From CLI to Web App: Making Archimedes Usable

Once the CLI proved the workflow, the next question was obvious:
how do I make this usable for people who are not me?

The answer was a web app.

A CLI is great for power users.
A web app is better when the goal is to reduce friction.
If someone wants to type a research question, watch progress, and get a result without learning commands,
the browser wins.

So Archimedes moved into a web-based experience.

The web version gave me a better way to handle the full user flow:

• enter keywords and a question
• optionally upload a file
• choose paper sources
• set the paper count
• start the analysis
• stream progress back to the UI
• review the final report
• download a PDF

That transition changed the feel of the project.
It was no longer just a tool I used.
It was something other people could understand at a glance.

I built the app with a modern stack:

• Next.js
• React
• TypeScript
• Tailwind CSS
• an LLM API for analysis

The important part was not the stack, though.
The important part was that the interface finally matched the workflow.

I also added a few product-ish pieces that made the app feel real:

• a launch-access flow
• a BYOK unlock path for users who want to bring their own model key
• lead capture for people who are interested but not ready to pay
• PDF export so the output is something you can actually keep

The web app was the point where Archimedes stopped being a personal experiment
and started looking like a thing someone could actually use.