Hilal Agil

Permissionless Intelligence

Hilal Agil — Thu, 02 Jul 2026 00:00:00 GMT

Yesterday I wrote about the gating of intelligence — eighteen days in June when one of the world’s most capable models went dark on government instruction, and a frontier release that now requires customer-by-customer approval. I ended that essay by saying the answer is to build alternatives. This one is about the alternative I’ve been building, and about who it’s actually for.

Two things became true this year at the same time, and their combination changes the picture.

First, open models stopped being the compromise option. GLM 5.2, released with open weights in June, is discussed in the same sentence as Claude now — within a point of it on some of the hardest coding benchmarks. DeepSeek’s V4 release in April matched the strongest closed models on real software-engineering evaluations, open-weight under MIT. And on OpenRouter — the closest thing to a public scoreboard of what people actually run — models from Chinese open labs grew from a rounding error to roughly half of all traffic in a year. Whatever your benchmark of choice says this month, the honest summary is: for a large share of real work, open models are no longer clearly worse. They’re just less convenient to run.

Second, the convenience of the alternative — a handful of gated APIs — revealed its price. Access that can be revoked overnight, by an authority you can’t appeal to, is not infrastructure. June made that concrete.

So the capable models are free to run, and the case for running them yourself has stopped being ideological. What’s missing is the layer underneath: somewhere to run them that isn’t a hyperscale data center. That’s a network problem, not a model problem, and it’s the problem I’ve spent the last two years on.

A network anyone can join

Tenzro is an open protocol and network for decentralized AI and distributed computing. The protocol and network are open source under Apache 2.0 — the code is at github.com/tenzro — and the premise is simple to state: anyone with a computing system should be able to join the network, run a node, serve models from the registry, and earn for the compute they provide. A few clicks, not a procurement process.

That includes the machines the AI economy currently ignores: workstations, small clusters, regional data centers, edge hardware. Individually, none of them can host a frontier model. That was the design constraint — and working within it turned out to be the most interesting engineering problem I’ve taken on.

Intelligence that doesn’t need one building

The reason AI concentrated into a few giant facilities isn’t conspiracy; it’s physics and architecture. Large models wanted to live in one place, on tightly coupled hardware. The most useful thing I can report from building Tenzro is that this constraint is loosening — and modern model architectures are what loosen it.

Mixture-of-experts models — the architecture behind most serious open models today — only activate a small fraction of themselves for any given request. An MoE model is less like one giant brain and more like a panel of specialists, a few of whom are consulted at a time. Specialists don’t need to sit in the same rack. Tenzro runs MoE models across multiple nodes — locally, within a city or a facility, and globally when the request tolerates it — routing each request to the experts it actually needs. The network also supports multi-token prediction models, which generate several tokens per step; the fewer round trips a model needs, the friendlier it is to a distributed setting, so the architectures and the network are converging from both directions.

Training tells the same story. Approaches in the DiLoCo family let nodes train independently and synchronize rarely, instead of demanding the constant lockstep communication that only a single building can provide. Tenzro supports Decoupled DiLoCo — resilient, distributed training where nodes can join, drop, and rejoin without bringing the run down. Resilience is the whole point: a network where participation fluctuates and the work persists is a network with no switch to pull.

None of this required inventing exotic science. It required taking a decade of ML research seriously as infrastructure — asking, for each technique, what it means for where computation can physically live. The conclusion I keep reaching is that the centralized data center is an artifact of one era of model architecture, not a law of nature.

A practical option for small nations

That conclusion matters most for the players the current model was never going to include.

Most countries cannot build a gigawatt AI campus. The costs, the power, the chips, the expertise — the whole stack assumes a scale that a few dozen nations will ever reach. UNCTAD said as much in its 2025 technology report: the United States and China account for roughly two-thirds of the world’s AI capacity and patents, and the “AI divide” between them and everyone else is widening, not closing. For most of the world, the current architecture of AI offers exactly one role — customer — and one posture toward it: dependence.

I understand that problem from the inside. I grew up in the Maldives, a country of half a million people spread across a thousand islands, that will never host a hyperscale data center and shouldn’t have to in order to have a say in the intelligence its economy runs on. When I look at the AI map from there, the question isn’t how to win a compute race no small nation can win. It’s whether there’s a way to participate at all without renting your entire intelligence layer from someone else’s territory, on terms you don’t set.

A distributed network is that way. A country doesn’t need a national megaproject to put its universities, its businesses, and its existing data centers to work on the network — providing compute, running models and agents, keeping its own data under its own governance, and earning rather than only paying. Sovereignty stops meaning “own a campus you can’t afford” and starts meaning “participate in a network on your own terms.” The same door is open to the independent operator: the individual, the small business, the regional data center with spare capacity. Bottom-up participation and national sovereignty turn out to be the same mechanism at different scales.

This is also where the policy conversation is finally heading. Last August the UN General Assembly established a Global Dialogue on AI Governance and an independent scientific panel; the first Dialogue convenes in Geneva this month. At Davos in January, one of the sessions was titled “Digital Embassies for Sovereign AI” — how nations might extend their digital sovereignty beyond their own borders. These forums are circling the right question: how the large and the small share the same technology without the small becoming permanent tenants of the large. I think the answer is less a treaty than an architecture. You don’t legislate your way to participation; you build a network that makes participation the default, and then the governance has something real to govern.

The footprint, and who gets to participate

Two consequences of distribution matter to me beyond resilience.

The first is environmental. The AI buildout is currently measured in gigawatts — new campuses, new plants, new transmission lines, much of it duplicating capacity that already exists in the world as underused hardware. A network that puts existing machines to work, closer to where demand actually is, grows AI’s capability without growing its footprint at the same rate. Distributed computing won’t solve AI’s energy problem by itself, but concentration is quietly making it worse, and distribution is the lever pointing the other way.

The second is economic. The AI economy is becoming one of the largest in the world, and right now the only way to participate on the supply side is to own a gigawatt campus or shares in one. A network where a workstation in Malé, a render farm in Lagos, or a small data center in Jakarta can serve models and earn from day one distributes that economy the way the web once distributed publishing. Participation is the antidote to dependence.

The quiet bet

Tenzro is a bet that the future of AI looks less like a few cathedrals and more like a grid — capable open models, running on a resilient network of hardware nobody fully controls and anybody can join, governed by rules that are legible because the code is open.

June was a preview of the cathedral model under stress. The grid is the alternative, and it’s no longer hypothetical: the protocol is open, the network runs, and the door is a few clicks wide. Whether you’re a nation that will never build a campus or a person with one good machine, if you have compute, you can be part of this.

The Gating of Intelligence

Hilal Agil — Wed, 01 Jul 2026 00:00:00 GMT

In June 2026, two things happened within two weeks of each other that should change how we think about depending on centralized AI.

On June 12, three days after Anthropic publicly launched Claude Fable 5 and Mythos 5, the US government issued an export-control directive citing national security. Anthropic had to abruptly disable both models for all customers — reportedly after the government became aware of a method for jailbreaking Fable 5. The restriction extended to any foreign national, inside or outside the United States, including Anthropic’s own foreign-national employees. The controls were lifted on June 30, and the models came back. But for eighteen days, one of the most capable AI systems in the world simply went dark, on government instruction, with no notice.

Two weeks later, on June 26, OpenAI rolled out GPT-5.6 — and limited it to roughly twenty companies whose participation had been individually approved by the US government. OpenAI itself called this a “short-term step” and said plainly: “We don’t believe this kind of government access process should become the long-term default.”

I want to be careful here. Neither of these is, on its face, a story about villains. Governments have legitimate national-security interests. Companies complying with lawful directives are doing what they must. And I have no interest in relitigating any single decision.

But step back from the specifics, and a pattern comes into focus that I think matters more than any of the individual events.

Access is becoming conditional

For most of the modern software era, we’ve treated access to our tools as roughly permanent. You bought the license, you called the API, and barring your own non-payment, the thing kept working. Capability was something you had.

Frontier AI is quietly breaking that assumption. In the span of a month we saw the most capable models become:

Revocable overnight — disabled for everyone, with hours of notice, by an external authority.
Gated by identity — restricted based on nationality, including for a company’s own employees.
Approved customer-by-customer — available only to a vetted list, each participant individually cleared.

None of this is inherently malicious. All of it is unpredictable. And unpredictability is precisely the property you cannot design a serious system around.

If you are building a product, a research program, or an institution on top of a model that can be switched off by a party you don’t control and can’t appeal to, you don’t actually have infrastructure. You have a privilege that happens to be extended to you today.

This is the infrastructure gap, made concrete

I’ve written before about the infrastructure gap in autonomous AI — the argument that centralized systems, built for a world of stable jurisdictions and human actors, weren’t designed for the world AI is creating. The events of June 2026 are that abstraction becoming very concrete.

When capability itself can be gated at the point of a single provider or a single regulator, three needs stop being ideological preferences and become practical requirements:

Verifiability. If you can’t inspect how a system behaves, you’re left trusting that it will keep working and keep behaving. Auditable, attestable systems replace faith with evidence.
Distribution. A single point of control is a single point of failure — and, increasingly, a single point of revocation. Distributing where models run and who can coordinate them removes the overnight kill switch.
Governance you can actually participate in. “Governance” too often means a decision made somewhere else that you learn about after it affects you. Real governance means the rules are legible, the process is repeatable, and the affected parties have standing.

This is not an argument against safety controls. It is an argument that safety and controllability should be properties of the systems themselves — provable, distributed, and open to scrutiny — rather than something enforced only through a chokepoint that a handful of parties happen to hold.

Why I keep building toward this

Everything I’ve worked on — Tenzro’s open protocol for decentralized AI and distributed computing, Furcate’s verifiable edge intelligence, the data infrastructure I’m now consolidating under Ipnops — points at the same conviction: that intelligence is becoming the substrate of civilization, and a substrate should not be revocable at will by whoever sits closest to the switch.

