By Jan Camenisch

Compute is the lifeblood of society. There’s hardly anything that we do today that does not involve software in one form or another.

Our current IT stack is complicated, sometimes fragile, and can no longer be reliably protected from cyber attacks. That’s why almost everybody has moved their software onto the cloud, trusting the cloud provider with that task.

The Cloud Control Problem

Now, this is problematic for two reasons. Handing over operations also means handing over the control of one’s software and data to these third parties. Worse, it’s not even working. The cost spent on IT is ever-growing, as is the damage caused by cybercrime. Clearly, a radically different approach is needed.

In fact, we all know a computer that is built very differently from the current IT stack and that does not suffer from these problems. Bitcoin.

Learning from Bitcoin’s Security

Bitcoin has never been hacked, despite the large amount of money at stake. You might think I’m crazy. Bitcoin wastes a ton of energy, it’s very slow, and it’s not even a computer but only a ledger that keeps track of who owns how much bitcoin. How could one possibly run all of the world’s software like that?

The reason that Bitcoin is secure is twofold.

How ICP Creates a Secure World Computer

Bitcoin is a computer (well, maybe more like a pocket calculator) that is created by a mathematical protocol. So, it’s pure math. It cannot be hacked; 2+2 will always be 4. But that’s not enough. You don’t want to trust a single entity running the protocol; it could be hacked, bribed, or be outright malicious. You want to have many parties running that same protocol, assuming the majority is honest. So if all of them, or at least the majority, tell you the same result, you’re good.

At DFINITY, we have taken that principle of building a computer from protocol and all the latest research in cryptography and distributed systems, and spent over a thousand person-years to design and build the Internet Computer Protocol. This protocol has been running in production since May 2021 without ever missing a beat nor ever being hacked. The protocol provides smart contracts with lots of storage, computational power, and communication bandwidth, just like any ordinary computer would.

The User Experience

This is how ICP delivers on the World Computer vision. Therefore, from an end user’s perspective, this feels and looks as if the software ran on AWS or Google Cloud. For instance, a user can just fire up a browser, enter the URL of an ICP smart contract, and bingo, they’re already using an application that runs on the Internet Computer. For the end user, the big difference is not visible. Because ICP is a protocol, the applications built by smart contracts on ICP are unhackable and unstoppable, just like the internet itself.

Benefits for Developers

This means that developers do not have to worry about the security and reliability of their applications and can focus on the application logic. That makes it far cheaper and more efficient to build and maintain applications. In fact, most applications that already run on the Internet Computer are built by very small teams.

Getting Started with ICP

I invite you to further explore ICP. Check out DFINITY’s YouTube channel for videos that explain how it works, fire up your favorite browser and try some of the many great dapps, or start building yourself by cloning some samples with ICP Ninja.

Join us in our journey towards the World Computer.

Learn More

• Discover the Internet Computer at internetcomputer.org.
• Try applications built on ICP in your browser at internetcomputer.org/ecosystem.
• For developers, explore the documents on building on the Internet Computer Protocol at internetcomputer.org/docs.

Other Articles from the World Computer Tech Talks Series

• Who Controls Your Digital Life? by Björn Tackmann
• The Bitcoin DeFi Renaissance by Thomas Locher

This article is adapted from a World Computer Tech Talk by Jan Camenisch, CTO at the DFINITY Foundation. You can watch the original video here.

Why ICP Is Essential was originally published in The Internet Computer Review on Medium, where people are continuing the conversation by highlighting and responding to this story.

The Ignition milestone is complete. This brings large language models (LLMs) to the Internet Computer, making it possible to use LLMs from within canister smart contracts with just a few lines of code.

What Is the Ignition Milestone?

The Ignition milestone, part of the Decentralized AI track of the ICP Roadmap, focuses on connecting canisters to off-chain LLMs. With this update, developers can use AI in their dapps without a complex setup, thanks to AI workers.

What’s New in Ignition

LLM Libraries for Easy Integration

To simplify connecting your canisters to LLMs, we’ve added libraries in three languages used on the Internet Computer: Motoko, Rust, and TypeScript.

