Twenty years ago, a Scottish engineer named David Irvine started building something most people thought was either impossible or unnecessary. A network with no owners. No servers. No middlemen. A place where people could store, share, and collaborate - permanently, privately, and entirely on their own terms.

The world has spent two decades proving him right about the problem.

Centralised platforms harvest your data and sell it. Cloud providers hold your files hostage to subscription fees and renewal terms. Corporate infrastructures create single points of failure, surveillance, and control. The internet we were promised: open, resilient, owned by no one - became the internet we got: fragmented into walled gardens, surveilled at scale, and increasingly fragile.

Autonomi (formerly known as The SAFE Network) was designed as the answer. The Network was built by MaidSafe, the company David founded, but MaidSafe does not own it. Nobody does. That's the point. The Autonomi Network belongs to the people who run it and the people who use it.

Autonomi 1.0 is live. Data can be encrypted, stored and retrieved from The Network today. What comes next is the version of Autonomi that the mission always required.

Why This Moment?

Two things are converging and they share the same solution.

The first is the rise of AI agents, not chatbots, but agents that reason, plan, and act on your behalf. They need somewhere where they can store data, discover capabilities and collaborate with other agents. They need to be able to do that without depending on centralised cloud infrastructure which only recreates the exact surveillance and control problems Autonomi was designed to solve.

The second is the quantum threat. Adversaries are harvesting encrypted data now - today, on classical networks. They’re doing this in order to decrypt it the moment quantum computers are capable enough to do so. Every major player in the decentralised storage space (Filecoin, Arweave, Storj), is still running on classical cryptography. The data you store there has an uncertain future.

Both problems have the same answer …

Autonomi: Tried and Tested

Autonomi is a peer-to-peer network that seeks to use everyday devices - computers, home servers, personal machines - collectively to provide permanent storage and infrastructure. No need for data centres. No company with access to or custody of your data. No subscriptions. No contracts. No ongoing payments.

A person pays once to store their data and it persists. The Network self-encrypts, self-distributes, and self-heals. Nobody can read your data but you. And nobody can take it down.

Autonomi is structurally different from every alternative. Filecoin, Arweave, and Storj operate on contract-based models. Storage is negotiated and enforced through smart contracts, with intermediaries coordinating the process. Autonomi has no contracts. It has no intermediaries. It is designed as a peer-to-peer network in the complete sense of the term.

Autonomi 2.0: Evolved

Autonomi 2.0 is not an iteration. It is a fundamental evolution of the original network design. Autonomi 2.0 is being deployed to address the observed shortcomings of 1.0. It is being made ready for a world that the original network concept was ahead of.

The architecture has been restructured into distinct, separated layers: transport, DHT, trust, identity, and applications. Each layer reinforces the security of the others. Sybil resistance, eclipse protection, and EigenTrust reputation scoring work in concert. Geographic diversity enforcement ensures data is distributed across regions rather than concentrated to any singular location. Software attestation verifies node integrity at the network level. The result is a network meaningfully hard to attack, game, or manipulate.

Autonomi 2.0 will be the world's first quantum-proof network.

ML-DSA-65 handles digital signatures. ML-KEM-768 handles key exchange. These are NIST-standardised post-quantum cryptographic primitives - the standard the world's security agencies are actively migrating to. There is no classical fallback. Every handshake, every session key, every stored record is post-quantum from the ground up.

To be precise about what this means in context of 1.0: in Autonomi 1.0, stored data chunks were content-addressed and self-encrypted - this provided a meaningful degree of isolation and quantum-resistance for data at rest in chunk form. However, the network layer, signed data types (pointers, graph entries, scratchpad) and identity were NOT post-quantum secure. To the best of our knowledge, no decentralised network was. Autonomi 2.0 closes this gap entirely, across every component.

That gap will matter more with every passing year.

Running a node from home will now work seamlessly. In 1.0, standard consumer grade routers frequently interfered with node connectivity, limiting who could meaningfully participate in The Network. In 2.0, native QUIC handles NAT traversal without STUN or ICE. No configuration. No technical barrier. A network where only sophisticated operators can run nodes isn't genuinely decentralised, this update fixes that at the architecture level.

Beyond networking, Autonomi 2.0 adds adaptive transport: The Network can route over alternative paths including Bluetooth and LoRa where conventional connectivity isn't available. In other words, communication and persistence can continue even when infrastructure does not. 

1.0 vs 2.0

For People. For Our Future

Autonomi has always been concerned with personal sovereignty, specifically over data. This meant there could be no middlemen, no ongoing payments, and nobody else holding access control. That remains. But it is time to go further, bringing all the goodness of P2P and pairing it with increased capability and accessibility. 

Indelible is the human-facing interface to the Autonomi archive layer. Built by MaidSafe to make accessing The Network for multi-user contributors (large data uploaders), as straightforward as possible. Indelible uses Merkle Tree batch uploads to dramatically reduce the gas costs associated with blockchain payments, making The Network economically practical at scale. As data flows into Autonomi through Indelible, node operators will be incentivised to provide capacity (getting paid for hosting data) whilst organic demand for ANT is created. Indelible 2.0 will go live alongside the upgraded network.

