This is a submission for the Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine
What I Built
An offline-first digital ofrenda for the stories we refuse to lose — built for the Ghost in the Machine open-source AI challenge.
Demo
Code
Ofrenda.
An offline-first digital ofrenda for the stories we refuse to lose — built for the Ghost in the Machine open-source AI challenge.
Why open innovation matters
Memory is intimate infrastructure. Ofrenda keeps its archive in the browser by default and optionally connects to an Ollama-compatible local model (llama3.2) on localhost. That means a family story can be transformed without creating an account, sending data to an unknown server, or paying per request. Because the model bridge is deliberately small, it can be swapped for any local open-weight model and the ritual's voice can be changed by the builder.
It also works with no model and no internet: a tiny deterministic fallback keeps the core experience useful offline.
Run it
No build step is required. Open index.html in a browser, or serve the folder:
python -m http.server 8000
For the optional local model:
ollama pull llama3.2
ollama…How I Built It
I have build it with github copilot and i have used ollama llama3.2 as localllm.i have embedded with web for easy serve
Why Does Open Innovation Matter?
Memory is intimate infrastructure. Ofrenda keeps its archive in the browser by default and optionally connects to an Ollama-compatible local model (llama3.2) on localhost. That means a family story can be transformed without creating an account, sending data to an unknown server, or paying per request. Because the model bridge is deliberately small, it can be swapped for any local open-weight model and the ritual's voice can be changed by the builder.
It also works with no model and no internet: a tiny deterministic fallback keeps the core experience useful offline.
The Spirit of It
Ofrenda is a quiet, private place for the people and stories we refuse to lose.
It turns a small memory—a familiar laugh, a favorite song, the smell of someone’s kitchen—into a living digital offering. Instead of sending those intimate stories to a distant server, Ofrenda keeps them in the browser and optionally uses a local Ollama model to reflect on them. No account, no cloud, and no loss of control.
The moment it lands is when the screen responds to an ordinary detail with something that feels deeply personal: a candle lights, a reflection appears, and a memory that might have faded suddenly has a place to rest. Even without AI or internet access, the ritual continues through its local fallback.
Ofrenda is not trying to recreate the dead or speak for them. It simply helps us pause, remember, and keep their presence alive in the details we carry forward.
Top comments (0)