TL;DR
I read widely and made copious notes. I then started experimenting with LLMs and retrieval-augmented generation (RAG) using those notes. I realised there was a strong overlap with the way UPSC/PSC preparation works. I eventually tooled this RAG pipeline into a chatbot experience for aspirants (covered in the other posts in this series).

For links to all the posts in this series, see Aanae.

The Starting Point

I joined Sequoia Capital in 2021. As part of the onboarding kit, I was given Measure What Matters by John Doerr. That book introduced me to goal-setting through OKRs (Objectives and Key Results).

One of my OKRs from that year looked like this:

OKR Example

Around this time, I realised I had slowly lost touch with my reading habit. I also felt a persistent urge to revisit my schoolbooks. Somehow, the curiosity that had been exorcised from my system through the vagaries of cramming and periodised testing came alive again.

I wanted to read:

  • History
  • Geography
  • Polity
  • The sciences

Anything that would help me understand the world better.

Side note: I had always assumed that you sharpen your understanding of the world as you grow older. My experience proved otherwise.

Early Attempts

I made well-laid plans:

  • Organised my reading material
  • Created an Excel sheet to track progress
  • Calculated Weekly Estimated Time (WET) to complete the NCERT syllabus in a year
  • Tried to balance this with work, life, and new interests (this was also when I discovered endurance sports and ultracycling)

I failed.

The Excel sheet gathered digital dust as work and life took over. I made incremental attempts to read more—with some success—but nothing felt structured or sustainable.

The 2025 Pivot

By 2025, during my last year at Sequoia, my interests began trending in a different direction.

Over the preceding three years, I had lived through what felt like epochal events:

  • ZIRP
  • The crypto boom
  • The advent of GenAI

I increasingly felt the need to slow down and just… study and learn.

The overarching goal was to understand:

  • Data analysis
  • Systems architecture
  • AI and LLMs

But I needed a gentler on-ramp. Something that would both rekindle the habit of studying and rebuild my ability to sit and focus.

I chose to start with NCERT books.

The Study Phase

  • By March 2025, I had completed the entire NCERT syllabus.
  • I then branched out to:
    • NIOS (National Institute of Open Schooling)
    • IGNOU (Indira Gandhi National Open University)
  • By May 2025, my urge to “study” had been largely sated.

More importantly, I now had copious notes across geography, history, polity, and science.

Detour: My Note-Taking Process

Taking notes was non-negotiable.

I knew I wouldn’t retain information beyond the event horizon of my memory. I needed a system that allowed me to revisit and refresh any granular topic quickly.

My evolution looked like this:

  • Handwritten notes
    • Too slow
    • Scaled poorly over long texts
    • My handwriting and note density were not conducive to quick comprehension
  • Typed notes
    • Better structure
    • Still slow

Eventually, I decided to augment my note-taking with ChatGPT.

Final Workflow

  • Use PDFs of books
  • Read an excerpt or chapter
  • Feed that text to the LLM
  • Prompt for structure, clarity, and coverage
  • Edit the output
  • Save with appropriate taxonomy in OneNote
    (This last choice turned out to be a mistake, as discussed later.)

Where I Ended Up

By the end of this phase:

  • I had (for now) satisfied my desire to study the foundational subjects that explain the world.
  • I had accumulated ~250 MB of typed notes, amounting to roughly 250,000 pages.

I began to think of these notes explicitly as my second brain.

The Aha Moment

The turning point came when I realised:

I could feed my notes into an LLM, use it as a daily driver for curiosity and queries, and let it paper over the cracks of my memory and recall—subject, of course, to how well I handled hallucinations.

That realisation was the seed from which Aanae grew.