These libraries make it quick to use LLMs in your dapps. For example, here’s how a canister can interact with Llama 3.1 using just a few lines of Motoko code:

import LLM "mo:llm";

await LLM.chat(#Llama3_1_8B).withMessages([
  #system_ {
    content = "You are a helpful assistant.";
  },
  #user_ {
    content = "How big is the sun?";
  },
]).send();

The above example can be found and tinkered with on ICP Ninja. It shows how a canister can send a query to an LLM and get a response, with minimal setup.

Here’s another example showing how to call LLMs with tools:

import LLM "mo:llm";

actor {
  public func example() {
    let response = await LLM.chat(#Llama3_1_8B)
      .withMessages([
        #system_ {
          content = "You are a helpful assistant."
        },
        #user {
          content = "What's the weather in Zurich?"
        },
      ])
      .withTools([LLM.tool("get_weather")
        .withDescription("Get current weather for a location")
        .withParameter(
          LLM.parameter("location", #String)
            .withDescription("The location to get weather for")
            .isRequired()
        )
        .build()
      ])
      .send();
  };
}

You can find a live demo of using tools here, as well as the source code of the demo here.

AI Workers

To connect canisters with off-chain LLMs, we’ve built a minimum viable product of AI workers, a simple system that lets canisters retrieve data from off-chain LLMs. AI workers handle the communication, so canisters can send requests and get responses from LLMs in real-time.

Currently, AI workers support Llama 3.1 8B, Llama 4 Scout, and Qwen 3 32B. These options give developers flexibility to choose the right LLM for their project.

Here’s a diagram illustrating how AI workers are used under the hood to relay prompts to off-chain LLM providers.

How it works:

1. Canisters send prompts to an “LLM canister” via the LLM libraries we mentioned. This is a special canister setup to receive LLM prompts.
2. The LLM canister stores these prompts in a queue.
3. AI workers continuously poll the LLM canister for prompts.
4. AI workers execute the prompts and return the response to the LLM canister, which returns it to the calling canister.

Real-World Use Cases

Developers are already using the LLM libraries to build creative dapps, such as:

• Wallets with chat features: Adding conversational AI to crypto wallets for better user interaction (soon to be launched in OISY)
• DAOs with sentiment analysis: Using LLMs to analyze community sentiment and guide portfolio decisions like in the Alice DAO.

These examples show how AI workers can support a range of applications on the Internet Computer.

Why This Matters and How to Get Started

The Ignition milestone makes it simple for developers to integrate LLMs into their Internet Computer projects, enabling new kinds of dapps, like chatbots, analytics tools, and AI-driven DeFi apps. By combining LLMs with Internet Computer features like Chain Fusion, HTTPS outcalls, and onchain randomness, developers can build creative and powerful solutions.

Ready to try it? Explore the LLM Chatbot project on ICP Ninja, check out the live demo, or dive into the code and examples in our repository.

Bringing LLMs to the Internet Computer was originally published in The Internet Computer Review on Medium, where people are continuing the conversation by highlighting and responding to this story.

Have you tried signing into Caffeine? You may have noticed something.

To login to Caffeine, you use Internet Identity 2.0. This is a new version of the authentication service for the Internet Computer, and it provides an upgraded experience.

This article explains what Internet Identity 2.0 is, its benefits, and how you can use the option to upgrade from Internet Identity 1.0.

What Is Internet Identity 2.0?

Internet Identity 2.0 is a complete redesign of the Internet Computer’s authentication service. It has been built to make the secure login process simpler and more user-friendly for everyone.

Key Improvements

The new authentication service brings several improvements.

• Passkeys: Internet Identity uses the latest passkey standards for enhanced security. Users can store their private keys on their devices, such as their smartphones, to use Internet Identity.
• No More Identity Numbers: With Internet Identity 1.0, users had to remember or store their identity numbers. This is no longer the case with 2.0. This is possible thanks to a technology called Discoverable Passkeys. The main benefit for users is a simpler login without needing to track a number.
• Sign-in with Google: The upgraded Internet Identity includes the option for users to use their Google Account as an authentication option. This makes it easy for the billions of Google Account users around the world to sign up to Internet Identity with a familiar method.
• A Redesigned Interface: The entire login process has a fresh design that is more intuitive to navigate.
• Multiple Accounts from One Identity: You can create various accounts for a single application from one identity. This removes the need to manage separate identities for different roles, like a test account.

