This is a submission for the Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine
What I Built
Ever opened a document full of notes, forgotten ideas, or information you saved months ago, only to realize that finding the important parts is almost as difficult as reading the whole thing?
Meet LazyDigest: a local-first AI summarizer that helps bring the important information buried in lengthy text back to life.
You can paste text or upload a .txt or .md file and get:
A short summary to understand the bigger picture.
Key points to recover the most important information.
Action items to identify tasks and next steps.
Important dates to bring deadlines and events to your attention.
Whether it's study notes, project documentation, personal writing, or information you've forgotten about, LazyDigest helps you rediscover what matters without having to reread everything from scratch.
The idea fits the Ghost in the Machine theme in a slightly different way: sometimes, the things we need most are already buried in our files. We just need a little help bringing them back.
Demo
Source code: GitHub — LazyDigest
LazyDigest runs locally and connects to a local Gemma inference server.
A public hosted demo is not currently available. To try the application, follow the setup instructions in the repository's README.
Code
Explore the project:
The project uses a straightforward architecture built around Python, FastAPI, and a lightweight browser interface.
How I Built It
The core of LazyDigest is Gemma, an open-weight AI model running locally through llama.cpp's OpenAI-compatible server interface.
The application uses:
Python and FastAPI for the backend.
HTML, CSS, and JavaScript for the user interface.
HTTPX to communicate with the local inference server.
Gemma to generate summaries, key points, action items, and dates.
Chunk-based summarization to process longer inputs in smaller sections before combining the results.
The workflow is simple:
Paste text or upload a supported file.
The application validates the input.
Longer text is divided into manageable chunks.
Gemma processes the text locally.
The application presents the resulting digest.
I wanted to keep the project practical and understandable rather than build a complicated system around it.
Why Does Open Innovation Matter?
For me, one of the most interesting things about open-weight AI is the ability to run it on your own machine.
Documents aren't always just documents. They can contain personal notes, unfinished ideas, private conversations, family stories, or information that shouldn't casually end up on someone else's server.
With LazyDigest, the model runs locally, so the text can be processed on the user's machine rather than sent to a third-party AI provider.
Once the model and dependencies are installed, the application can work without an internet connection, provided the local inference server is running.
That gives users several advantages:
More control over their data: The application doesn't need a hosted AI service to summarize text.
Offline capability: Summarization can continue without an internet connection after setup.
Freedom to experiment: Developers can explore different compatible models and change how summarization works.
Less dependence on external services: The core workflow doesn't require a paid hosted AI API.
A closed API can be convenient, but it also introduces a dependency on an external provider and its pricing, availability, and data-handling policies.
Open innovation gives developers the freedom to build tools that they can inspect, adapt, and run themselves.
For a project that processes potentially personal information, that freedom matters.
Prize Categories
Best Use of Gemma — using an open-weight Gemma model as the core of a locally run summarization application.
Final Thoughts
The Ghost in the Machine theme made me think about how much useful information we leave behind in our files.
Ideas get buried. Notes get forgotten. Stories sit unread.
LazyDigest is a small attempt to bring that information back into focus, using AI that users can run on their own machines.
No elaborate infrastructure. No required hosted AI API. Just your text, a local model, and a little help finding what matters.
Sometimes, the most useful thing AI can do is help us remember what we already know. 👻
Top comments (0)