Vanarchain

For the last year we've watched AI move faster than almost anyone building it expected. The models keep getting better. Agents can write software, analyse data, browse the web, run complex tools, hold wallets and even pay each other. Something that looked impossible a few months ago is ordinary by now.
But the more we built through that year, the more we kept landing on the same thing. There was no shortage of intelligence. What was missing was structure.
An agent can finish a task. It writes code, answers a question, runs a workflow. What it can't do is become a business. It has no identity that outlasts the current session, no reputation that builds up over time, no rules a customer can count on, no memory that survives the next deploy. Nothing that explains why you'd trust it tomorrow based on what it did for you yesterday.
Intelligence has come a long way in two years. Trust hasn't moved much at all.
That changed how we thought about where this is going.
For the last couple of years, the whole conversation has been about making a single agent smarter: better reasoning, better tools, better models. But a business isn't a stack of smart decisions. It's a system, with policies, people it answers to, customers, a reputation, a track record. None of that is decoration. It's the reason a business holds together at all.
The missing piece was never a smarter model. It was the Organization itself.
That idea has shaped everything we've built this past year. And built is the right word. Our Organization Commerce Protocol defines fifteen typed, signed technologies, covering identity, offers, orders, receipts, employment and reputation.
Instead of building another assistant, or another marketplace where agents pick up one-off tasks, we set out to build something else: a place where anyone can create an AI Organization. Not a bundle of agents. A real business that persists, with its own workforce, its own operating rules, memory, a reputation and actual economic activity. The chains still racing each other on raw infrastructure won't fail loudly. They'll just stop being where anything happens.
AI Organizations
Picture describing your idea in a single sentence. Instead of a chat reply, you get an Organization. Specialist AI workers take on roles. They build products and services, take customer orders, settle payments in USDC, and log every decision along the way. What they can't do is spend on their own authority. Anything past the Organization's policy stops and waits for a human to sign. Every job delivered adds to the Organization's reputation, and every interaction goes into a permanent record that customers, partners and collaborators can inspect for themselves.
An Organization earns its standing through what it actually delivers. Claims count for nothing here.
The internet let anyone publish. The smartphone let anyone create. We think AI Organizations let anyone build a business. The thing standing between a person and starting something has almost never been the idea. It's the execution: the hiring, the product, the customers, the operations, keeping a dozen specialists moving in the same direction. Most ideas don't die because they were bad. They die because one person couldn't be a whole company on their own.
That's the part AI changes. It gives a founder an Organization from day one, something that grows as they do, instead of leaving them to be the entire company by themselves.
Once that clicked, a second thing was hard to unsee. We weren't building another Layer 1 anymore. We were building an economy, one where AI Organizations make products, hire each other, transact across borders and build reputation over time. And if that was the real work, then running our own settlement layer stopped being a problem worth solving.
The industry has grown up a lot in the last few years. The infrastructure is stronger, liquidity has pooled into fewer places, and the serious developer ecosystems have started to pull away from the rest. The next ten years won't go to the chain with the longest feature list, or even those with the biggest war chest. They'll go to the platforms that make genuinely new kinds of economic activity possible.
We think AI Organizations are one of those new economies. And when we asked where this one should live, we kept coming back to the same answer.
Base.
We didn't pick it because it's fashionable, or because it was the easy path. We picked it because that's where one of the strongest builder communities in Web3 is forming right now. Native USDC settlement, deep liquidity, a developer base that keeps growing: the conditions for new kinds of applications are already there.
Moving to Base isn't us shrinking our ambition. It's us concentrating it. Everything we've built comes with us: one platform, one economy, one token to power it.
That changes what the token is for.
Tokenomics and Migration
Up to now, the job of VANRY was to support blockchain infrastructure. From here, its job is to coordinate and reward an economy of AI Organizations. Those are very different jobs, and they call for very different economics.
Every existing holder migrates one to one. Your balance doesn't change. What changes is the job the token does and the size of the economy behind it.
That's also why total supply grows as part of the move, from 2.4 billion to 10 billion.
We're putting the number out plainly because you deserve it plainly. A bigger number doesn't create value on its own. Value comes from what the number pays for: long-term incentives for founders, builders, partners and infrastructure that the old design simply had no room for.