How to Upgrade Your Internet Identity

The upgrade process is straightforward and does not affect your data or login information. After the upgrade, your identity works with the new system while remaining fully usable with existing applications.

There are two options.

• Visit id.ai directly and follow the upgrade steps.
• If you already have an Internet identity, upgrade the next time you log in to Caffeine or any app using id.ai by following these three upgrade steps.

Steps to upgrade:

1. Click on “Upgrade.”

2. Enter your Internet Identity number. You will then be prompted to authenticate with your current passkeys.

3. Enter a name for your identity. Internet Identity 2.0 introduces names for the identities. These are not unique names, but names that help you identify your Internet Identity. You will be prompted to create a new passkey after this step.

That’s it! Simple and quick.

Key Questions About the Upgrade

What happens now if I do not upgrade from Internet Identity 1.0? You cannot use your existing identity to sign into id.ai or use applications that require the new login system, including Caffeine.

Will I lose my login information or data if I upgrade? You won’t lose anything. Your login information remains unchanged, and there is no data loss.

Will apps using Internet Identity 1.0 stop working? No. Apps using Internet Identity 1.0 will continue to work as usual. Your upgraded identity is compatible with both the existing and new systems.

What is the “primary account” I see after upgrading? This is the same account you used with Internet Identity 1.0. To continue using your existing account with your applications, you only need to select this default option in Internet Identity 2.0.

Should I upgrade? For now, you only need to upgrade for certain apps like Caffeine. As the Internet Identity 2.0 is rolled out, in the future, other apps will also be updated to use 2.0.

You can find more information on the the Internet Identity upgrade FAQ page.

For Developers

To find out more about integrating Internet Identity 2.0. in your apps, read these posts on the Internet Computer Developer Forum:

Internet Identity 2.0

Existing Users Upgrade

Ready to upgrade or create an identity for the first time? Visit id.ai.

Internet Identity 2.0: The New User Experience was originally published in The Internet Computer Review on Medium, where people are continuing the conversation by highlighting and responding to this story.

By Thomas Locher

Bitcoin DeFi is an important development in the crypto space, so let’s start with a quick introduction.

Bitcoin is the first and most well-known cryptocurrency. Even though there are many other crypto projects, Bitcoin still commands a large share of the market cap of the entire blockchain space, and its popularity continues to grow.

Another area that has seen remarkable growth in the last few years is decentralized finance, or DeFi for short. DeFi is a term used for financial services and products that operate without any centralized entities, like banks, and that enable peer-to-peer transactions and decentralized applications.

The Challenge of Bitcoin DeFi

Bitcoin DeFi is simply DeFi that makes use of bitcoin, or BTC for short. However, it is actually not that simple to use bitcoin in DeFi because Bitcoin, the network infrastructure, was intentionally designed to be simple and limited in functionality.

This limitation restricts the use of bitcoin in decentralized applications, including DeFi. The significant growth in DeFi has happened to a large extent without the involvement of bitcoin. Obviously, it would be great if bitcoin’s large market cap could be unlocked and used in DeFi applications. So far, most attempts at Bitcoin DeFi use some sort of centralized intermediary to make it work, which goes against the decentralized nature of Bitcoin.

A Solution on the Internet Computer

How can the Internet Computer help here?

The Internet Computer is directly integrated with Bitcoin at the networking level, which means there is no need for any centralized intermediaries, making the interactions reliable and secure. This integration makes it possible for smart contracts on the Internet Computer, called canisters, to get information about the Bitcoin blockchain, such as the balances of bitcoin addresses. It is also possible for canisters to have bitcoin addresses themselves and make bitcoin transactions.

In short, canisters can hold and transfer bitcoin, just like a wallet. However, because they’re programmable, they can implement any program logic. In other words, any decentralized application, and in particular any DeFi application, can be built around bitcoin on the Internet Computer.

