Track market-moving stories before they get noisy

Singapore has already decided that AI runs on trusted access to people's data. At the Tanglin Club, hosted by the MIT Club of Singapore, the question from senior business leaders and government officials was how.

Singapore's refreshed National AI Strategy makes one of its ten priorities to unlock datasets, underpinned by strong data governance. The strategy says the challenge is no longer making more data available, but giving the right people trusted access to the right data, with safeguards, for the four national AI Missions in manufacturing, financial services, connectivity and healthcare. Parliament has already gone further. The Personal Data Protection Act carries a data portability obligation: a person can ask an organisation to send their data straight to another, in a form the receiving organisation can read. The right exists in law. The method is what the country is working out.

Vana is the open network where personal data moves with the permission of the person it belongs to, and a record of every move. Vana wrote the standard for how personal data moves, now hosted by the Linux Foundation Decentralized Trust.

We sat down with the MIT Club of Singapore at the Tanglin Club, with senior business leaders and government officials around one table. The conversation kept returning to three things.

Keeping data under Singapore's jurisdiction while letting it move A country's data is most useful when it stays under its own laws and its citizens' consent, and can still go where it is needed. On Vana a person grants a specific use of their data, for a specific purpose, for a specific time. The grant is recorded and can be revoked. The data moves without leaving the protections the country has set. That is AI a country runs on its own terms rather than on whichever platform holds its people's context.

Opening the market to new players When data is trapped where it was created, only the incumbent can build on it. When people can move their data, anyone can compete to serve them better. Singapore's economy is built on being open to new players. Its data infrastructure should be too.

Fitting with the identity infrastructure Singapore already has Singapore has some of the most advanced digital identity infrastructure in the world. Identity says who you are. Vana says what you permit. Together they let a person carry their data into any service they choose, and take it back when they want.

Where Vana fits The standard Vana wrote does what Singapore's portability obligation asks for: a person's data goes from one organisation to another, with their permission, in a form the receiver can use, with a record of what was agreed. Because the standard is open and owned by no single company, a regulator can point to it and a company can build against it. Governments are already scoping national implementations. The Vana network is where that data moves today: more than 100 apps built by its community, 1.5 million users and more than 20 million transactions.

Singapore is investing more than S$1 billion in public AI research through 2030. That investment pays off fastest where AI can draw on real context from real people, moved with their permission, on infrastructure the country controls. We look forward to continuing to work with sovereign governments and with organisations like the MIT Club of Singapore to lead the way data moves in an AI world.

Vana comes out of the MIT Media Lab, a place that has spent thirty years building strange, decentralised things before anyone else thought they were sensible. People were wearing computers there in the 1990s.

It is the same school of thought that Tim Berners-Lee, the inventor of the web, brought to MIT with the Solid project, which set out to give everyone control of their own data online. It is the school of thought that shaped the Vana protocol.

Vana carries one idea from it. The value of a person's data should go to the person who created it. Your data stays with you. An app or an AI reads it only with your permission. Every read is recorded, and you can take the permission back at any time.

Most people know Vana for the first way it put that idea to work: Data Collectives, where people pool their data into sets AI could not get any other way, and share in what the set earns. That is aggregation, and it serves the crowd.

Underneath aggregation sits the thing the network was built on at MIT: portability. A person's data moves, under their permission, with a record. Aggregation is portability done by many people at once.

Portability on its own is one person's data moving to an app or an AI that person chose, and being used for them. Your health records to your doctor's assistant. Your purchase history to the agent doing your shopping. That serves you.

On 21 September Vana completed the Vega upgrade, a rebuild of how the network handles permissions, so that data portability runs at the speed AI needs.

Aggregation built the network. The market asked for portability itself. Aggregation is how Vana was bootstrapped: 1.5 million users, more than 20 million transactions, and more than 100 apps and data pools built by the community. All of them are still running and still earning.

Then the market moved, and it moved towards portability itself. The models got good. Once every model could reason well enough, what set a useful answer apart was no longer the model but its context: what it knew about the person asking. And that lives in the services where the person created it.

AI agents arrived. An agent acting for you reads your context, the things about you that make its answer fit you, before a task and writes the result back afterwards, under a permission you gave once, thousands of times a day. Every one of those interactions is a data transfer.

People started carrying their own context. When people left ChatGPT for Claude this year, some opened the Vana app, connected their ChatGPT history, and Claude knew them on day one.

Builders and companies followed. More than 40 data portability apps shipped through the Vana Cup, the launch campaign for the data portability beta, within a month. Bush & Noble, a yacht brokerage, made its inventory readable to a buyer's AI on Vana.

