Colab Gemini DSA though doing so would be helpful for debugging errors, running it again in future and documentation.