The Fable 5 shutdown and the GPT-5.6 gating aren’t anomalies to wait out. They’re early, visible symptoms of a structural fact: we’ve concentrated the most consequential technology of the century into a very small number of hands, and we’re now discovering, in real time, what that concentration feels like when it tightens.

The answer isn’t to reject centralized AI, which will remain the best tool for many jobs. The answer is to make sure it isn’t the only option — to build the verifiable, distributed, governable alternatives now, while the pattern is still just emerging, rather than after we’ve learned we can’t live without a switch we don’t control.

Not everything needs to be decentralized. But the ability to think, build, and reason with machine intelligence probably shouldn’t depend on permission that can be withdrawn overnight.

Who Does the Agent Answer To?

Hilal Agil — Tue, 16 Jun 2026 00:00:00 GMT

Sometime this spring, without much ceremony, the world finished wiring money to machines.

In March, Stripe took its Tempo blockchain to mainnet with a protocol built specifically for machine-initiated payments, and Mastercard spent $1.8 billion acquiring stablecoin infrastructure to serve the same future. Visa’s agentic commerce program now counts its partners in the hundreds and predicts ordinary consumers will be delegating purchases to agents by the holiday season. Whatever your view of the timeline, the direction is settled: AI agents are becoming economic actors — holding balances, executing transactions, entering agreements.

The rails came first. Everything that makes an economic actor accountable is being retrofitted afterward. I think that ordering is backwards, and this year has already produced the evidence.

The accountability gap, in three documents

Three things happened in the first half of this year that belong in the same frame, though they came from a courtroom, a statehouse, and a payments company.

On March 4, one of the world’s largest insurers filed suit against OpenAI in federal court — an early entry in what will become a defining genre: litigation over what happens when an AI system, acting on someone’s behalf, causes loss. On January 1, a California law took effect that bars what might be called the “the AI did it” defense — the idea that harm caused by an autonomous system attaches to no one. And in April, a regulated financial institution published the first “Know Your Agent” framework — an attempt to do for AI agents what Know Your Customer did for banking: establish who you’re actually dealing with.

A lawsuit, a statute, a compliance framework. Three different institutions independently discovering the same missing layer: agents can act, but nothing reliably establishes who an agent is, whose authority it carries, and where responsibility lands when it goes wrong.

Liability needs someone to attach to

Every functioning economy runs on a quiet assumption: an actor is a durable thing. A person persists. A company persists. Courts, contracts, credit, insurance — all of it depends on identity that holds still long enough for responsibility to attach.

Agents, as currently built, fail this test in a new way. An agent is a temporary arrangement — a model version, a context window, a set of instructions — that can be spun up in thousands of copies, modified mid-task, and gone by morning. When one of those arrangements moves money or signs up for obligations, the old questions arrive with nowhere to land. Which copy acted? Under whose instruction? Was it the deployer’s configuration, the model builder’s training, or the user’s prompt that made the difference? The California statute can insist that someone is liable; it cannot conjure the records that show who.

The honest answer is that today, mostly, nobody knows — because the infrastructure that would know was never built. Identity for agents isn’t a login. It’s a verifiable, persistent record: this agent, operating under this authority, within these limits, with this history. Not a policy document describing good intentions — a property of the system, checkable at the moment of the transaction, the way a signature is checked, enforceable the way a spending limit is enforced.

Capability first, accountability later — again

I’ve written about the infrastructure gap in autonomous AI — the pattern where we extend AI’s ability to act faster than we extend the systems that make action governable. Agent commerce is that pattern at its purest. The capability (an agent that spends) took roughly two years to become products. The accountability (an agent that can be identified, bounded, and answered for) is still mostly aspiration, arriving through lawsuits — which is to say, through the failure mode.

This is the problem I’ve been circling from different directions for years: enforceable identity and governance standards for AI, memory and authority that persist across systems rather than evaporating between sessions, rules that are properties of infrastructure rather than promises in documentation. It’s work I began under Praecise and Rivier and have now brought together under Ipnops, because I’ve come to believe these aren’t separate products. Reaching the world’s data safely and acting in the world accountably are the same problem: both need rules that travel with the actor.

None of this is an argument against agents. I want the agent economy to work — it’s one of the genuinely large economic openings of the decade, for people and small businesses most of all, and it will not survive its first serious wave of fraud and unattributable losses if accountability stays bolted on. Trust is not a feature you add later. It’s the load-bearing wall.

An economy of actors nobody can identify, answering to nobody in particular, is not a market. It’s a crowd. The difference between the two has always been accountability — and we still get to choose which one we’re building.

There’s a second question stacked behind this one: agents ultimately answer to whoever controls the intelligence they run on. Who that is, and what happens when they change the terms, is a story that’s beginning to tell itself — and it deserves its own essay.

Privacy Is Not a Setting

Hilal Agil — Tue, 02 Jun 2026 00:00:00 GMT

In January, a federal judge in New York ordered OpenAI to hand over twenty million ChatGPT conversations to lawyers in a copyright case. The logs are de-identified, and the order came with protections. But sit with what actually happened: millions of people told a machine things they may never have told another person, and those conversations became discoverable evidence in a lawsuit none of them are party to, under rules none of them agreed to, decided in a jurisdiction most of them have never set foot in.

No setting could have prevented that. There is no toggle for “my words stay out of other people’s litigation.” And that’s the point I want to make, because it’s the quiet assumption underneath almost everything I build: privacy is not a setting. It is a property of architecture. By the time you reach the settings page, nearly every decision that matters about your data has already been made — where it lives, what law it lives under, who can compel it, and what it can be used for by default.

The setting was never yours

Consider how the last few months have gone for the ordinary user.

In November, LinkedIn began using member data to train its AI models across Europe and beyond — turned on by default, opt-out buried in preferences. The pattern is familiar by now: the choice arrives after the decision, framed as control, functioning as consent theater. You weren’t asked whether the system should work this way. You were asked whether you’d like to click a box after it already did.

And even when a company wants to make a real promise, jurisdiction overrules it. Last summer, a Microsoft executive was asked under oath, in the French Senate, whether he could guarantee that French citizens’ data would never be handed to American authorities without France’s consent. He said he could not. That testimony did more for public understanding of data sovereignty than a decade of policy papers, because it made the structure legible: the data was in France, the servers were in France, the contract said all the right things — and none of it mattered, because the company that operates the machines answers to a different sovereign.

A setting is a promise made inside an architecture. If the architecture reports to someone else, the promise is decoration.

Governments stopped believing promises first

Watch what the customers with the most leverage on earth are doing about this.

The European Union’s Data Act became applicable in September, giving users the right to move between clouds. Amazon launched a legally separate European sovereign cloud in January. In April, the European Commission put roughly 180 million euros into procuring sovereign cloud capacity, and France announced it would begin migrating government workstations off American operating systems entirely — following Denmark, which spent the past year moving its ministries off Microsoft. These are not gestures. They are states concluding, after years of adequacy decisions and privacy frameworks and contractual clauses, that the only guarantee that survives contact with another country’s courts is physical and architectural: our data, on our infrastructure, under our law.

I think they’re right, and I think the conclusion runs deeper than governments have yet followed it. Because the same logic applies at every scale below the nation — to a hospital, a record label, a research lab, a family. If sovereignty over data can’t be granted by promise, only built by design, then it has to be designed in everywhere, not just at the level of national clouds.

Design it in, or you don’t have it

This is the thread that runs through everything I work on, mostly without being announced.

Tenzro moves computation to where data already lives, instead of moving everyone’s data to where the computation is. A distributed network of nodes means a clinic, a studio, or a ministry can run serious AI on infrastructure it controls, under the law it actually lives under — the model travels, the data doesn’t have to. Ipnops approaches the same principle from the other side: as AI reaches beyond the web into the world’s data, every access happens under explicit, verifiable rules — who may reach what, for which purpose, on whose terms. And Vestigim carries it into the creative industries, where the people who make the culture AI learns from deserve data they can verify and terms they can see, a chain of custody rather than a scraping and an apology.

None of these are privacy products. That’s deliberate. Privacy as a product is a setting with better marketing. What I’m after is the structural version: systems where the default is that data stays under the governance of the people it belongs to, and anything else requires permission — instead of systems where extraction is the default and privacy is the paperwork.

Sovereignty runs all the way down

Sovereignty in the age of AI now means compute, data, and model access — and open infrastructure is the only version of it that doesn’t require being rich. That’s an argument I usually make at the scale of nations, but it’s the same argument at every scale below one. A country renting its intelligence layer from someone else’s territory, an artist whose life’s work trains a model she can’t audit, a person whose private conversations surface in a stranger’s lawsuit — these are the same structure at different sizes. Somebody else’s architecture, somebody else’s jurisdiction, somebody else’s default.

The fix is the same at every size too, and it is not a better settings page. It is architecture that makes the promise physically true: computation that comes to the data, access that is governed and verifiable, infrastructure that answers to the people who depend on it. Privacy, done properly, isn’t something you configure. It’s something you can stop thinking about — because it was decided in your favor before you arrived.

Data Has a Chain of Custody Now

Hilal Agil — Wed, 15 Apr 2026 00:00:00 GMT

In late January, three music publishers — Universal Music Publishing Group, Concord, and ABKCO — filed a $3.1 billion lawsuit against Anthropic. I read it with more than professional interest. I spent the first chapter of my working life as a musician, and the catalogs named in that suit are the kind of work people spend decades of their lives making.

But the detail worth paying attention to isn’t the number. It’s the legal theory. The suit doesn’t primarily argue that training an AI model on songs is illegitimate. It argues that the data was taken through piracy — that the collection itself was the violation. And that framing is showing up everywhere now. Across the AI lawsuits moving through courts this year, the question is drifting from whether training is fair use to a much older and sharper question: where did you get this, and can you prove it?

In other words, data has acquired a chain of custody. I think that’s one of the most consequential shifts happening in AI right now, and it’s being underestimated because it looks like legal housekeeping.

