How Archimedes Started: A Research Tool I Built for Myself

Harshal Rudra — Sun, 12 Apr 2026 08:05:28 +0000

How Archimedes Started: A Research Tool I Built for Myself

I did not start Archimedes because I wanted to launch a SaaS.
I started it because I was sick of doing the same research ritual over and over:
search a paper, open five tabs, skim abstracts, save PDFs, forget half the useful bits,
and then try to stitch everything together into something that actually answers the question.

That sucked.

Archimedes began as a very personal fix for that exact mess.
I wanted something that could help me search academic papers, analyze them with an LLM,
synthesize evidence, and spit out a clean answer I could actually use.

The first version was not elegant.
It was just me trying to make one painful workflow less painful.
But that is usually where the good stuff starts: not with a business plan,
but with a tiny recurring annoyance that refuses to shut up.

The core idea was simple:

search academic sources
identify the most relevant papers
analyze them with an LLM
collect evidence in a structured way
export the final result as a PDF

That is Archimedes in one sentence.

I built it because I wanted to stop doing research like a caffeinated raccoon,
opening random tabs and pretending I had a system.
I wanted one place where a question could become a useful answer.

The funny part is that the more I worked on it, the less it felt like a toy.
The pipeline started making sense.
The workflow became repeatable.
And once something becomes repeatable, it starts becoming a product.

This is the first post in the story of Archimedes.
The next ones are about the ugly middle bits: looking at existing solutions,
building a CLI, showing it to friends, and slowly realizing the thing had legs.

DEV Community: Harshal Rudra

How Archimedes Started: A Research Tool I Built for Myself

How Archimedes Started: A Research Tool I Built for Myself