To give you a few examples, bitcoin can be used to build tokens that are backed by real bitcoin in a decentralized manner.

For instance, ckBTC is such a token on the Internet Computer. The ckBTC minter canister guarantees that one ckBTC is always backed by one bitcoin. But unlike bitcoin, ckBTC can be transferred in a few seconds and at low cost, which makes it suitable for micropayments.

Another example is decentralized exchanges that can offer trading pairs involving bitcoin or fungible tokens on Bitcoin. You can build lending and borrowing services where you can earn interest on your bitcoin or use it as collateral. You can also build bitcoin-based escrow services with automated payments.

To summarize, Bitcoin DeFi is possible on the Internet Computer and without any centralized intermediaries, offering exciting opportunities for developers and users.

Join us on our journey toward the World Computer.

Learn More

• Discover the Internet Computer at internetcomputer.org.
• Watch Liquidium discuss how they addressed bridging problems with native asset lending on the DFINITY YouTube channel.
• For developers, explore the documents on building on Bitcoin at docs.bitcoindefi.dev.

Other Articles from the World Computer Tech Talks Series

• Who Controls Your Digital Life? by Björn Tackmann
• Why ICP Is Essential by Jan Camenisch

This article is adapted from a World Computer Tech Talk by Thomas Locher, Staff Researcher at the DFINITY Foundation. You can watch the original video here.

The Bitcoin DeFi Renaissance was originally published in The Internet Computer Review on Medium, where people are continuing the conversation by highlighting and responding to this story.

By Björn Tackmann

As more elements of our lives move online, the amount and value of the data we amass also grows rapidly. Our communication in emails, messengers, and social media apps, our identity information, and just so much more — all this data is now online, moving seamlessly between our devices.

To achieve this, however, all that data must be stored somewhere, on some computer. If that’s not your computer, it is someone else’s. That means this “someone else” who controls that computer is ultimately in control of your data. Someone else also decides what you get to do with your data. You might even lose access.

But there’s more: that someone else who is in control also decides who else gets access. Your data has long been used to target you with advertisements to make you buy more stuff you don’t need. If you’ve read any of the end-user license agreement updates you’ve probably signed off on over the last year, you will have seen that many companies now also claim the right to train their AI on your data.

What’s the Alternative?

Running your own services is actually hard. Operating your own infrastructure involves hours and hours of work.

The Internet Computer Protocol (ICP) enables services to be built in a different way. It allows you to store your code and data without giving up control, so you are still in charge. It also accomplishes this without requiring you to operate your own infrastructure.

Fundamentally, you can think of ICP as a computing platform built around the concept of ownership for code and data: a sovereign cloud, a true World Computer.

This is achieved by building on a decentralized network. ICP is based on the same underlying principle as Bitcoin. However, more than a thousand person-years of research and development efforts have enabled the Internet Computer Protocol to expand how decentralized technology is used. This technology is no longer a simple token ledger for monetary transactions. Its use now includes a full-fledged computing platform that can host all your code and data and keep you in control.

Data sovereignty is in your hands with ICP.

Join us on our journey toward the World Computer.

Learn More:

• Discover the Internet Computer at internetcomputer.org.
• Watch tutorials on the DFINITY YouTube channel.
• For developers, start building today at icp.ninja.

Other Articles from the World Computer Tech Talks Series

• The Bitcoin DeFi Renaissance by Thomas Locher
• Why ICP Is Essential by Jan Camenisch

This article is adapted from a World Computer Tech Talk by Björn Tackmann, Head of Research at the DFINITY Foundation. You can watch the original video here.

Who Controls Your Digital Life? was originally published in The Internet Computer Review on Medium, where people are continuing the conversation by highlighting and responding to this story.

This year, we launched the World Computer Tech Talks series with a clear mission: shine a light on the groundbreaking technology of the Internet Computer Protocol. Across eleven episodes, experts from the DFINITY Foundation broke down core concepts that power the World Computer vision.