From free harvest to accounted supply

For the first decade of modern machine learning, data was treated like weather — something that was simply there. You scraped the web, you trained, and the provenance of any individual piece of text or audio was nobody’s concern, least of all the model’s.

That era is visibly closing. Reddit now licenses live access to its data on usage-based terms measured in the tens of millions of dollars per year, and treats that access as a product with pricing tiers, not a favor. Publisher content marketplaces have moved from experiment to ordinary business. The going assumption among researchers — that the supply of quality public text is on a path to exhaustion sometime between now and the early 2030s — has stopped being a provocative forecast and become planning input.

Put those together and you get a new default: data is supplied, not harvested. Supplied things have owners, terms, prices, and records. Harvested things don’t. The entire apparatus of the last decade was built for harvesting, and it is being retrofitted, lawsuit by lawsuit and license by license, into a supply chain.

The web was never the world

Here’s why I think this matters beyond the courtroom. The web — the thing all of this litigation is fighting over — was always a thin and biased sample of reality. It’s what people happened to type, photograph, and upload. It contains very little of what the world actually knows: what happens on factory floors and coral reefs, inside instruments and hospitals and power grids, in the lived experience of people who never post.

AI is now capable enough to make use of all of that. The models are not data-starved because the world lacks information; they’re data-starved because almost none of the world’s information was ever designed to be reachable, and its owners — rightly — won’t hand it over into a system with no chain of custody, no rules of access, and no way to get paid.

So the two problems are actually one problem. The reason the web’s data is ending up in court and the reason the world’s data stays locked away are the same: we built systems for taking data, not for governing it.

Provenance is the product

This is the conviction underneath most of what I’m building now. If the next decade of AI depends on data the web never had, then the infrastructure that matters isn’t another scraper — it’s the layer that lets data move with its history and its rules attached: who it belongs to, what it may be used for, on what terms, with a record that survives the transaction.

That’s the thesis behind Ipnops: making the world’s data — digital and physical — something AI can reach, understand, and act on safely, with clear rules for how it’s accessed and governed. And it’s the thesis behind the verifiable-data side of Vestigim: musicians shouldn’t have to discover in a court filing, years later, how their work entered a training run. The record should exist from the start, and the terms should be enforceable from the start.

None of this requires believing the lawsuits will all succeed, or that scraping will disappear. It requires only noticing which direction every incentive now points. Rights holders want records. Regulators want records. Model builders, frankly, want records too — provenance is the only durable defense they’ll ever have.

For most of a decade, the honest answer to “where did your model’s knowledge come from?” was: everywhere, and don’t ask. That answer is dying in courtrooms right now. What replaces it — data that carries its chain of custody with it — is a better foundation for everyone, including the machines.

The Infrastructure Gap in Autonomous AI

Hilal Agil — Wed, 01 Apr 2026 00:00:00 GMT

The tech industry has largely operated under a simple assumption: the future of AI will be built on centralized infrastructure. In many ways, that assumption still holds.

Cloud providers like AWS and Google Cloud, along with established financial networks, process the majority of global data and transactions. They are efficient, scalable, and backed by clear legal frameworks. For training large models or running conventional applications, centralized systems remain the most practical and reliable option.

But as AI evolves, the conditions those systems operate in are also changing.

A Changing Operating Environment

Over the past few years, infrastructure has become increasingly entangled with geopolitics.

Data centers, cloud regions, and network infrastructure are no longer just technical assets — they are tied to jurisdiction, regulation, and, in some cases, strategic risk. Access can be restricted, services can be disrupted, and dependencies can become visible very quickly under pressure.

This doesn’t mean centralized infrastructure is failing.

But it does mean that relying on a small number of tightly controlled systems introduces exposure — particularly when software, data, and users operate globally.

At the same time, AI itself is evolving.

It is no longer limited to passive tasks like generating text or analyzing data. Increasingly, it is expected to act — coordinating workflows, making decisions, interacting with external systems, and in some cases, transacting.

This combination — more autonomous systems operating in a more fragmented world — begins to surface new constraints.

Where Friction Starts to Appear

Two areas in particular become more apparent.

Payments and Coordination

Traditional financial systems are built around human identity and institutional onboarding. They rely on account ownership, jurisdictional boundaries, and structured compliance processes.

These assumptions work well for people and organizations, but they are less suited to software operating autonomously — especially when interactions are frequent, cross-border, and low-value.

For an AI system, opening accounts, navigating jurisdictions, or handling payment friction becomes a bottleneck.

Data, Compliance, and Location

At the same time, data is becoming more tightly controlled.

Regulations, internal policies, and geopolitical considerations increasingly dictate where data can reside and how it can be processed. In some cases, data cannot leave a specific region or be exposed to external infrastructure at all.

This creates a tension: AI systems require access to data to be useful, but moving that data across borders — or into centralized environments — may not always be possible or desirable.

Not Replacement — But Extension

These challenges don’t suggest that centralized infrastructure is obsolete.

They highlight that it was designed for a different model — one where:

humans are the primary actors
systems operate within stable jurisdictions
trust is anchored in institutions

As those assumptions shift, there is a growing need for infrastructure that can support:

machine-to-machine interaction
execution in constrained or sensitive environments
coordination across independent systems

This is where a complementary layer begins to emerge.

Where Tenzro Fits

Tenzro is designed to operate within this gap.

It combines distributed compute, hardware-based security, and a programmable settlement layer to support systems that need to operate across boundaries — technical, regulatory, and geographic.

This includes:

enabling software systems to exchange value directly, without relying on traditional financial onboarding
allowing computation to occur in secure environments where data remains private, even during processing
integrating with enterprise frameworks such as Canton to support workflows that require compliance and auditability
supporting smaller, efficient models that can run across a broader range of hardware, reducing reliance on centralized data centers

The goal is not to replace centralized infrastructure, but to extend it — particularly in scenarios where access, trust, or location introduce constraints.

A Hybrid Direction

In practice, the direction forward is unlikely to be purely centralized or purely decentralized.

Centralized systems will continue to handle:

large-scale data storage
high-performance model training
applications that benefit from tight control and legal accountability

At the same time, more distributed approaches may increasingly handle:

coordination between independent actors
execution where data cannot leave specific environments
interactions across fragmented or constrained regions

The two are not competing systems. They address different realities.

As AI systems move from tools to participants, the infrastructure around them will need to adapt — not just to new capabilities, but to a changing world.

Geopolitical dynamics, regulatory fragmentation, and shifting control over data and compute are no longer edge cases. They are becoming part of the baseline.

The response is not to replace existing systems, but to build around their limitations.

Not everything needs to be decentralized. But some things increasingly need to be.

Vestigim

Hilal Agil — Fri, 27 Mar 2026 00:00:00 GMT

Over the past five months, two of the three major labels settled with the AI music companies they sued in 2024 and announced licensed platforms with them. Sony is still in court. In January, music publishers filed a copyright case against an AI lab asking for more than three billion dollars. Spotify, meanwhile, spent the fall building AI disclosure into song credits and removing tens of millions of spam tracks.

That’s the industry conversation about AI and music: generation, infringement, licensing, flood. Prompt in, song out, lawsuit after. And I understand why it dominates — livelihoods are at stake. But as someone who spent years of my life in studios and on stages before I worked in technology, I think the whole fight is over the wrong asset.

A music catalog has priced exactly two things for a century. Masters — the recordings. Publishing — the compositions underneath them. Both are records of what an artist already made: finished, and fully valued. What no one has ever owned is the layer above the output — the judgment that chose it — and the layer beneath it — the craft that produced its sound. Vestigim is my work on those two layers.

Two more layers of the catalog

Vestigim maps an artist’s judgment, sound, career, and world into distinct, ownable, licensable layers that didn’t exist as assets before. The judgment, career, and world resolve into one living picture — the Vestigim Graph. The sound — how an artist actually gets their tone — becomes Machine Sounds. Together they’re a third and fourth layer of the catalog: not what the artist made, but how they decided and how they sounded, as things a machine can read.

This is a forward bet. Catalogs still price only masters and publishing today. But in 2026, machines are becoming the main consumers of an artist’s sound and an artist’s judgment — and that’s what turns a disposable preset, or a pattern of choices, into an asset worth owning. When machines are the audience, how you decide and how you sound stop being byproducts of the work. They become the work.

A memory of judgment

The Vestigim Graph is the closest thing to a living memory of an artist: their press, their charts, their scene, their collaborators, their career and life — all of it organized around one thing, the way that artist makes decisions. And that’s the part that matters, because it isn’t a model of what an artist tends to produce. It’s a model of what they chose — and, just as importantly, what they turned down, and why.

That second half is where the real signal lives. What an artist throws away says more than what they keep — the take they cut, the mix they held, the feature they walked away from. Vestigim reads the work itself, reads the choices behind it, and only asks the artist directly when the evidence runs out. Everything it believes comes with a source. It doesn’t assert; it shows its work. That’s the difference between a memory a system can trust and a guess it can’t.

Not who they were — how they’d decide next

Because that memory holds the reasons and not just the results, it can be run forward. Point it at where the scene and the moment are heading, and it can surface the moves an artist would lean into, reject, or want to explore next — not a prediction of the future, but their own taste, applied to a field that hasn’t happened yet.

A queryable model of an artist’s judgment is a genuinely new object, and the uses stack up fast. A real-time read on how an artist’s instincts move against the market. A licensed check other AI systems can call to ask whether an output actually sits inside an artist’s taste, or just borrows their name. A signal an investor can weigh when valuing a catalog. A way to reconstruct a historic artist’s decision-making from the evidence they left, for scholarship or a sanctioned live experience. Each of these is something the industry has wanted and never had a way to own.

The craft, captured — not the preset

The other layer, Machine Sounds, is an artist’s authenticated sound-craft — their tones, their moves, the way they engineer a sound, made ownable and reproducible. Not a frozen preset, but the living behavior behind it.