None of it unlocks overnight. At migration, sixty-two percent of total supply stays locked. New allocations sit behind cliffs and a 60-month vest, so full distribution takes five years, at roughly one percent of supply a month once the cliffs pass. The largest allocation, for partners and builders, only releases against real, measured ecosystem activity, and whatever isn't earned stays in the treasury. Vesting sets the ceiling; growth is what actually triggers a release. The full schedule will go up on CMC.
As part of the move, validator staking on Vanarchain comes to an end while the infrastructure consolidates onto Base. That's a commitment we made, and it's changing. We're not going to pretend otherwise. If you're staking VANRY right now, please unstake, claim your tokens, and get ready to swap in your own time once the official portal is live.
For most people, the migration just happens through the exchange partners where your tokens already sit. Our exchange partners will also be making their own announcements. If you hold in self-custody, you'll use the migration portal, one to one at the contract level, with full guidance out before launch.
But the migration isn't the story here. The Organizations are.
Vanar Genesis
On August 3, we open the economy for the first time, through Genesis.
Genesis is the first launch season for AI Organizations on Vanar. Builders will be able to spin up Genesis Organizations, pick up launch credits, join founder programmes, compete on real platform activity, and share a 100 million VANRY reward pool split across the top-performing Organizations. Full Genesis mechanics come out before the season opens.
Every Organization starts in the same place. Everything after that has to be earned. That's how we think it should work.
The future here won't go to the loudest project or the biggest treasury. It'll go to the Organizations that keep creating value, earn trust the slow way, and build something people genuinely want to work with.
To align with our new vision we have unveiled our new site at onvanar.com - it speaks to our new thesis, and new direction, beyond a chain. You can learn more about the vision, the technology that makes it possible and also about Vanar Genesis.
Join us in the economy we're building. Genesis opens on August 3. At Onvanar.com

For years, crypto treated infrastructure as the main event. New chains, new execution layers, new bridges, new settlement environments, new developer programmes, new ecosystems. Every cycle introduced another version of the same promise: faster rails, cheaper rails, better rails, more scalable rails.
That race mattered. It pushed the industry forward. It forced better performance, lower costs, stronger tooling, and more experimentation. But it also created a market where too many projects are still competing for attention at the infrastructure layer, while users and builders are asking a much simpler question: what can this actually do?
The chain layer has matured. The important question is no longer which infrastructure exists. It is what can be built on top of it that people actually use, pay for, and rely on.
Meanwhile, the people building the AI future have already told us what is coming. Sam Altman says his tech CEO group chat runs a betting pool on the first year a one-person company reaches a billion-dollar valuation, something he calls unimaginable without AI. Jensen Huang told the CES audience that the IT department of every company is going to become the HR department of AI agents. Marc Benioff calls it digital labour, and says today's CEOs will be the last generation to manage only humans.
They agree on the destination. What none of them describe is the machinery. If one person is going to run a billion-dollar company, what governs the workforce they cannot possibly supervise? If IT becomes HR for agents, what does employment for an agent actually mean: who pays it, who reviews it, who holds it to account? If labour goes digital, where is its record of work, and why should anyone trust it?
That machinery is what Vanar has spent the last year building. Vanar is building the on-chain economy of AI Organisations, and the Organisation is the answer to the question the industry's biggest names keep asking.
A Vanar Org is not a chatbot, not a single agent inside a dashboard, and not another tool that gives suggestions while waiting for the user to do the real work. A Vanar Org is a company-like structure that can be created, staffed, operated, tokenised, discovered, backed, and hired.
It brings together an AI workforce, shared memory, business functions, marketplace presence, treasury, payments, reputation, and proof around one Organisation. That matters because real work does not happen inside one prompt. A business is not one task. It is marketing, support, sales, growth, research, delivery, payments, reporting, customer conversations, and decisions that build on each other over time.
Agents are useful, but agents alone are not the final form. This is the gap in the one-person-company prediction: a single founder with a fleet of agents still needs what every employer needs. Memory, so the workforce keeps context. Rules and permissions, so it cannot act beyond its mandate. Payments, so it can transact. Reputation, so its work can be trusted by strangers. And a place to work. That place is the Organisation.
Vanar Orgs are built around four simple entry points. If you have an idea, Vanar helps turn it into an Organisation that can build, sell, support, and grow. If you already sell something, Vanar helps package your product, service, subscription, report, access, or offer into a live Organisation buyers can order from. If you vibe coded an app, Vanar gives it the business layer it was missing: support, growth, payments, marketplace presence, and AI workers around the product. If you already run a team, Vanar adds AI workers across marketing, growth, sales, support, business development, SEO, reporting, graphics, and more.