With the first season now complete, we want to take a look back at key themes our speakers explored. After the incredible feedback we received from the community for the Tech Talks, we also want to announce that we will replay the entire first season as we prepare for the next.

The Big Picture: Why ICP is Essential

To understand the Internet Computer, it is important to first grasp the problems it solves. In his episodes, the DFINITY Foundation’s CTO Jan Camenisch explained that our current IT stack is fragile, complex, and cannot be reliably protected from cyberattacks. He positioned ICP as a revolutionary solution, taking the security principles of Bitcoin and applying them to general-purpose computing to create a true “world computer.”

He described ICP as fundamentally different from other blockchains, a platform that provides developers with immense storage and computational power while remaining unhackable. As Jan put it, the protocol “has been running in production since May 2021 without ever missing a beat nor ever being hacked.” This foundation of security and reliability is what makes everything else possible.

Digital Sovereignty and Security

Building on that foundation, a central topic throughout the series was digital sovereignty.

In his episodes, Head of Research Björn Tackmann tackled this from two angles: the personal control of our data and the organizational need to defend against insider threats. He argued that the current internet model forces a trade-off. “If that’s not your computer, well, then it’s someone else’s,” he stated in his talk on digital life. “But that means that this ‘someone else’ who controls that computer is ultimately in control of your data.”

His other talk focused on how ICP’s architecture solves this at an enterprise level. By removing centralized points of control, the protocol provides security guarantees that traditional IT cannot. This is because, as Björn explained, the integrity of code and data on ICP is cryptographically secured by the network itself, making it resilient to both external attacks and malicious insiders.

A Deep Dive into True Decentralization

What does decentralization really mean?

In his talk, VP of Engineering Samuel Burri went beyond the buzzword to explain ICP’s unique approach. He emphasized that building a resilient platform requires eliminating every single point of trust. “To achieve that, we require a compute and storage platform that is extra secure and resilient,” he said. “There must be no single point of failure.”

His session detailed how ICP’s deterministic decentralization, which uses a network of vetted, independent node providers, provides robust security without the immense energy costs of other systems.

Viewers also learned about the Network Nervous System (NNS), the onchain DAO that governs the protocol, executing upgrades autonomously and transparently.

Uniting a Fragmented Web3 with Chain Fusion

Several episodes were dedicated to Chain Fusion, ICP’s technology for seamless cross-chain interoperability.

Senior Director of Engineering Manu Drijvers explained how this feature allows ICP smart contracts to natively control assets on other blockchains, including Bitcoin, Ethereum, and Solana. “With Chain Fusion on ICP, we can build native multichain DeFi, bringing DeFi across chains together in a secure and seamless way,” he highlighted in his talk.

The series showed how this technology directly addresses the fragmentation of the current Web3 space, eliminating the need for risky centralized bridges.

In his episode on the Bitcoin DeFi renaissance, Staff Researcher Thomas Locher demonstrated the practical applications. “Canisters can hold and transfer Bitcoin just like a wallet,” he noted, “but because they’re programmable, they can implement any program logic.” This unlocks Bitcoin’s vast liquidity for a new wave of decentralized finance.

Reimagining the User and Developer Experience

A recurring message in the series was the importance of accessibility.

Several speakers, including Director of Engineering Georgi Harizanov, explained how ICP was designed for a low-friction experience. He pointed out that a true World Computer “should be appealing to all different types of users, including those that are not technically savvy.”

The talks highlighted key features that make this possible:

• Browser-Based Access: Users can interact with dapps on ICP using a standard web browser, with no special software needed.
• Reverse Gas Model: Developers prepay for application resources, so users can engage with services for free.
• Full-Stack Onchain Development: Samuel Burri’s episode on building dapps showed developers how they can host both an application’s frontend and backend entirely onchain, a unique capability of ICP.

Exploring the Frontiers of Onchain Technology

Finally, the series offered a glimpse into the future.

Islam El-Ashi, Head of the AI team, dedicated his talk to sovereign AI, explaining how users will soon be able to run personal AI agents on ICP. “Your assistant and agents will work for you and only you,” he assured viewers, emphasizing the data privacy benefits.