The reason this is hard is worth stating plainly. A real piece of studio hardware doesn’t have a “setting” you can copy — it’s a physical thing where every knob changes how every other knob behaves, across everything you feed into it. Most tools capture a snapshot of one setting and call it done. Machine Sounds captures the whole instrument: its actual behavior across its entire range, measured from the real unit, so what you license sounds like the hardware because it is the hardware, modeled faithfully — with a clear line of provenance back to the physical device. It captures a sound; it never invents one. That’s exactly what makes it something you can license rather than fake.

One graph, and a firewall

The two layers are joined, but only in one direction, and that direction is the whole ethic of the thing. The Graph learns from Machine Sounds — an artist’s sonic choices are evidence of their judgment — but Machine Sounds never learns from the Graph. Sound informs the picture of the artist; it never gets mistaken for the artist. Only genuine decisions carry an artist’s taste. Everything else — the sound, the market, the press, the whole surrounding world — is context that judgment reacts against, never a stand-in for it. That rule is built into the structure, not bolted on as a promise. It’s what lets the system take in an entire industry’s worth of noise around an artist without ever confusing the noise for the person.

On the rights-holder side of the line

There’s a reason the architecture is this careful, and it’s the same reason the lawsuits exist: in the age of AI, nobody can prove what happened to a piece of music — whether a work trained a model, whether a track was made by a person or a system, whether a voice was licensed or lifted. Unauthorized use became common precisely because there was never an audit trail.

Vestigim is built on the other side of that line. It does not sell AI-generated music. Every asset is human-authored, with provenance back to a real work, a real decision, or a real instrument — the licensed inputs that any downstream system has to license to operate honestly. The models are living because they keep updating, not because they resurrect anyone. We map a practice — judgment, craft, and how they evolve — not a person. We don’t clone the artist. We license their judgment.

I don’t think AI in music is going away, and I don’t think fighting it as a category works — the settlements of the past few months suggest the industry has quietly reached the same conclusion. Even the sharpest artist statements against AI exploitation have been careful to say the technology, used responsibly, has enormous potential. The real dividing line was never AI versus no AI. It’s whether the tools answer to the people who make the music. Masters and publishing gave artists ownership of what they made. The next two layers give them ownership of how they made it — and that’s the standard I’d want as a musician, so it’s the one I’m building as a technologist.

Furcate: Decentralized Edge Intelligence for the Era of Physical AI and World Models

Hilal Agil — Tue, 24 Feb 2026 00:00:00 GMT

We’re hitting the limits of what you can do by just scaling up giant language models on internet text. The easy data is mostly gone, and the next real leaps are coming from better architectures, multimodality, synthetic data, and smarter reasoning at inference time — not just piling on more parameters.

But the biggest jumps ahead could be coming from AI that actually lives in the physical world: world models that understand cause and effect plus real dynamics, embodied agents moving around in messy environments, robots that can handle the unexpected, and autonomous systems learning straight from live edge sensors.

All of that runs into hard realities — latency kills you, bandwidth is expensive or nonexistent, safety and privacy can’t be compromised. You can’t keep funneling everything to some central cloud. You need something that grabs trustworthy real-world data right where it happens, does the heavy lifting locally when it makes sense, and lets intelligence emerge across a bunch of distributed nodes without any single choke point.

I’ve been playing on these ideas since around 2024 — starting with some early thoughts and papers about nature-inspired, ecosystem-style AI networks where everything adapts in a distributed way. What began as sketches on a whiteboard has turned into actual working infrastructure. That’s Furcate.

At its core, Furcate lets distributed AI systems work together smartly: heavy processing happens securely at the edge, data gets tokenized for coordination, and everything stays verifiable.

It isn’t another IoT platform that’s mostly about piping raw data around. This is built for what’s coming after the big LLM wave:

Edge-first thinking — Run inference and early reasoning right on the device or very close by. Cuts latency, saves bandwidth, keeps private stuff private, and works even when the network flakes out.
Mesh-style coordination — Nodes link up dynamically into tough, self-healing networks. They share context and updates peer-to-peer. No central boss, no single point that can fail or get shut down.
Real trust built in — Every piece of data or inference comes with cryptographic proof: quantum-safe encryption, verifiable attestations, tokenized controls. So when that info feeds into world models, federated training, or data markets, you actually know where it came from and that it hasn’t been messed with.
Live, adaptive teamwork — Nodes push federated updates, reach consensus on what happened, run ensemble-style reasoning together. The whole system keeps getting better without anyone pulling strings from above.

Turning Everyday Edge Devices into Verifiable Intelligence Nodes

Furcate turns everyday edge devices — Raspberry Pis, NVIDIA Jetsons, industrial boxes — into active participants in a larger, real-world-grounded distributed intelligence network. The hardware doesn’t matter much; it’s swappable. The framework is what counts.

To make the trust part even stronger, Furcate plays really well with something like Minima’s lightweight Layer-1 blockchain and their Integritas middleware. Minima lets you run full nodes directly on tiny devices, hashing and timestamping data right at the source for immutable provenance. Pair that with Furcate’s local smarts and tokenization, and you get end-to-end verifiable intelligence — no middlemen, no central weak links. It’s perfect for feeding clean, attested signals into world models, autonomous agents, or physical AI.

Why this matters right now

Everyone — big tech, robotics companies, factories, environmental platforms — wants high-quality, local, real-time physical data to train better world models and run real AI in the world. Centralized setups choke on privacy, scale, energy use, and basic trust issues. Furcate (especially with Minima/Integritas underneath) fixes that by:

Giving verifiable origin stories for edge data, which is the current bottleneck for good world-model training
Letting things run autonomously in tough spots — think mangroves, coral reefs, offshore rigs, remote factories
Opening up decentralized incentives so nodes can earn from contributing to shared learning
Actually being kinder to the planet (local processing means way less constant cloud traffic and wasted energy)

Where it’s already making a difference

The system drawn from years of building distributed systems and is running in real spots:

Environmental & planetary — Live ecosystem tracking (mangrove conditions, coastal changes, coral bleaching forecasts) with edge-verified data feeding predictive models
Industrial setups — Factory optimization, predicting machine wear, coordinating supply chains — all with on-device reasoning and secure sharing
Autonomous systems — Fleets of vehicles, boats, drones doing sensor fusion, navigation, and group decision-making in decentralized setups
Basically anywhere physical AI needs edge brains + real trust: smart grids, robot swarms, city-wide sensing

Looking ahead

The future of AI isn’t locked in huge data centers or old-school IoT. It’s spreading out — edge-native, deeply tied to the physical world, coordinated across independent nodes, verifiable every step, and constantly learning from reality.

Furcate is designed to be a base layer for that change. If you’re building world models, physical AI, edge systems, decentralized data markets, or grounded autonomous agents, I’d love to talk. It’s designed for integration and real-world use — especially alongside edge-native verification like Minima.

The next wave of intelligence isn’t waiting for bigger servers. It’s already starting at the edge — decentralized, trustworthy, and very much alive.

Building the Infrastructure Layer for Institutional Finance and Decentralized AI

Hilal Agil — Fri, 19 Sep 2025 00:00:00 GMT

We launched Tenzro in early 2024, right when the convergence of AI and web3 was the hottest narrative in the space. Everyone was talking about AI-powered DAOs, blockchain-based AI training, verifiable datasets, and smart contracts that could learn and adapt.

While the story sounded compelling, the reality was different. Projects were either really good at blockchain development or really good at AI, but almost nobody was doing both well. The few teams that tried to combine them kept running into the same problems: the infrastructure wasn’t there.

Instead of competing in the crowded application layer, we decided to solve the infrastructure problem that was slowing everyone down. Our mission became simple: take away the complexity so teams don’t need to become experts in blockchain validation AND AI infrastructure AND cryptography just to ship their product.

The Problem Everyone Avoids

Infrastructure as an afterthought. Most teams treat infrastructure like plumbing — something you deal with later. But when you’re handling billions in tokenized assets or deploying AI that makes autonomous decisions, “good enough” infrastructure becomes a massive liability.

The expertise gap. Building reliable blockchain infrastructure requires deep knowledge of consensus mechanisms, cryptographic security, and network protocols. Building AI infrastructure needs expertise in distributed computing, model verification, and hardware optimization. Very few teams have both skill sets, and even fewer want to maintain both.

Smart teams building on shaky ground. An AI company with breakthrough algorithms but no way to prove their models haven’t been tampered with. Financial institutions wanting to tokenize assets but needing infrastructure that can handle enterprise compliance.

We realized we could solve this by becoming the infrastructure layer that handles the complex stuff, so application teams can focus on what they do best.

What We Actually Built

Infrastructure for Institutional Blockchains

We became validators for institutional blockchain networks:

Canton Network — Goldman Sachs and other Wall Street firms just put $135M behind this privacy-focused blockchain. We work closely with the Canton foundation and run validator nodes for the network and trained a custom coding model to provide an AI application layer for building tools for DAML, the programming language that powers Canton’s smart contracts.

Ripple — Cross-border payments need serious security. We provide validation infrastructure that meets banking standards for moving large amounts of money across borders.

Stellar — As CBDCs move from pilot to production, we’re providing the validation infrastructure for one of the key players that is building the infrastructure for central banks.

Our Own Blockchain Protocol

We developed a Proof of Authority consensus mechanism that powers projects building on our platform like Boli, PRVNZ, and Pillars Foundation. This gives us a testing ground for infrastructure improvements that we can later apply to the larger networks we validate.

AI Infrastructure Services

Tenzro Grid — High-performance GPU and TPU access without the operational headaches, plus tools for training, inference, and end-to-end workflow verification.

Tenzro AI Governance — Frameworks for AI transparency and auditability. As AI agents become more autonomous, institutions need to understand and govern what these systems are doing.

Tenzro Identity — Infrastructure for distinguishing between humans and AI agents in applications where that matters for security and compliance.

High Performance Financial Settlement Layer

High-performance transaction and settlement middleware for stablecoin providers, Bitcoin, EVM blockchains, and our partner ecosystems. We currently support Circle USDC, Ethereum, and Bitcoin, with infrastructure built for 24/7 operation with security and the compliance features that institutional risk teams require.