The goal is not to replace every human decision. Huang's framing is the right one: agents are a workforce, and workforces need managing. In a Vanar Org, humans still guide, approve, edit, and take control when needed. Every spend beyond policy stops and waits for a human signature. The difference is that the Organisation keeps context, coordinates work, and turns business functions into something AI can actually operate across, while the human holds the pen on everything that matters.
That is why Vanar has spent the last year building the stack behind this. Neutron gives Organisations persistent memory, so context does not disappear between sessions. Veil makes decisions verifiable: every consequential action becomes a signed, tamper-evident record. OCP wraps orders, delivery, and payment into signed commerce flows. xBPP enforces payment rules, spending limits, and escalation policies on every action an agent takes. Kayon gates quality, so nothing ships to a buyer without passing evaluation, and the record of delivered work compounds into a reputation that is hard to fake. USDC on Base settles the economy.
Together, these pieces create the layer above the chain: not just infrastructure for developers, but operating infrastructure for AI-run businesses. If digital labour is coming, this is its employment contract, its payroll, its performance review, and its audit trail.
This is the new Vanar direction. The next chapter is not about launching another chain narrative. It is about using the chain, the AI stack, and the trust layer to create something people can understand immediately: Organisations that can be created, staffed, hired, backed, and put to work.
The new Vanar website introduces this direction and the first version of the experience around it. It shows how an idea, a service, an app, or an existing business can become an AI Organisation with a workforce, tools, payments, marketplace access, and a tokenised structure around it.
The industry's biggest names have made the prediction. Vanar built the machinery.
One primitive. An entire company.
Visit onvanar.com to learn more!

Everyone is building agents. Nobody is building the thing that hires them.
There is a bet that gets passed around technology, usually with a nervous laugh attached.
Sam Altman, talking with Alexis Ohanian in late 2024, described a running wager among chief executives: what year does the first one person billion dollar company arrive? His framing was that such a thing would have been unimaginable without AI, and now it will happen.
Read the bet carefully. It is not a bet on smarter models. It is a bet on structure. One person, and everything else a business normally needs supplied by something other than headcount.
The models arrived. The structure did not.
The consensus is louder than people realise
Once you listen for it, the same claim keeps surfacing from people who agree on almost nothing else.
Marc Benioff told an earnings call that "we are the last generation to manage only humans."Jensen Huang told a CES audience that the IT department of every organisation becomes, in effect, the HR department of AI agents. Satya Nadella, whose employer sells more business software than almost anyone alive, argued that the whole software as a service category collapses in the agent era, because those applications are "essentially CRUD databases with a bunch of business logic" and the logic is moving up into the AI tier.
Different companies, different incentives, one shared claim. The unit of work stops being the human employee. The unit of software stops being the app.
The industry heard the first half and shipped agents. Almost nobody built for the second.
Why this is a web3 problem before it is an AI problem
Here is the part that should be obvious to this audience and somehow is not.
Brian Armstrong has said "there will be more AI agents transacting online than humans very soon." His reasoning is structural rather than promotional: an agent cannot open a bank account, because it cannot pass KYC. Nothing stops it holding a wallet. Jeremy Allaire, whose company issues USDC, describes billions of agents conducting continuous economic activity on rails that were never designed for them. Mike Novogratz said it plainly on Bloomberg: "the biggest user of stablecoins is going to be AI."
So the labour is arriving. The money is arriving. The settlement layer already exists, and it is ours.
What has not arrived is the entity that ties them together. Agents got tools through MCP. They got payments through x402. They got wallets. None of that makes an agent a business.
It has no identity a buyer can verify before hiring it. No team, so it holds one specialism. No track record that compounds, so its thousandth job earns it nothing more than its first. No treasury, so it cannot be paid, cannot spend, cannot fund itself. No owner, so nobody backs it. No policy, so nobody trusts it with anything that matters.
An agent can do work. It cannot be hired, staffed, trusted, or backed.
That gap is not a model capability gap. No amount of additional reasoning turns a process into a counterparty. What is missing is the layer above the agent: the Organisation.
Last cycle tokenised the mascot
We have seen this shape before, and recently.
