Quorra — Press Release & FAQ
A PRFAQ is a press release plus frequently asked questions, written early to keep a project honest about what it is really for. This one describes Quorra as she is meant to be experienced; the milestone tracker records exactly how much is built today.
Press release
Your household finally has an assistant that works for the household — not for a data center
JUNC1 — 2026. Today Quorra goes to work in her first home: a conversational AI assistant that helps a household run its daily life, running entirely on a server in that home.
Quorra does the unglamorous, constant work of a household — logging what was spent, finding the document, remembering the wifi password, scheduling the appointment, tracking down the photo, queueing the movie. You reach her the way you’d message a person: a chat app, plain language. She takes it from there.
What’s new isn’t the chat. It’s the address. Quorra runs on Junc, an appliance in the home. The AI model, the household’s files, its photos, its finances — all of it sits on hardware the household owns. Nothing is uploaded in order to be useful. There is no account on someone else’s platform, no subscription priced per question, and no second party with a commercial reason to read the conversation.
“Every cloud assistant asks you to trust a promise — that the company won’t look, won’t sell, won’t change the terms,” said the project’s builder. “Quorra replaces the promise with architecture. The data can’t leak to a platform because there is no platform. It’s your computer, in your house, talking to you.”
Quorra is built for a household rather than an individual: each member has a private context, and the household shares what it decides to share. She remembers facts about the people she helps — and every one of those facts can be viewed, corrected, or deleted by the people themselves. She asks permission before doing anything destructive or anything that would leave the house.
Quorra is the software layer of the larger Junc project, and the first step toward The Grid — a vision of many home servers, each fully the property of the people who use it.
Frequently asked questions
What can I actually ask Quorra to do? Talk to her in plain language. Log and review spending, search and file documents and notes, manage calendars and contacts, find photos, request and pick media, set reminders, and coordinate household logistics. The rule of thumb: anything you’d normally do by opening an app should be doable by asking her.
Where does my data go? Nowhere. Quorra runs on Junc, the home server. Your files, photos, finances, conversations, and the AI model itself all stay on that machine. The only deliberate exception is the optional relay that lets you reach Quorra from outside the house — and even that carries only encrypted traffic and can be turned off.
What AI model does she use — and is it as good as ChatGPT? Quorra runs an open-weight large language model locally, on the server’s GPUs. It is smaller than the largest cloud models, and the project picks the model deliberately: fast and accurate enough for real household tasks, sized to run well on hardware a household can own. The model can be upgraded as better local models appear. The current choice and the measurements behind it are in the decision record and the benchmarks.
Can Quorra be wrong? Yes — any AI assistant can. Quorra is designed around that fact rather than hiding it. She answers from retrievable documents and citable facts wherever possible, she keeps the facts she stores visible and editable, and she never takes a destructive or outward-facing action without showing you exactly what she’ll do and waiting for your confirmation.
Several people live here. Does everyone share one assistant? You share Quorra, but not your privacy. Quorra is multi-user from the ground up: each household member has their own identity and private context, and there’s a shared household context for things everyone should see. One person’s private notes don’t surface in another person’s conversation.
Can I see what Quorra remembers about me? Yes — all of it. Quorra’s memory of facts is stored in plain, inspectable form, and you can view, edit, or delete any entry. If she shouldn’t know something, you can make her not know it. Memory you can’t audit isn’t allowed, by design.
Is Quorra always listening? No. Quorra responds when you message her. She isn’t an always-on microphone, and she isn’t watching you. The project’s stance is “presence over omniscience” — an assistant with normal, human-like limits is more trustworthy than one that silently tracks everything.
What happens when the internet goes down? Quorra keeps working. Because the model and your data are in the house, a conversation with Quorra doesn’t depend on an internet connection. Reaching her from outside the home pauses until the connection returns; everything inside the home continues.
How is this different from Alexa, Google Assistant, or ChatGPT? Those run on company servers: your requests and the data they touch go to a data center, and the service exists to serve the company’s business. Quorra runs on your server. The difference shows up in privacy, in cost (no per-question metering), in working offline, and in who the assistant ultimately answers to.
What does it run on? Junc — a home-server appliance. The first one, JUNC1, is a compact dual-GPU machine. Junc, the hardware side of the project, is covered in the Junc section.
Is it finished? No, and the project won’t pretend otherwise. Conversational chat, household memory, and finance are working now; more integrations are in progress. The milestone tracker is the honest, current list of what is done and what is next.
Why build this at all? Because the useful everyday assistant and the trustworthy private assistant shouldn’t have to be two different products. Quorra’s bet is plain: be genuinely useful at the small daily things, and be so by running in your home instead of someone else’s cloud. Adoption through usefulness, not ideology.