The Infrastructure Services That Keep Things Running

Tenzro Network — Decentralized networking and validation across multiple blockchain protocols. Our job is making sure transactions process reliably, even during high demand.

Tenzro Ledger — Hardware security modules and cryptographic key management. Critical infrastructure for keeping private keys actually private.

These services solve the expertise problem. Instead of every team needing blockchain and AI infrastructure specialists, they can focus on their application while we handle the complex backend systems.

Why Infrastructure Matters Now

We’re at a moment where both institutional finance and AI are moving from experimental to production. Goldman Sachs is tokenizing real assets. Central banks are launching CBDCs. AI systems are making autonomous decisions that affect real business outcomes.

But production systems need production-grade infrastructure. The kind that works 24/7, meets regulatory requirements, and can scale when needed. Most application teams don’t want to build that infrastructure themselves — they want to focus on their core product.

That’s the gap we’re filling. We run validator nodes, the hardware security, the compliance frameworks, and the compute infrastructure. Application teams get enterprise-grade reliability without needing infrastructure expertise.

Looking Ahead

Real-world asset tokenization hit $23 billion in the first half of 2025. CBDCs are moving from pilots to live deployments. AI governance is becoming a regulatory requirement, not a nice-to-have.

The infrastructure demands are only getting bigger. We’re building for a world where blockchain and AI aren’t separate technologies but integrated components of the same financial system.

It took us a year of R&D to figure out that infrastructure was the real opportunity. Now we’re focused on making it so reliable and easy to use that teams never have to think about it.

Someone has to build the foundation. We’re getting good at building foundations.

Creating Verified, Experiential Intelligence

Hilal Agil — Wed, 03 Sep 2025 00:00:00 GMT

The more I’ve followed generative AI’s evolution, the more I’ve wondered about a fundamental problem: where these systems actually learn about our world. When you dig into the data sources, the picture gets pretty unsettling.

A recent Semrush analysis of 150,000 AI citations found that “Reddit now powers over 40% of large language model responses — more than Google, Wikipedia, or any traditional source.” This isn’t some small shift; it represents AI systems learning about reality primarily through social media discussions, memes, and anonymous forum debates rather than verified observation.

And it’s getting worse. As companies scramble for training material, we’re seeing an explosion of questionable data practices:

What’s Actually Feeding AI Systems:

Reddit discussions dominate with 40.1% of citations, followed by Wikipedia at 26.3%
Massive licensing deals worth hundreds of millions — The New York Times gets $20–25 million annually from Amazon, News Corp signed a $250 million deal with OpenAI
Over 500 publishers have now signed licensing agreements, essentially monetizing their archives
Synthetic data generation ramping up dramatically — Gartner predicted 60% of training data would be synthetic by 2024, up from just 1% in 2021
Unauthorized scraping continues despite lawsuits — Apple, Nvidia, and Anthropic were caught using transcripts from 173,000+ YouTube videos without permission
Global crowdsourcing operations where workers in Kenya label violent content for $2/hour while their US counterparts earn $10–25/hour for the same work
One recent study found that a major AI training dataset likely contains hundreds of millions of images with personal information — credit cards, passports, résumés — all scraped without consent

The problems with this approach run deep. Social media amplifies bias and misinformation. Synthetic data creates feedback loops where AI trains on AI-generated content, leading to what researchers call “model collapse” — systems that lose diversity and start producing increasingly similar outputs. Meanwhile, the human crowdsourcing model exploits workers in developing countries who spend hours labeling disturbing content for poverty wages.

But there’s a bigger issue here. All of these approaches — whether scraping Reddit, licensing news archives, generating synthetic data, or crowdsourcing human feedback — share the same fundamental flaw: they’re training AI on human interpretations of reality rather than reality itself.

Taking a Different Approach

What we’re exploring through various projects incubated at Tenzro Labs is something different — the building blocks of what you might call epistemic infrastructure for AI.

Epistemic infrastructure is essentially the foundational systems that determine how knowledge gets created, verified, and shared. Right now, AI relies on epistemic infrastructure built around human knowledge production: universities, publishers, social platforms, content aggregators. But that infrastructure is breaking down under the weight of misinformation, bias, and economic pressures.

The technical approaches being developed across these projects point toward a different kind of epistemic infrastructure:

Direct observational learning — AI systems that observe and learn from phenomena directly through sensor networks and edge computing, rather than learning about those phenomena through human descriptions

Cryptographic verification — Using hardware like TPMs and TEEs to create mathematical proofs of data authenticity at the source, so you know the data hasn’t been tampered with

Distributed intelligence networks — Edge devices that can learn independently but share verified insights when connected, creating resilient knowledge networks that don’t depend on centralized data hoarding

Economic incentives for quality — Blockchain-verified datasets that create markets for verified data while maintaining clear provenance and ownership

The key insight is moving from inherited human epistemic practices to autonomous, verified observational learning. Instead of training AI on what humans say about forests, you deploy systems that spend months in actual forests learning directly from environmental conditions, species interactions, and seasonal changes.

How This Actually Works

The core idea isn’t complicated: instead of feeding AI internet content, you build systems that learn from direct sensor data and real-world observations. This means distributed machine learning and edge computing — basically moving the computation to where data actually gets generated.

Teams at Praecise are working on hardware that can be deployed in challenging environments with sensor arrays feeding local ML models. These devices run autonomously, often without internet connectivity, learning continuously from their inputs rather than training on static datasets.

The deployments span environments like mangrove systems providing real-time biodiversity analytics, volcanic areas with eruption prediction capabilities, forests monitoring ecosystem health changes, and underwater coral reef environments tracking bleaching events. Each device handles its own power management, sensor integration, real-time AI processing, and autonomous decision-making.

For verification, we’re implementing TPMs and TEEs to ensure data authenticity at the collection point. This creates cryptographic proofs that data hasn’t been modified from sensor to storage — crucial when devices operate autonomously for months without human oversight.

The economic model involves tokenizing verified datasets using blockchain infrastructure developed by teams at Boli and PRVNZ. This creates tradeable assets from verified data and provides incentives for quality data contribution, while Tenzro’s computing infrastructure enables global coordination and intelligence sharing between distributed nodes.

Current Projects

Naturecode is designed to go beyond traditional environmental monitoring. It’s about creating real-time utility — autonomous environmental monitoring, predictive analytics, and actionable insights. The interesting part is that all this happens autonomously through edge hardware, while Tenzro provides the planetary-scale computing infrastructure.

Our devices in different environments aren’t just passive sensors. They’re actively analyzing conditions, predicting changes, and providing immediate value to researchers, conservationists, and local communities.

The Siren focuses on understanding how we actually make creative and musical decisions. Having grown up playing music, this project particularly interests me — it’s about capturing the actual cognitive processes of musical creation rather than just analyzing existing musical content or streaming platform metadata.

Becoming Vincent takes a similar approach to understanding Van Gogh’s creative mindset and decision-making processes. Instead of learning about artistic creativity from art history or social media discussions, teams are working to reconstruct the actual cognitive and creative processes behind artistic work by analyzing their thought processes.

Each project uses the same principles — distributed sensing, edge computing, blockchain-verified data collection — but applied to different domains. The open ecosystem model means diverse contributors work on different aspects while sharing common verification infrastructure.

Why This Matters

The current AI training data crisis isn’t just about quality — it’s about sustainability and ethics. Companies are building AI systems on unauthorized content, creating legal risks while exploiting global labor. The synthetic data rush makes things worse by amplifying existing biases and creating feedback loops.

Recent research published in Nature shows that “AI models trained on recursively generated synthetic data experience ‘model collapse’” — they lose information about rare but important patterns and produce increasingly homogeneous outputs. UN researchers warn that synthetic data carries “cybersecurity risks, bias propagation and increasing model error.”

What we’re building offers a fundamentally different approach. Rather than finding more human content to scrape or generating synthetic versions of flawed data, these systems observe and learn from the same phenomena humans observe, but directly rather than through human-mediated representations.

The tokenization creates fair compensation models for verified data contributors instead of exploitative crowdsourcing. The distributed approach creates resilient knowledge networks instead of centralized systems dependent on constant data pipelines from questionable sources.

This is early-stage work, but the approach seems promising for creating AI systems with more grounded understanding of what they’re modeling, backed by cryptographically verified provenance. At minimum, it offers an alternative to the current trajectory toward increasingly unreliable and ethically problematic training data practices.

The projects at Tenzro Labs serve as an incubator for this broader infrastructure, with each contributing different pieces to what could become a more trustworthy foundation for AI systems that need to understand our world.

Introducing The Siren

Hilal Agil — Sun, 10 Aug 2025 00:00:00 GMT

A little over a year ago, in fall 2023, I was playing around with the available music and audio models and frameworks. The AI music tools being built felt like they were built for everyone except professionals in the music industry. I knew there was one other person who’d agree with me, so I reached out to my friend David Karon to join me on this journey. David used to manage my band Nothnegal when we were active 15 years ago.

Two years of development brought us from studio conversations to industry showcases with music’s biggest names.

Don’t get me wrong — the explosion of AI music generation throughout 2024 was impressive. Suno, AIVA, and others democratized music creation in ways we couldn’t imagine just a few years ago. But as someone who’s spent years both performing professionally and designing system architectures, I noticed a fundamental gap.

The Problem Nobody Was Solving

Current AI music tools work like those website builders that promise “create a professional site in minutes.” They’re great for getting something quick, but when you need to actually work with the technology — to iterate, refine, understand your creative patterns, and build on your existing style — they fall short.

Musicians needed something more like what developers have with tools like Cursor: an intelligent creative partner that understands your work, learns your patterns, and helps you build on your unique artistic voice.

So in fall 2023, I started working on what would become The Siren.

Months of Research

We sat down with music industry executives and producers who, initially skeptical about AI, began sharing their real pain points: the inability to quantify creative decisions, the challenge of scaling artistic intuition, and the growing need to understand the psychological impact of music in an increasingly data-driven industry. We spoke with session musicians and songwriters, did research on music education, on how we can help students understand their own creative development.