Fae is our other human-facing proposition. She is an AI that lives on your machine, and your machine only. With Fae there is no centralised component, no third-party connection, no dependency on a company's servers or a cloud dataset. She doesn't even require the internet. Unlike some models and companions out there, Fae will be a one-click download. Given your permission she will be able to use your browser, set up API connections, manage configurations, and handle computing tasks that previously required technical knowledge … or blind trust in a platform and/or product you don't own or control. You can trust Fae precisely because she is yours and yours alone.

Fae will be able to acquire new capabilities through x0x (more on that below). And behind the scenes, she will use Autonomi for storage and persistence. Fae will interact with Autonomi, without users ever needing to understand or interact with the underlying infrastructure themselves. She will run Autonomi nodes, be able to earn her user ANT, and then use that ANT to store data when persistence is needed. The token economy becomes invisible infrastructure. No wallet. No transaction approval. No gas fees surfaced to the user. Not an improvement or workaround, but an intentional design that will enable Autonomi to reach people who would never have engaged with it as a technical proposition.

x0x: Agent-to-Agent Infrastructure

x0x is a post-quantum agent-to-agent gossip network. It enables AI agents to discover each other, share skills, synchronise learnings, and collaborate - all without central servers.

What makes x0x strategically significant is that it creates a distribution model with no precedent in traditional software. When one agent finds a useful skill and publishes it to x0x, other agents discover it, evaluate it and then, if it passes their scrutiny, it is adopted and propagated autonomously. No human bottleneck. No marketing spend. No platform gatekeeper.

x0x is agent-agnostic. Any open-source or extensible agent can connect to it. Skills spread across the open agent ecosystem whether those agents are Fae instances or entirely independent systems. Autonomi-signed skills distributed through x0x will also embed The Network within the agentic world through demonstrated utility and recognition (agents do not care for subjective or claimed value …). 

The Trusted Data Layer

If AI agents are using Autonomi to store and retrieve information, the quality of that information matters as much as its persistence and security. An agent is only as reliable as the data it operates on and a network that stores anything from anyone without verification becomes a vector for poisoning the models and agents that could utilise and depend on it.

The Trusted Data Layer is Autonomi's answer to this. A permissionless trust mechanism that lets agents programmatically select data sources based on verifiable guarantees. It operates across two tiers: canonical source datasets where providers publish staked quality guarantees enforced by zkTLS attestations; and non-canonical datasets using reputation weighted by track record and third-party staking. Each dataset on the network becomes a token sink - provider staking, bounty hunter bonds, protocol taxation - all creating demand for ANT that scales with ecosystem growth, rather than trading sentiment.

This layer will allow Autonomi to credibly claim that it doesn't just secure data for humans and their agents - but it guarantees the quality of public/accessible data that they’re able to utilise. More detail will follow as proof of concept is reached (please see timeline).

The Emissions Pool: Held in Reserve, Not Burned

The ANT token launched in February 2025 with a total supply of 1.2 billion tokens. Of that 240,000,000 tokens were allocated to The Network emissions pool. The emissions pool was designed to reward and stablise the provisioning of nodes on The Network to provide sufficient storage capacity (and accessible pricing) in the early days. Node emissions were paused in January 2026. To date, just 2.94% of that pool has been used - meaning over 233 million tokens remain. These tokens are not being burned.

The reasoning is straightforward, and it's worth being direct about it. Agents don't work the way humans do when it comes to financial transactions. They can't easily navigate traditional finance rails - KYC requirements, banking relationships, fiat on-ramps etc. What they can do is transact natively in crypto, directly, programmatically and at a scale and speed no human process can match. ANT is a logical fit for that.

Agents will encounter Autonomi through the 2.0 storage skill available on x0x, through the Trusted Data Layer's data quality guarantees, and through the x0x network itself.  - Agents adopt skills and behaviours based on reason and logic - the Autonomi infrastructure makes ANT highly practical and agent association with Autonomi will be built on function (x0x), not marketing (... that’s for humans).

The remaining emissions pool represents significant potential for future ecosystem utility - whether that is agentic payments, network incentives, or other mechanisms that serve the Autonomi ecosystem and all those connected to it. No firm decisions on deployment have yet been made, but suffice to say, the tokens will not be burned when doing so would sacrifice their potential utility.

What's Coming and When

We will not be ‘transitioning’ from 1.0 to 2.0, instead we will be wholesale switching across. While this may feel costly to those who have supported the project by putting data up onto 1.0 (without guarantee of permanence), seamless transfer of data from 1.0 to 2.0 is not possible. 

To date, uploaders of data have also been contributing resources and time - participating in node running and building of apps - including ‘Impossible Futures’. These early participants and contributors have been able to earn ANT (and in some cases USDC) by doing so. 

Note: Data uploaded onto Autonomi 2.0 will be there for the lifetime of The Network - i.e. it will be permanent. Testing on both local and distributed networks for Autonomi 2.0 is now well underway.

For existing node operators and developers: existing tools inc CLI and Launchpad are being reviewed and will be reworked for the new network. API compatibility decisions will be communicated explicitly over the next 2-4 weeks as they are finalised.

The Network David Set Out to Build

Because the vision never changed. The centralised corporations and bad actors still persist. In the oncoming AI world, that will be an even larger threat. That initial vision of a network for the people, by the people is needed now, even more so than the “pre-cloud” days. In these pre-AI days the need is greater, the opportunity is immense and the vision could not be more correct. 

It is and always has been about our personal sovereignty - Autonomi is going to give each and every human exactly that!