This is a submission for the Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine

What I Built

Echoes of the Forgotten is an AI-powered historical exploration application that helps people discover, understand, and reconnect with stories from the past.

The idea behind it is simple: history is full of stories that deserve to be remembered, but sometimes they get buried in old documents, forgotten records, and scattered information.

I wanted to build something that could bring those stories a little closer.

Users can upload historical documents or paste text and let AI help them explore the material through summaries, evidence, uncertainties, timelines, and follow-up questions. There is also a creative feature that generates fictional echoes inspired by the source material, clearly separated from historical facts.

For me, the theme Ghost in the Machine represents something interesting: the past may be silent, but technology can help us hear its echoes again.

Demo

Youtube Video:

Code

GitHub Repository:

Built with Python, Gradio, and Groq.

How I Built It

I built Echoes of the Forgotten using Python and Gradio for the interface, with Groq powering the AI responses.

The application allows users to:

• Upload PDF, TXT, and Markdown documents or paste text.
• Generate summaries of historical material.
• Explore supporting evidence and uncertainties.
• Build timelines from the provided content.
• Ask follow-up questions about the material.
• Generate fictional creative interpretations, clearly labelled as fiction.
• Export structured results as JSON.

For the AI layer, I used the open-weight openai/gpt-oss-20b model through Groq.

I wanted AI to be more than a chatbot attached to a webpage. The model is central to how the application interprets documents and helps users explore their contents.

Why Does Open Innovation Matter?

This is probably one of the most important parts of the project for me.

Open innovation matters because AI should not be limited to a handful of closed platforms or controlled entirely by services that developers cannot inspect or adapt.

Using an open-weight model gave me more flexibility in choosing the model behind my application and building around it. It also gave me a clearer path towards experimenting with different inference providers and potentially running the model locally in the future.

There is still room to improve: my current version uses Groq's hosted inference, so it requires an internet connection and does not yet offer fully offline processing.

But that is exactly why open models matter. They leave developers with more choices. The same application can evolve towards local inference, different hosting providers, or other compatible models without having to rebuild its entire purpose around one closed AI service.

For a project about preserving stories, that flexibility feels especially meaningful. I don't want the ability to explore and understand those stories to depend forever on a single service.

My Agent Session

[Add your DevRelay agent session link here, if available.]

The Spirit of It

Imagine uploading a document that has been sitting unread for decades.

A few moments later, you have a timeline, a summary, questions to investigate, and a better understanding of what the document contains.

The document hasn't changed. Its words were always there.

What changed is that someone can finally explore them.

That is the idea behind Echoes of the Forgotten: sometimes, the most interesting stories aren't new ones. They are the ones we almost forgot.

Prize Categories

[Add the applicable partner prize categories here, or remove this section if none apply.]