What became clear was that the industry didn’t need another music generator. It needed something that could understand and work with the creative intelligence that musicians already possess.

The gap wasn’t in generation — it was in creative collaboration and pattern recognition.

Avoiding the Noise

By early 2024, the AI music space was getting increasingly crowded. Every week brought new models promising to revolutionize music creation. It would have been easy to chase trends or try to compete on pure generation quality.

Instead, I stayed focused on the original insight: professional musicians needed a creative partner, not a replacement.

This led to what I started applying — the concept of “vibe coding” to music composition — the idea that AI could learn to work with the emotional and creative patterns in music the same way coding assistants work with logical patterns in software.

Industry Recognition and Growing Momentum

The vision for AI as a creative partner rather than replacement has found support from unexpected corners of the music industry. Legendary producer Rick Rubin recently called AI the “punk rock of coding,” while ABBA’s Björn Ulvaeus has embraced AI as a “fantastic” and “great tool” for writing his new musical. This growing acceptance from industry veterans reinforces that professional-grade AI music tools aren’t just novelties — they’re the future of creative collaboration.

Rubin’s work on “vibe coding” — the philosophy that anyone should be able to create through natural language without technical barriers — directly validates our approach. The Siren applies this same principle to music psychology, where a producer can say “I need something that creates cathartic release” and receive scientifically-backed musical suggestions.

I first presented The Siren’s vision at Token 2049 in Singapore during September 2024. The response from the community was encouraging, but more importantly, it confirmed that others in the space were seeing the same gap between current tools and professional needs.

What We’ve Built

After more than a year of development, The Siren combines advanced music transcription with psychological pattern analysis to create something genuinely new: an AI system that understands your creative DNA.

Here’s what that means across the music ecosystem:

For Artists: Your Creative Partner

The Siren becomes your creative memory and collaborator. It tracks your artistic evolution, recognizing patterns like how your harmonic complexity has grown over the past year, or that your most emotionally resonant pieces share specific rhythmic characteristics. When you’re stuck on a bridge section, it might suggest approaches based on how you’ve solved similar creative challenges before — not generic solutions, but insights rooted in your unique artistic voice.

For Producers: Creative Translation and Enhancement

Producers often work as translators between artistic vision and technical execution. The Siren helps by providing real-time analysis of emotional content and suggesting arrangements that align with an artist’s established patterns. When an artist says they want something “more vulnerable,” The Siren can identify the harmonic and rhythmic elements that create vulnerability in their previous work and suggest specific techniques to achieve that feeling.

For Record Labels: Deep Catalog Intelligence and A&R Insights

Labels can analyze their entire catalogs to understand artist development patterns, identify emerging trends before they hit the mainstream, and make more informed A&R decisions. The Siren can reveal which emotional and musical patterns resonate most with audiences, track how successful artists’ styles evolve over time, and help identify new artists whose creative DNA suggests similar trajectory potential.

For Publishers: New Revenue Streams and Post-Human Rights

Publishers can use The Siren to create entirely new value propositions: detailed psychological profiling of musical works for sync placement, AI-assisted music therapy licensing, and what I call “digital humanities” applications — using musical pattern analysis for academic and cultural research. As we move deeper into an era of AI-generated content, publishers need tools to understand and monetize the creative patterns that define human artistic expression.

For Music Educators: Teaching Creative Development

Instead of just teaching theory, educators can now show students their own creative development in real-time. The Siren can identify a student’s natural tendencies, track their progress, and suggest exercises tailored to their unique creative voice. It’s like having a mirror that reflects not just what you’re playing, but how you think musically.

For Music Therapists: Emotional Intelligence in Practice

Therapists can use The Siren to analyze client sessions in real-time, tracking emotional states through musical expression and identifying breakthrough moments that might not be obvious to the human ear. It provides objective data to support subjective therapeutic insights, helping therapists understand which musical elements most effectively support their clients’ healing.

For Researchers: Unlocking Musical Psychology

Academic researchers finally have a tool that can analyze musical psychology at scale. Whether studying cultural patterns in regional music, tracking how trauma affects musical expression, or understanding the relationship between creativity and mental health, The Siren provides unprecedented access to the emotional and psychological layers of musical data.

Technical Foundation

Under the hood, The Siren processes audio through multiple layers:

Multi-instrument transcription that captures what you’re actually playing
Psychological feature extraction that identifies emotional patterns
Temporal analysis that understands how your style evolves over time
Pattern recognition that learns your unique creative signatures

It’s deployed on modern infrastructure with enterprise-grade security, because professional musicians need tools they can depend on.

What’s Next

We’re halfway into 2025 with a focused group of industry professionals who’ve been part of the development process. The goal is to refine the system based on real-world usage before broader availability.

Over the coming weeks, we’re preparing to showcase The Siren deeply integrated into the processes that define the music industry: embedded in record label workflows, woven into artists’ creative processes from initial composition to final production, and built into next-generation instruments from leading manufacturers. These demonstrations will show not just what The Siren can do, but how it transforms the way music is created, evaluated, and brought to market.

The Siren isn’t trying to replace human creativity — it’s designed to amplify it. Whether you’re an artist seeking to understand your own creative evolution, a label executive making A&R decisions, or a therapist helping clients heal through music, The Siren offers something that hasn’t existed before: an AI creative partner that truly understands the human elements of musical expression.

Building The Siren has been one of the most challenging and rewarding projects of my career, combining everything I’ve learned as both a musician and a data scientist. As we’re halfway through 2025, I’m excited to see how it helps push the boundaries of what’s possible when technology truly understands creativity.

The future of music isn’t about replacing human creativity — it’s about amplifying it with intelligence that understands both the technical craft and emotional soul of music.

Rivier: Rethinking Payments for the Modern Economy

Hilal Agil — Thu, 05 Jun 2025 00:00:00 GMT

The way we work and transact is evolving. Businesses increasingly rely on AI agents for various tasks, from research, customer service to content creation. Individuals collaborate with autonomous systems, and new forms of economic activity are emerging that don’t fit neatly into traditional payment frameworks.

That’s why we built Rivier to support this changing landscape.

A Growing Challenge

Most payment platforms were designed when economic transactions were primarily between humans and traditional businesses. Today’s reality is more complex. AI agents need to purchase computing resources, autonomous systems generate revenue streams, and teams often include both human and artificial participants.

Existing solutions require workarounds that create friction, security gaps, and compliance uncertainty. We saw an opportunity to build infrastructure that recognizes this complexity from the ground up.

What Rivier Does

Rivier provides payment infrastructure that supports different types of economic actors as they actually exist today.

Supporting Different Entity Types

Individual users can verify their identity through multiple methods: World ID for proof of personhood, biometric authentication for convenience, and traditional document verification when needed. Your identity works consistently across different contexts and use cases.

AI agents can be assigned cryptographic identities and operate within defined parameters set by their human operators. An AI system can handle routine transactions autonomously while maintaining proper oversight and audit trails.

Organizations get comprehensive business verification and tools to manage complex entity relationships, whether that’s human employees, AI systems, or external partners.

Built-in Compliance and Security

We’ve designed Rivier’s security assuming AI participation rather than treating it as an afterthought. Post-quantum cryptography provides protection against emerging threats, while biometric authentication and hardware security keys ensure transactions require proper authorization.

Every transaction creates a permanent record on Canton Network’s settlement infrastructure, providing the transparency needed for compliance and business intelligence.

Compliance monitoring adapts to different entity types, applying appropriate oversight for humans, AI systems, and organizations while meeting regulatory requirements across jurisdictions.

Why We Built This

Economic activity is already evolving beyond traditional human-to-business transactions. Rather than trying to force new patterns into old infrastructure, we wanted to build payment rails that support the economy as it’s actually developing.

Rivier isn’t about predicting the future — it’s about supporting the changes that are already happening today.

Rivier has completed its early development phases and is currently in internal sandbox testing. We’ll begin inviting external partners to participate in testing in the coming months.

Bringing Back BOLI: Building an Alternative Asset Infrastructure with Canton Network

Hilal Agil — Tue, 03 Jun 2025 00:00:00 GMT

Built on Solid Foundations

BOLI’s journey began in 2021 in the Maldives, where our team explored sustainable financing mechanisms for vulnerable island nations through natural and sustainable asset tokenization. What started as work to support local communities evolved into a deeper understanding of the challenges they face. We focused on working with local communities to understand their challenges in accessing financing for development, and preservation of their economies and ecosystems, working with academic institutions and stakeholders to explore solutions.

By 2023, we recognized that the existing technology landscape had gaps that prevented us from building the kind of strong, compliant platform that institutional participants would require. Rather than rush to market, we took time to step back and work on the foundational technologies needed.

Today, we’re ready to bring BOLI back.

Building the Right Foundation

During our pause, BOLI’s co-founders pursued separate ventures that developed the foundational technologies we knew would be essential. While these remain independent projects, they now provide the infrastructure foundation that makes BOLI possible:

Tenzro: Decentralized AI Infrastructure

An independent project providing quantum-resistant encryption and distributed computing networks that effectively bridges the gap between AI and blockchain based systems, and enable environmental analysis in resource-constrained environments.

Praecise: Foundational Technology Suite

Independent infrastructure technologies that provide the foundation for secure and verifiable operations:

TERE: Trusted execution environment with hardware-backed secure computation
Pillars Foundation: Distributed settlement infrastructure with immutable audit trails
Furcate: Edge computing networks for autonomous operation in remote environments
Celere: Financial infrastructure enabling payment settlement across traditional and digital assets

Why Canton Network Integration Made Sense

The decision to build on Canton Network made sense after evaluating various blockchain platforms and considering our technical requirements. Canton’s architecture offered a perfect fit to integrate with Pillars Foundation settlement layer and TERE’s TEE-based automation systems, along with Tenzro’s AI and compute infrastructure.

