Check a forecast. Write some code. Keep your photos and files under your control. A modest device should be a starting point, not a limit on what you can do.

VOLPAROSSA is building an open-source, participant-operated network that brings shared connections, storage and computing to familiar applications. The aim is practical: useful AI beyond one device's capacity, private data without dependence on one always-on personal server, and tools people can inspect and help shape.

One reusable core connects the applications, so they can draw on the same shared services rather than each needing its own infrastructure. This is a Decentralized Intelligent Cooperative Network (DICN)—built around what people want to do, and what they can contribute together.

Development preview: selected workflows have been demonstrated; there is no supported end-user release.

Help build one useful next step: Find a contribution · Inspect a working milestone · Ask a question

Our purpose · The network · Applications · Our principles

Learning a new skill, understanding a document or working through a coding problem should not require owning the largest computer. Our goal is for a modest device to reach models and agents working across shared capacity, with suitable tasks divided between participants.

That is why shared AI belongs in the common core: the browser, mail client and coding assistant should benefit from the same improvements, rather than each building an isolated assistant. Dependable shared AI and protected execution of private tasks are still being developed. More participants alone do not prove better answers.

People already rely on Firefox, Signal, Thunderbird and other open-source applications. We are integrating shared services with those foundations so people can keep recognizable tools while gaining more choice over where their work and data live.

The goal includes private photo and file storage without needing to operate a permanently running server yourself. Messages, files and journeys must not become the price of admission or silently become shared training data. Complete server-independent services and their maintenance remain unfinished.

Here, sovereignty means practical control: inspect the software, understand its limits, choose how resources are shared and retain the freedom to move to other tools. It is a design commitment, not an unconditional privacy guarantee.

A verified public forecast can serve more than one weather client. A storage repair should benefit both a file library and a photo library. A better task scheduler should help every application using shared compute. Building these services once makes each contribution relevant beyond a single app.

We want to use existing devices and suitable spare capacity responsibly, with the owner's activity taking priority. Reusing eligible content and models may avoid repeated work, but transfers, recovery copies and computation also have costs. Lower energy use, lower bills and fewer data centers are goals to evaluate, not demonstrated benefits.

Joining is voluntary. Production network use is reciprocal: consumers must offer nonzero relay capacity, and nodes with a usable independent internet uplink must also offer exit service. Local-only nodes contribute their links without providing an exit. Reachable links, power and someone's real internet uplink are still necessary; sharing does not create free connectivity.

Our goal is wider access to education, independent journalism, science and culture, with resilience against arbitrary censorship by governments or platforms.

We want people to compare sources, examine evidence and encounter perspectives beyond their immediate surroundings. The hope is to support informed dialogue and reduce polarization—not to make everyone agree or claim that a network can settle every dispute.

That also means respecting creators' rights and supporting fair, sustainable access to quality media. Shared distribution is not permission to bypass copyright or paid access. The network's own restrictions must be transparent and open to challenge; disagreement alone should not be treated as abuse.

The VOLPAROSSA core brings together five complementary layers:

These are shared services, not separate networks for every application. Public caching, private storage and message delivery have different permissions and lifecycles. Resource sharing should serve the device owner as well as the wider network.

One compatible background core should serve several applications and continue its accepted work when a frontend closes. Application access and explicitly chosen system-wide operation remain separate permissions.

Private storage follows one shared redundancy policy. Its contribution rule counts the physical space used on other devices, including recovery copies and overhead: a 1 GB encrypted file with two copies requires roughly 2 GB plus overhead of space for others. If usage falls, that contribution target falls too; existing fragments must be safely handed off before occupied capacity is released. Automatic accounting and maintenance are still being integrated.

Shared AI is the default architecture, not a collection of isolated local assistants. The goal includes protected private tasks, images, audio and spoken interaction, plus autonomous research and software maintenance under the same resource and privacy boundaries. Authorized updates should eventually be distributed and installed automatically, with independent checks and recovery. These capabilities are not all operational yet.

Financial participation is separate from network contribution. Joining VOLPAROSSA does not open a financial account, invest money or authorize payments.

Shared access is not connectivity created from nothing. Devices still need a reachable connection, external internet access still needs someone's real uplink, and hardware and electricity still have costs. We are working to share that infrastructure more fairly, not promising unlimited free capacity or coverage everywhere.

The application family builds on existing open-source work rather than asking everyone to abandon familiar tools. These are integration projects at different stages of development, not finished replacements for their upstream applications.

These are independent VOLPAROSSA integrations, not official products of the upstream projects. Upstream attribution and applicable licenses must be preserved.