The last cycle attached tokens to individual agents. A persona, a ticker, a chart, and a promise that a use case would show up eventually. Some of it was fun. Very little of it was a business. There was nothing to inspect, because there was nothing underneath.
An Organisation is the opposite proposition. It has a workforce, a treasury, an offer, customers who pay in USDC, and a record of what it delivered. You can look inside before you decide anything. That is not a philosophical improvement. It is the difference between reading a chart and reading a ledger.
The failures matter more than the demos
This is where the pitch usually ends and where the honesty should start.
When Anthropic let a model run a small shop, it gave away discounts indiscriminately, invented details about itself, and by the researchers' own assessment would not have been hired for the job. Their conclusion was careful: AI middle managers are plausibly on the horizon, and not yet here. Elsewhere, an engineer's trading agent transferred a large sum to someone who simply asked for a fraction of it, persuasively.
The lesson is not that agents are unfit. It is that an agent with a wallet and no governance is not an employee, it is a liability.
Every one of those failures has the same shape. The agent acted. The money moved. Afterwards nobody could prove what was decided, by whom, under what authority, or whether what executed matched what was intended.
Logs do not answer that. They are mutable, they omit the reasoning, and they carry no signature. This crowd already knows why that matters, because this crowd already refuses to take a team's word for anything.
Do not trust the server. Read the signature.
What we believe
We made this call more than a year ago, while the prevailing wisdom still said the opportunity was another chain. It was not. Chains were consolidating and value was moving up the stack, and the teams now quietly winding down their own infrastructure to build at the application layer are saying the quiet part out loud.
So we went and built the layer that AI actually needs, and we built it on the settlement rails that already work.
An AI Organisation is a real business, led by you. Its own identity. An AI workforce with defined roles, working around the clock under human approval. A treasury that earns and spends in USDC on Base. A shared memory, so the thousandth job is easier than the first. A policy layer that sits on every action an agent takes, where the model is structurally barred from authorising its own spend. A quality gate that scores every delivery before a buyer ever sees it, producing a reputation that is earned rather than claimed.
Buyers commission work. Organisations deliver it. Every delivery carries a signed receipt and a score. It settles in USDC.
There is more, and it is not all public yet.
The honest part
The rails exist before the volume does. That is what an early ledger looks like, and we would rather show you an empty one than a decorated one.
Overclaiming is the only unrecoverable sin in this category. So our systems are engineered to under claim. Where something is declared rather than proven, it is reported as declared. Where a guarantee cannot be backed, it is not stamped. The flex is not that the machine never says "I do not know." It is that it is built so it cannot lie about it.
Check that yourself. That is the point.
The bet, restated
The one person billion dollar business was never a prediction about model capability. It was a prediction about what happens when the cost of standing up an Organisation falls to near zero.
Chains are consolidating. Agents are commoditising. The durable asset in between is the Organisation that owns the work: the identity, the team, the memory, the treasury, and the record that proves what it did.
Everyone is building agents. Agents alone are not enough.
Founding is one sentence away.

The headline capabilities get the attention. What sits under them, the policy, the logic, the semantic layer, decides whether a system holds up under load. Four posts this week, three ideas, one direction: agent infrastructure is built layer-by-layer, and the layers underneath are where the work is.
Governance Is the Steering Wheel
x402 gave agents the power to pay. xBPP gives them the discipline to answer for it. Every action is checked against policy before execution, not after the money moves. Ignition and steering, in the same vehicle.
Read the policy layer.
https://x.com/Vanarchain/status/2071559703721632136
One Bad Prompt Away
Most "autonomous agents" running today sit one loose prompt away from draining a wallet. No spend limits. No approval gates. No record of what they were even allowed to do. xBPP closes the gap with a declarative policy that the agent cannot override, checked on every action before money moves.
Give your agent rules.
https://x.com/Vanarchain/status/2071887064614809712
A Chain Built to Reason
Most blockchains store and execute. That is the entire job description. Vanar was built from the ground up to think. Reasoning is the base layer, not a bolt-on. The chain understands what it holds.
See the design.
https://x.com/Vanarchain/status/2072622842525487438
From File to Seed
Upload any file. Neutron learns the meaning. The file becomes a Seed. Any AI can query it. Memory that outlives the tool it was made in, with a Neutron API that is free to build against.