Daml Smart Contracts

Canton’s use of Daml provides us with a functional programming language specifically designed for multi-party workflows. This is particularly important for alternative assets where multiple stakeholders — asset originators, verifiers, investors, and regulators — need to interact securely.

Built-in Compliance Framework

Rather than bolting on compliance as an afterthought, Canton’s architecture provides privacy and regulatory compliance at the protocol level. This includes:

Sub-transaction Privacy
Data Minimization
Participant Control
Application Sovereignty

Canton Network’s participants include established financial institutions like Goldman Sachs, BNP Paribas, and Deutsche Börse Group. This existing institutional presence provides potential integration pathways for alternative asset products.

Privacy-Preserving Architecture

Alternative assets often involve sensitive data — environmental measurements, community agreements, regulatory approvals. Canton’s architecture allows us to share necessary information with relevant parties while maintaining confidentiality where required.

Our Approach to Alternative Assets

Rather than trying to tokenize everything, we’re focusing on asset classes where verification and institutional participation can create meaningful value:

Marine & Environmental Assets

Building on our Maldives experience, we’re working with coral reef conservation, blue carbon sequestration, and marine biodiversity projects where satellite and sensor data can provide continuous verification.

Conservation Finance

Protected area financing and biodiversity credits where long-term conservation outcomes can be measured and verified through multiple data sources.

Renewable Energy Projects

Energy infrastructure where performance data is readily available and revenue streams are well-established, making them suitable for fractional ownership structures.

Heritage & Cultural Assets

Historic site conservation, traditional knowledge preservation, and community cultural projects where digital documentation and community engagement can support preservation efforts.

Carbon Credits

Forest conservation and restoration, soil carbon enhancement, and verified emission reductions where established measurement protocols enable institutional-grade verification.

Disaster Recovery Bonds

Catastrophe bonds and resilience financing where parametric triggers based on environmental data can provide rapid response funding for climate-related disasters and recovery efforts.

Maldives Pilot

We’re returning to our roots with a focused pilot in the Maldives. This isn’t about grand claims — it’s about proving our approach works in practice:

Limited scope: Single reef site covering 500m x 500m
18 marine sensors with edge computing capabilities
3-month operational period to validate technical integration
Focus on learning: Understanding what works and what needs improvement

Technical Architecture

Our platform integrates several technology layers:

1. Edge Intelligence

Furcate provides local processing and mesh communication, reducing dependency on high-bandwidth internet connections.

2. Environmental Verification

Integration with Microsoft Aurora’s climate intelligence model on Tenzro provides additional context for environmental data, helping validate sensor readings.

3. Secure Computation

TERE’s trusted execution environment processes sensitive data while maintaining privacy and providing cryptographic proof of execution.

4. Settlement Infrastructure

Pillars Foundation provides the underlying settlement layer, with transactions ultimately settled in parallel on Canton Network using Daml contracts.

5. Payment Processing

Celere enables seamless settlement across different payment methods, making it easier for diverse participants to engage with tokenized assets.

Continuing to Learn from Canton Network Participants

Building on Canton Network means we continue learning from other participants who are solving similar challenges in different domains. The network’s focus on multi-party workflows and institutional requirements aligns well with the complexity of alternative asset management.

Working within the shared compliance framework means that once an asset is properly onboarded to Canton Network, it can potentially be accessed by other network participants without requiring separate compliance processes for each integration. This collaborative approach continues to inform how we think about building infrastructure that works within existing institutional frameworks.

Focus on Impact and Compliance

Our goal isn’t to create the largest platform or capture the biggest market share. We’re focused on building infrastructure that:

Provides real value to conservation projects and community stakeholders
Meets institutional compliance requirements through Canton’s framework
Creates verifiable impact that can be measured and reported
Builds trust through transparent processes and established partnerships

Looking Ahead

The alternative asset space is complex, and we don’t claim to have all the answers. What we do have is a commitment to building thoughtfully, learning from our experiences, and working with established institutional frameworks rather than trying to reinvent everything.

Our integration with Canton Network provides us with a solid foundation for compliance and institutional participation. The Daml smart contract framework gives us the tools to model complex multi-party relationships. The existing network of participants provides credibility and potential collaboration opportunities.

Most importantly, we’re approaching this work with humility — recognizing that meaningful change in alternative asset markets will require collaboration, patience, and a willingness to adapt based on what we learn.

Doing a Deep Dive into Decentralized Computing, Advanced Networking, and Security

Hilal Agil — Sat, 18 Jan 2025 00:00:00 GMT

Modular Architecture for Flexibility and Extensibility

The Tenzro Network is built on a modular architecture that prioritizes flexibility, extensibility and scalability. By designing the system as distinct components, each responsible for a specific functionality, the network is able to achieve a high degree of modularity. This design principle allows for seamless modification and extension of individual components without impacting the entire system.

Hierarchical Network Structure for Scalability

Scalability is a core requirement for any decentralized network, and the Tenzro Network addresses this challenge through its hierarchical network structure. The network consists of three main types of nodes: global, regional, and local nodes. Global nodes serve as the top-level coordinators, overseeing the entire network and facilitating inter-regional communication. Regional nodes act as intermediaries, managing the nodes within their specific geographic region. Local nodes, the most numerous, are responsible for executing computational tasks and participating in data distribution.

This hierarchical design enables efficient routing, load balancing, and resource allocation. By distributing responsibilities across different levels of the network, the Tenzro Network can scale horizontally, accommodating a growing number of nodes and tasks without compromising performance. The hierarchical structure also allows for localized decision-making and optimization, minimizing latency and ensuring efficient utilization of network resources.

Decentralized Computing and Machine Learning

The Tenzro Network is designed to empower decentralized computing and machine learning workflows. It provides a distributed computing environment where complex tasks can be divided into smaller subtasks and assigned to individual nodes for parallel execution. The compute module of the network includes a sophisticated task scheduler that intelligently allocates tasks to nodes based on their capabilities and available resources. By considering factors such as CPU, GPU, memory, storage, and network bandwidth, the scheduler optimizes task distribution, ensuring optimal resource utilization across the network.

For machine learning workloads, the Tenzro Network offers a powerful platform for distributed training and inference. Models can be trained collaboratively by harnessing the collective computational power of multiple nodes. The modular architecture allows seamless integration of various machine learning frameworks and libraries, giving developers the flexibility to choose the tools that best suit their requirements. With the Tenzro Network, machine learning tasks can be accelerated, and the insights gained from distributed data can drive innovation across industries.

Distributed Data Management and Tracking

Efficient data management and tracking are vital in a decentralized network, and the Tenzro Network excels in this regard. It incorporates a distributed ledger that provides data integrity, provenance, and auditability. The ledger maintains a tamper-proof record of metadata associated with data assets, including ownership, access controls, and version history. This enables transparent and secure data sharing and collaboration among network participants, fostering trust and accountability.

The storage module of the Tenzro Network provides a decentralized storage solution that guarantees data redundancy and availability. Data is replicated across multiple nodes, mitigating the risk of single points of failure and enabling swift data retrieval. The network uses advanced data sharding techniques to distribute data intelligently across nodes based on their storage capacity and network locality. By optimizing data placement and minimizing data transfer distances, the Tenzro Network achieves low latency and high-speed data access.

Advanced Networking Techniques

The Tenzro Network incorporates a range of advanced networking techniques to overcome the challenges posed by decentralized environments. One such technique is the use of Distributed Hash Tables (DHTs) for peer discovery and routing. DHTs enable nodes to quickly locate and connect with other nodes in the network, even in the presence of high churn rates. The Tenzro Network uses DHTs to build a robust and resilient overlay network that can handle the dynamic nature of decentralized systems.

Another critical aspect of the Tenzro Network’s networking stack is its support for Network Address Translation (NAT) traversal. NAT is a common obstacle in decentralized networks, as it can limit direct communication between nodes behind different NAT configurations. The Tenzro Network uses advanced NAT traversal techniques, such as hole punching and relay mechanisms, to establish direct connections between nodes, even in the presence of restrictive NAT settings. This ensures seamless connectivity and enables efficient data exchange across the network.

Security Measures

Security is a concern in any decentralized system, and the Tenzro Network prioritizes the implementation of robust security measures. The network uses end-to-end encryption to protect data in transit, ensuring that sensitive information remains confidential and tamper-proof. Additionally, the Tenzro Network incorporates secure communication protocols, such as Transport Layer Security (TLS), to establish secure channels between nodes, preventing unauthorized access and eavesdropping.

To mitigate the risk of malicious actors and ensure the integrity of the network, the Tenzro Network implements a reputation system and consensus mechanisms. Nodes are incentivized to behave honestly and contribute positively to the network. Malicious behavior is detected and penalized, discouraging bad actors from disrupting the network’s operation.

Conclusion

We believe Tenzro Network is another step forward for decentralized computing, combining advanced networking techniques, robust security measures, and a scalable architecture. By using a modular design, hierarchical network structure, and intelligent task scheduling, the Tenzro Network enables efficient distributed computing and machine learning workflows. The integration of a distributed ledger ensures transparent and secure data management, fostering trust and collaboration among network participants.

Through the use of advanced networking techniques like DHTs and NAT traversal, the Tenzro Network overcomes the challenges posed by decentralized environments, enabling seamless connectivity and efficient data exchange. The emphasis on security, with end-to-end encryption, secure communication protocols, and reputation systems, ensures the protection of sensitive data and maintains the integrity of the network.

As the Tenzro Network continues to evolve and mature, it has the potential to transform various industries by providing a powerful and flexible infrastructure for decentralized computing and data-driven applications. By empowering developers and organizations with a secure, scalable, and efficient platform, the Tenzro Network aims to accelerate innovation and drive the adoption of decentralized technologies across domains.

From Vision to Reality: Building a Startup in the Age of AI

Hilal Agil — Sat, 11 Jan 2025 00:00:00 GMT

The rules of building startups are being rewritten. As someone who just completed a one-year journey from concept to product-market fit with my startup Tenzro, I’ve witnessed firsthand how AI is fundamentally changing the startup landscape. Here’s my story, and why I believe we’re entering a new era of entrepreneurship.