The project has working components and tested application workflows, but no supported end-user release. The main milestones are:

• Shared infrastructure: scoped trials demonstrate real multipath network traffic, verified content retrieval from peers, encrypted storage recovery, adapter training and public tasks on peer workers. These are component results, not a completed network service.
• Weather: Drizz.li and Overmorrow web clients display and refresh real weather and particulate forecasts delivered through protected peers. Their first retrieval currently precedes opening the UI. The installed Overmorrow Android app also retrieves real forecasts on demand through the core in an emulator, including refresh and return from the background while forecasts remain fresh. Expiry with real products, physical phones and iOS remain open.
• Code: OpenCode is now the basis of VOLPAROSSA Code on its main branch, and the native CPU backend is integrated into core main. A real model-driven trial reads and edits a code fixture, runs an approved test and passes an independent check through the core's local CPU backend. This bounded task takes about 37 minutes and uses its original pinned core revision; the model trial has not been repeated on the newly integrated core. It proves an executable workflow, not competitive performance, general coding reliability or confidential processing on other people's devices.
• Bank and Citizen: a core trial transfers signed TEST units across four validators, rejects conflicting spends and tests partitions, restart and replay after a committed transfer without a second debit. Bank currently connects to the owner-local transaction API. Citizen has an executable referendum-procedure simulation with fictional people. Neither provides a live financial service, verified real-person voting or recognized public authority.

The next shared dependencies are private-storage maintenance and protected peer compute. The complete Image/Cloud publication, repair, repeated restoration and deletion lifecycle remains unverified. Confidential execution of private tasks on other people's devices is not yet available; source checks and public guest trials do not establish that protection.

See the core development summary for the short status, the current execution plan for upcoming milestones, and the evidence ledger for exact tested versions and remaining gaps.

Technical capability is not enough. We also want fair, responsible use of digital services: generosity without exploitation, freedom with responsibility, and cooperation without surrendering personal dignity.

Ancient ethical writings and philosophical traditions inspire this purpose. We seek principles that can serve people across cultures and beliefs—not a religious membership test, a claim that every tradition agrees, or an AI appointed as an unquestionable moral authority.

The project's seven-virtue framework gives that ambition a practical vocabulary:

• Humilitas / Superbia · Humility / Pride Admit uncertainty, accept correction and avoid unsupported authority.

• Humanitas / Invidia · Humanity & Kindness / Envy Respect dignity and help others participate.

• Mansuetudo / Ira · Gentleness / Wrath Respond proportionately and de-escalate conflict.

• Diligentia / Acedia · Diligence / Sloth & Apathy Carry out entrusted work carefully and report failures honestly.

• Liberalitas / Avaritia · Generosity / Greed Share fairly within available means, without exploiting others.

• Temperantia / Gula · Temperance & Moderation / Gluttony Respect resource limits and prefer useful work over endless consumption.

• Castitas / Luxuria · Chastity / Lust In the project's broader interpretation: respect consent, dignity and personal boundaries; reject exploitation and respect people's private lives.

The intended shared immune system applies these principles through mutual review across all five layers, their applications and software maintenance: agents examine work, compare evidence and help correct harmful or unreliable behavior. Consequential decisions should have a clear scope, be explainable and remain open to correction. Privacy must not become a pretext for abuse, and safety must not become a pretext for silencing disagreement.

This oversight is still being developed. No automated system can be assumed infallible, and a signature does not make a judgment true. The project is not a promise of unrestricted access or exemption from applicable law. Current principles and policy boundaries are documented openly.

You do not have to wait for a finished product to help shape it.

• Build something people can use. Developers, designers, translators and educators can improve one integration, clarify a guide or remove an accessibility barrier. The contribution guide explains how to choose a bounded task; documentation work does not require running the network.
• Put an assumption to the test. Researchers can examine a recorded workflow, challenge a privacy or resource-sharing assumption, or propose a fair comparison. Bring evidence and limits to the project discussion; interest is welcome without a partnership commitment.
• Help the work reach the right people. Supporters and prospective users can explain a real need, review an explanation or share a milestone with its qualifications. The supporting guide covers useful help. Financial sponsorship remains pending; joining the network requires no payment, token or investment.

Working components and scoped trials already exist, but there is no supported end-user release. Complete applications, dependable shared AI, confidential computation on other people's devices and automatic resource maintenance remain in development. Do not use development builds as your only backup or rely on them to protect sensitive activity.

Explore the core · Documentation guide · Frequently asked questions · Contribute · Community guidelines · Accessibility

Our aim: more people able to connect, more freedom to build, and more of the benefits shared.