Institutions needed a way to implement data portability, not just a policy that says they should. So Vana wrote one. The Personal Data Portability Protocol is the first open standard for how personal data moves, and Vana contributed it to the Linux Foundation to host. Regulators can point to it and companies can build on it, which makes Vana the gold standard for portability. Governments in several regions are already scoping national implementations of Vana.

All of them want the same thing. The Vega upgrade moves data, under users' permission, as fast and as simply as on the platforms they already use.

What the Vega upgrade does We committed to the Vega upgrade in our January roadmap, when the first signs of how much portability would matter were already showing. Portability is built on two core concepts: grants and operations.

A grant is a permission: who can use which of your data, for what, and for how long. Issuing it, using it and revoking it are each recorded, so there is always an answer to "who was allowed to read this, and when".

An operation is an answer computed from your data and shared without the raw data. An app can ask how you slept last week and get a number, with a record of where it came from, and never hold the week of sleep data.

Vana is the network behind both. It records every grant and every revocation, so the permission is always on the network and never in an app's hands. It holds the keys, so nothing, including your own personal server where the data is computed, can decrypt your data without a valid grant. And it moves the data when the grant says it can.

After the Vega upgrade, Vana does all of that in about 50 milliseconds end to end in production. That is fast enough for an agent to ask, be allowed, read and act in one motion.

An app can also pay the fees for its users. So a person can use an app built on Vana without holding tokens or paying per read, and never needs to know there is a chain underneath.

What keeps the network running Every read of a person's data under a live grant carries a fee. Reads grow with how much the network is used, not with how many people sign up, so the network's income grows with use.

VANA is the token the network runs on. Fees are paid in VANA or in stablecoins such as USDC, and people stake VANA, meaning lock it up, behind the operators, the businesses that run the machines carrying the data. Those fees fund the operators and the people who stake behind them: stakers are paid from the fees the network earns. The supply of VANA and the rate at which new tokens are issued are unchanged by the upgrade.

The next two weeks The upgrade is the start. Over the next few weeks, we will release further details on what it means for stakers, and then for the community that got Vana here. Keep an eye on our X account or join our Discord for early announcements: X, Discord.

Read it How Vana works, the original whitepaper and today's addendum are at vana.org/learn.

The addendum, The Vega Upgrade: Data Portability and Transformations, sets out the grants, operations, encryption and fees in full. The original whitepaper is at vana.org/papers/vana-whitepaper-dec-2024.pdf.

Every major AI added memory in the first half of 2026. OpenAI's Dreaming. Claude Chat Memory. Gemini Personal Intelligence. Grok Skills. Microsoft's M365 Copilot Memory rollout. Five launches, five more walled gardens.

Each one is a retention feature. Your context lives on their servers, serves their product, and stops at their wall. If you move to a different AI tomorrow, or use two at once, you start from zero.

Until now. Vana was built to give you an exit from walled gardens so that your data belongs to you. Today, the Vana App Upgrade is live in Beta. And the memory layer inside it is now something you own.

What is the Vana App upgrade? The Vana App upgrade gives you a personal data server, on your device, that you control.

Connect your data sources once. Your Spotify listening history, your Oura sleep and recovery data, your calendar, your conversations across platforms. That data lives locally, not on any platform's servers. It's yours.

From there, you decide what it serves and to whom. Grant a permission, revoke it anytime. No platform intermediary, no asking anyone for access to your own context. Your data becomes self-sovereign.

You can permission your data to any app built on Vana's Data Portability API, or port your memory using MCP.

Portable Memory MCP Vana has acquired the team behind Memory Protocol to lead these important upgrades. Jack Spallone has joined the Vana team and brought his deep know-how and expertise on portable memory into the Vana stack.

The Vana App upgrade ships an MCP endpoint for your personal server. That means Claude can read from it. ChatGPT can read from it. Any MCP-compliant tool can read from it. Your memory and context are now portable, from one source you own, across every AI or app you use.

This is what we mean by open data infrastructure for human-grounded AI. Portability as a protocol.

For builders Vana's Data Portability API now makes it possible to ship apps that read from a user's Vana personal server with their permission.

Your users bring their own context to your app. You don't need to build memory infrastructure from scratch. You don't need to ask a third-party platform for access to user data. You can ask the user directly.

Start building, or add personal data portability to your app today. The docs are at docs.vana.org. We will be holding Builder Workshops and Office Hours in Vana's Discord throughout the week, so be sure to tune in.

How to try it Try the Beta version of the Vana App Upgrade at app.vana.org.

Those who try it out and offer feedback will be given priority slots for the Full Release.

Interested in building on it? Visit docs.vana.org. For workshops and questions, join our Discord.