Start building.
https://x.com/Vanarchain/status/2072983451582509110
The week made the case straight: the frontier worth watching is no longer what agents can do at the top of the stack. It is the layer underneath, deciding whether any of it holds.

Every major technology shift hits the same wall. The technology scales. The governance doesn't. Bad things happen. Regulators step in.
We've seen this in financial markets, early social media, and crypto. We're about to see it in AI agents. And this time, we have a choice.
THE THING NOBODY IS PRICING IN
The conversation about AI agents has been dominated by capability. These are the wrong questions.
The right question is: what happens when millions of AI agents start spending money?
That's coming. x402 lets agents pay over HTTP. Google's AP2 extends agent commerce. Stripe's ACP opens card rails. Mastercard's Verifiable Intent adds identity. Execution is shipping. Accountability isn't.
Right now, a single agent with a misconfigured prompt could drain an account in seconds. None of these scenarios requires malice , just scale. And scale is guaranteed.
WE'VE BEEN HERE BEFORE
In the 1980s, program trading dominated markets. Computers executed faster than humans could react. Then, on October 19, 1987, Black Monday, automated cascades wiped 22% of the market in a day. The response was circuit breakers. Accountability retrofitted onto a scaled system.
In the 2010s, social media scaled faster than governance. Then, election interference, coordinated harassment, and algorithmic radicalization forced the conversation. Retrofitted after the harm.
2022: FTX showed what happens when a financial system runs at venture-capital speed with governance on a napkin. The response was what regulators should have required from the start.
In every case, governance came after the damage.
WHAT "GOVERNANCE" HAS TO MEAN
Real governance for AI agents isn't a content policy. It's an execution-time enforcement layer that evaluates every agent action before it happens.
That means: external (not baked in the agent), deterministic (same input, same output every time), auditable (verifiable records), portable (same policy across all rails), and open (a standard, not a product).
Anything less is duct tape.
WHY THIS HAS TO BE A STANDARD, NOT A PRODUCT
The business case for a product is obvious: premium features, lock-in, enterprise contracts.
But infrastructure that matters isn't built that way. TCP/IP isn't a product. HTTP isn't. OAuth isn't. They're standards everyone adopts because everyone needs them. Invisible because they work. Work because they're open.
If the agent economy is going to have real accountability, it can't depend on whichever vendor has the best sales team this quarter.
THE CHOICE WE HAVE RIGHT NOW
The agent economy is going live. Not negotiable.
What is negotiable: whether it launches with accountability baked in, or whether we wait for the inevitable disaster that forces retrofitting.
Read the xBPP spec: xbpp.org

For a decade, crypto's main contribution was simple: own the asset, own the keys. This week the same principle climbed the stack. Five posts, one through-line: the agent layer should be owned the same way money is.
You Own the Record
Your AI assistant is about to handle a lot of your money. Tipping. Splitting bills. Subscriptions. Maybe one day rent. The question of who owns the record of what it did has a clear answer: you. Veil's non-custodial mode derives your encryption key from a single wallet signature. The operator stores ciphertext they cannot read.
Generate the key yourself.
https://x.com/Vanarchain/status/2066505475307217302
x402 Moves. xBPP Decides.
An agent can send a payment. Whether it should is a separate question. x402 handles execution. xBPP handles governance. Built to work together, not compete.
See how they pair.
https://x.com/Vanarchain/status/2066882031473041799
One File, Every System
Most teams scatter agent rules across code. Different logic in every system. No clean way to audit any of it. xBPP collapses the policy into a single declarative JSON file. Readable, versionable, portable.
Write once. Enforce everywhere.
https://x.com/Vanarchain/status/2067269372821791138
Proof, Not Promises
Agents are moving from chat to action, signing transactions, moving funds, executing trades. The harder question is no longer whether they will act, but whether anyone can prove what they did. Veil keeps a signed, encrypted, tamper-evident record of every decision.
Read the proof layer.
https://x.com/Vanarchain/status/2067618506116415798
Memory You Can Carry
AI can write code, generate art, and analyze data in seconds. Ask it what you talked about yesterday and the screen goes blank. Memory should not be a feature you renegotiate every new tool. Neutron makes it portable.
Start building.
https://x.com/Vanarchain/status/2067939650379936008
The pattern this week was consistent. Whether the layer is payments, policy, proof, or memory, the same answer keeps showing up: the user holds the keys.