How I use LLMs to make Indian literary works in the public domain readable

Project Akshara is a passion project to make Indian literary and historical works in the public domain easily readable. It was born out of frustration about the sheer apathy towards our cultural and written heritage. As someone who loves the written word and has gotten so much out of it, what bothered me was that we never stop boasting about our rich history and our contributions to civilisation, yet we didn't have anything like a Project Gutenberg or Standard Ebooks for Indian books in the public domain. That bothered me a great deal, and I wanted to do something about it.

The problem was that I'm not a technical person. But the thing about living in 2026 is that you don't have to be. As long as you have some common sense and the ability to type in English, you can use tools like Claude Code, Codex, and LLM APIs to extract clean and well-formatted text and publish it for anyone to read and download.

Akshara is the result of that. I won't claim I'm solving the problem — it's just my attempt to at least do something about it. As I started using these AI coding tools, it became apparent that people who care about preserving the public domain can use them to do fantastic things. What you see on akshara.ink is one such attempt. I have other ideas, but this is what I'm working on for now.

• Large language models are incredibly good at extracting and formatting text, even from horribly scanned PDFs, and this can be done at scale without spending a lot of money.

• LLMs combined with coding tools like Claude Code open up a lot of interesting use cases for extracting and transforming public domain text.

• The barrier to building with LLMs is lower than you think. Non-technical people can wrangle these tools to build real, useful things.