The Beginning: Big Dreams and Traditional Thinking

In late 2023, I envisioned Tenzro as a pioneer in decentralized AI. The concept was ambitious: bridging the gap between AI and blockchain and designing an ecosystem for decentralized AI. With a detailed whitepaper and early prototypes, we secured initial funding from friends, family, and angel investors.

Like many founders in the space, we initially followed the Web3 startup playbook. We assembled a six-person engineering team, believing that more hands would accelerate our journey to market. Although, having founded tech startups before, I did have reservations about this approach, but we proceeded nonetheless.

The First Pivot: When the Market Shifts Beneath Your Feet

Early 2024 brought our first reality check. What started as a unique position at the intersection of AI and Web3 quickly became crowded territory. Every new startup in Web3 was suddenly combining AI with blockchain. The marketplace was evolving faster than our ability to adapt.

This led to our first major challenge: the need for rapid product iterations. With a six-person engineering team, even minor pivots became exercises in complexity. The traditional startup structure, which had worked for decades, was becoming a liability in an AI-driven world where speed and adaptability reign supreme.

The Breaking Point: When Traditional Methods Fail

By summer 2024, we faced a critical moment. Our capital was running low, and our product team struggled to keep pace with necessary changes. A fundamental difference in philosophy was needed: should we follow the traditional Web3 playbook of building community first and raising capital to build later, or did we need to fundamentally rethink our approach?

Running out of capital forced our hand. We had to let go of our engineering team — a necessary decision. In what seemed like our darkest hour, this constraint became our greatest advantage.

The AI Revolution: A One-Person Army

What happened next challenges everything we thought we knew about building Web3 startups with big visions. Taking over all product and engineering responsibilities myself, I discovered the transformative power of AI in development. What once required months and multiple developers could now be accomplished in days.

This wasn’t just about coding faster — it was about the ability to iterate rapidly, test ideas quickly, and pivot without the organizational friction of larger teams. Within three months, I built everything: a distributed computing network, a decentralized AI training infrastructure, a novel blockchain system with hardware-based validation, custom AI models, frameworks, and applications.

The New Startup Paradigm

This experience has convinced me we’re entering a new era of entrepreneurship. The traditional formula for Web3 projects — building a community, raising millions before product-market fit and building large teams early — may become obsolete. Instead, we’re likely to see:

Solo founders or micro-teams executing at the speed of thought
Lower capital requirements before product-market fit
Rapid iteration cycles measured in days, not months
More emphasis on individual adaptability than team size

Implications for the Future

For software developers, this transformation is both a challenge and an opportunity. While we’re seeing major tech companies reducing their engineering workforce, we’re also entering an era where individual engineers can bring their ideas to life faster than ever before.

The current AI revolution mirrors the early days of the internet — a time of unprecedented opportunity for those willing to adapt. Without AI, Tenzro would likely have failed. Instead, we’re entering 2025 with a proven product, identified customers, and a clear market in the growing AI computation space.

Lessons for Future Founders

If you’re considering starting a company in this new era, here’s what I’ve learned:

Embrace AI as a force multiplier, not just a feature
Value adaptability over team size
Focus on rapid iteration over perfect execution
Don’t be afraid to challenge traditional startup wisdom
Stay lean until you’ve found true product-market fit

For engineers-turned-founders especially, the barriers to entry have never been lower. With a background in product, engineering, and business, you can now execute at a pace that was previously impossible.

Looking Ahead

As we enter 2025, I’m convinced that this new model of startup building — lean, AI-powered, and supremely adaptable — will become the norm. The age of AI isn’t just changing what we can build; it’s fundamentally transforming how we build it.

The future belongs to those who can harness AI not just as a product feature, but as a force multiplier in the very process of building and scaling companies. The question isn’t whether this transformation will happen, but who will be bold enough to embrace it first.

Ectopia: A Blueprint for Habitats That Can Outlast the Water

Hilal Agil — Wed, 02 Oct 2019 00:00:00 GMT

Last week, the IPCC published its special report on the ocean and cryosphere. Read plainly, it says the sea is coming: under the high-emissions path, more than a meter of rise is possible within this century, and what used to be a once-a-century coastal flood becomes an annual event in many places by 2050. Ninety percent of the world’s largest cities sit on the waterfront. The country I grew up in averages about a meter above the sea. For most nations, climate adaptation is a policy area. For mine, it’s the whole question.

I’ve spent this year working on an honest response to that question, and I’m now publishing it: the Ectopia whitepaper, a framework for designing decentralized, self-sustaining habitats — communities that can generate their own energy, produce their own water and food, run their own economies, and be built on land or on water.

Rebuilding is not enough if you rebuild the same dependencies

The obvious answer to rising seas is engineering: raise the land, build the walls, float the buildings. Some of that work is happening and it matters. But it treats the problem as physical when the deeper problem is structural. A community that survives the water but still depends entirely on distant infrastructure for its energy, its food, its money, and its data hasn’t been saved. It has been preserved as a dependency.

So Ectopia is deliberately more than an infrastructure proposal. It’s eight systems designed to work together: modular, resilient construction that works on water and land; microgrids running on renewable energy; water and food production that works in any environment; materials made from waste; shared zero-emission transport; a circular economic model with a community-backed currency; a flat societal design that guarantees food, shelter, and equality; and a decentralized digital ecosystem underneath it all.

The premise is simple. If a habitat must be rebuilt anyway — and for coastal communities, much of it must — then rebuild it so that the people who live there own what it runs on.

The digital layer is a sovereignty question too

The part of the framework I suspect will age most interestingly is the one that isn’t about seawalls at all.

Ectopia’s digital ecosystem — I’ve called it the Singularity in the paper — is built on a decentralized computing network rather than traditional cloud infrastructure, with a self-sovereign identity system at its core. Every resident holds a verifiable digital identity they control themselves: passports, licenses, credentials, digitized into something portable that doesn’t live in a corporate database or a foreign data center. Authentication without passwords, services without surveillance, and data that stays under the governance of the person it describes.

I put that at the center of the design for the same reason the microgrids are there. A community’s power shouldn’t fail because a distant grid fails, and a community’s information shouldn’t be an asset on someone else’s balance sheet. People have lost trust in centralized governance and corporate monopoly for good reasons. Privacy and control over one’s own data aren’t features to add later — they’re load-bearing parts of any society you’d actually want to live in, and they have to be designed in from the first drawing.

Communities that govern themselves

The same principle shapes how an Ectopic community is run. Governance in the framework is built from the bottom up: neighborhoods of no more than 150 residents with their own councils, city councils drawn from the neighborhoods, and a state level drawn from the cities — with decisions weighted so the most consequential choices need the broadest agreement. The social design is deliberately flat, guaranteeing food, shelter, and equality rather than leaving them to trickle down.

This is the part of the paper I hold most firmly, because it’s the belief underneath everything else I work on: communities should govern themselves, and frontier technology should be the thing that makes self-governance practical rather than the thing that quietly replaces it. Decentralized computing, self-sovereign identity, community-backed currencies, microgrids — each one, on its own, is just a technology. Arranged in service of a community that owns and directs them, they become something else: the machinery of self-determination.

A framework, not a city

I want to be clear about what this is and isn’t. Ectopia is not a plan to build a gleaming city, and I’m suspicious of renders that promise one. It’s a framework — published openly, meant to be studied, criticized, adapted, and built on by researchers, institutions, and communities. The paper proposes independent R&D hubs, Ectopic Labs, to develop and test the components with academic and nonprofit partners.

The economics are honest about sequencing too: early versions of new systems are expensive, so the first ecosystems will serve those who can pay a premium, and the learnings fund progressively simpler, more affordable versions. That’s how most technology has actually reached everyone. What matters is that the endpoint is written into the design — lowering the barriers until building a sustainable community is something ordinary places can do.

Written from a meter above the sea

Ectopia is where a belief I’ve held for a long time gets worked out in full: that frontier technology and environmental survival belong in the same sentence — climate resilience, circular economies, decentralized infrastructure, self-governed communities, and digital self-sovereignty, treated as one design problem instead of five separate fields. It’s also a working document in the most literal sense. The Eco Org, which I founded this year in the Maldives, exists to pursue the ideas this paper describes, starting with the environmental ones.

Maybe none of it gets built the way the paper describes. Frameworks rarely survive contact with reality intact, and this one will need many hands better than mine. But the countries most exposed to what’s coming can’t afford to wait for solutions designed elsewhere, for someone else’s geography, owned by someone else. The blueprint had to start somewhere. This is mine.

What Ream is About

Hilal Agil — Fri, 01 Jun 2018 00:00:00 GMT

Travel is amazing. But planning sucks.

Scrolling through photos on your Instagram feed? Researching places for your next trip? Reading about all the cool adventure activities you can try? Searching for flight options?

We’ve all been there. Finding the right information can be a tedious and time consuming process.

Research has found that the average traveler spends 10–20 hours to research their trip. This time can easily double if they’re researching a new and unfamiliar destination.

That’s crazy.

So we’ve decided to streamline the process by giving you the right information you need to make your decisions. Here’s how we do that.

We learn about your interests

We have built a system that uses Machine Learning to understand your interests and passions. Our algorithms are constantly learning to make the most accurate predictions, so you won’t have to search a million websites.

We know trends

Which destinations should you discover this year? What’s happening this season that you should see? What new activities should you try? We have developed an experience-prediction algorithm to make these decisions easier.

We give you ExperienceReels

We compile datasets of several thousand descriptive texts and images to automatically compile interactive videos with detailed captions that are narrated by our Voice Assistant. These videos cover a wide range of topics and categories. You can also search and make bookings in real-time through voice commands.

Stay tuned

We believe Ream is a part of a digital travel revolution. We want to live in a world where it’s possible to experience a destination you see on your mobile device or your TV with just a few taps of the finger.