Everyone needs their personal AI knowledge sidekick. I built my own.
Empowering curiosity with personal AI knowledge sidekick.
The need for an AI knowledge sidekick
Having my personal AI knowledge sidekick has always been a dream of mine. From the days of reading sci-fi in high school to studying AI/ML over two decades ago at the University, I've been looking forward to my own AI knowledge sidekick. A true AI partner.
And its even more critical now. The volume of information is increasing at a breathtaking pace and its accelerating. Making it harder for the super curious. Its both the most exciting times and stressful simultaneously.
If you're anything like me, your curiosity leads you to subscribe to lots of newsletters, follow experts on X/twitter, save articles, Reddit scroll, listen to podcasts & watch YouTube videos, all in the pursuit of staying sharp. Just the content from Lenny Rachitsky is more than 20 hours long, and he keeps posting maximally exceptional content every week!
We have better content, better archives, better search bars, and even better RAGs (e.g., ChatGPT connected to your documents) & apps, such as NotebookLMs and LennyBot, to manage and learn from it all.
But it feels like we're still lacking something. Its not just about finding information faster; it's about having intelligent systems that help us understand all the stuff we’re consuming on a deeper, more connected level.
An itch I had to scratch
This isn't just a hypothetical for me.
My Top 5 CliftonStrengths or StrengthFinder (a better name, tbh) profile is a combination of Learner, Input, Analytical, and Ideation. Meaning I am personally driven to gather and connect information.
But this also means that I am subscribed to way too much stuff. Remember the 20+ hours of Lenny’s content? And I’ve had to whittle this down, and this is just the tip of the iceberg.
Thankfully, I found some solace in Readwise Reader1, where I can funnel almost everything I want to read or watch in one place. Its been enjoyable to read / watch and take personal highlights in one place. I even forward my fav newsletter emails or subscriptions to Reader, so it all funnels to one beautiful app.
But the Reader app also created a paradox:
The more I collected to fuel my learning (
Learner), the less I could effectively process. Hindering my ability to analyse (Analytical), and Connect (Ideation).My own nature was at war with my capacity.
I needed something that kept me at the bleeding edge, ideally, without going insane.
An AI sidekick that reads and reasons for me
So I rekindled my dream, my crazy thought experiment about having my own personal knowledge AI sidekick. And I spent a couple of weekends designing and building it.
The system is designed to be far more than a simple summarisation engine. It runs a multi-stage intelligence pipeline:
It ingests every content I save in Reader.
It analyses each one for key tags, entities, and critical insights and extracts them.
It clusters all these tags to discover emergent themes.
It synthesises deep-dive briefings on each theme, complete with citations.
The output isn't just "less to read." It's a deeper, richer & more connected level of understanding, delivered when I need it.
One nice side effect is that my processed Reader content is now beautifully tagged, analysed, and embedded. Meaning it's interconnected, and can do even more cool stuff in future.
Why Now?
What was once a daydreamy thought experiment is now a reality, thanks to the accessibility of powerful LLMs.
With the Google Gemini API's generous free tier2, I realised that building this sophisticated system was not only possible but practical.
The era of the personal AI knowledge sidekicks is here. I built mine, and I am super excited!
In my next post, I'll share the architecture, how I stitched together Streamlit, Gemini, Readwise Reader, and a smart processing pipeline to build my AI knowledge sidekick.
Stay tuned!
Oh, one more thing, I am open-sourcing the repo, so everyone can build and leverage it.
About me
Hi, I am Suhel.
I am writing to jot down things I’ve learned and noticed on my journey. My posts aren’t intended to be polished, shiny pieces of writing; consider them works in progress. They may be interesting to you; if so, enjoy them and feel free to share if you found them valuable. Feedback is always welcome, please let me know.
https://readwise.io/read. Smooth reader, native apps, and highlights are also consolidated to Readwise for regular review. Must have knowledge junkies.
https://ai.google.dev/gemini-api/docs/rate-limits. e.g., Tagging uses Gemini Flash-lite, which has 1000 free calls per day!