What’s Next? Get Ready for Season Two

The first eleven episodes of the World Computer Tech Talks laid the foundation for understanding the power and potential of the Internet Computer. We are grateful to everyone who tuned in and engaged with the series.

The conversation is far from over. Work on a second season is already underway. We plan to dive deeper into the Internet Computer and beyond. For example, the topics will include how storage works on ICP, an introduction to the Motoko programming language, the Network Nervous System (NNS), decentralized autonomous organizations (DAOs) more broadly, and more.

In the meantime, you can watch the entire first season on the DFINITY YouTube channel. You can also catch the episodes as they are rereleased each Friday on X and LinkedIn.

Join us on our journey toward the World Computer.

From Decentralization to AI: World Computer Tech Talks Recap was originally published in The Internet Computer Review on Medium, where people are continuing the conversation by highlighting and responding to this story.

For many developers, the modern cloud is a double-edged sword. It offers unprecedented power and scale, but often at the cost of control. Data is opaque, policies are complex, and the platform can change the rules at any moment. It’s a frustration that seasoned developer David Dal Busco knows all too well. Instead of accepting the status quo, he embraced the classic engineering mindset: “I’m going to do it myself.”

The result is Juno, a platform-as-a-service built on the Internet Computer that offers a powerful alternative to traditional cloud providers. The project was born from a desire to give developers the tools they need without forcing them to give up ownership. We sat down with David to learn more about the origins of the project, the idea behind Juno, and how it empowers developers on the Internet Computer.

The Birth of Juno

David is a freelance developer with nearly a decade of experience, much of it spent building mobile applications for clients. For years, he has balanced his client work with a passion for building his own projects on the side. His dual perspective as both a service provider and an independent creator gave him a unique insight into the needs and pain points of the modern developer.

“When I build mobile applications, I use serverless providers. I used Firebase, mainly,” David explains. “And there wasn’t that construct on the IC at the time… If I want to build an app over the weekend, I don’t want to reinvent the wheel. I want to start with some tools.”

This need for a streamlined, ready-to-use toolset, combined with a growing frustration with the existing developer experience on Web3, sparked the idea for Juno. His motivation went beyond simply filling a market gap; he was building the solution he wished he had.

What is Juno?

At first glance, Juno might seem familiar to developers. In the past, David has described it as the Google Firebase alternative for Web3, and the analogy holds to a point. Like Firebase, Juno provides a comprehensive suite of backend services out of the box, including authentication, data and file storage, hosting, and serverless functions, all designed to let developers build and launch applications quickly.

But as David is quick to point out, the comparison only goes so far.

“I would say it’s a serverless cloud provider as they know, but there is a twist,” he says, “and the twist is that they are in charge.”

This is the core of Juno’s philosophy: empower the developer. With Juno, there is no central intermediary. Developers maintain full control over their applications, their data, and their funds. The platform’s architecture, built entirely on the Internet Computer, ensures that Juno itself has no access to or control over a developer’s creations.

Why the Internet Computer?

The choice to build on the Internet Computer was fundamental to achieving this vision. The Internet Computer provides the decentralized foundation necessary to deliver on Juno’s promise of true ownership.

“When you use the IC, it’s transparent that you have the security and that you have the control,” David notes. While one can trust the security of major cloud providers, the Internet Computer’s architecture makes that security inherent and verifiable, moving it from a policy promise to a technical guarantee.

Juno leverages the Internet Computer’s unique capabilities, like its smart contracts, which are called canisters. As described by David in his white paper for Juno, this allows Juno to give each developer their own “Mission Control” smart contract. This acts as a central hub for all their projects and even functions as a wallet, ensuring that control remains firmly in their hands. The Internet Computer’s ability to serve web content directly from the blockchain at “lightning speed,” as David says, enables Juno to offer a seamless, 100% onchain experience.

David’s ultimate vision is to abstract the complexity of the blockchain away entirely. “My dream goal is having a web developer that uses it that doesn’t even know that there is a blockchain behind it.”

From a Blank Page to a Full Dapp

Juno is designed to address the countless small but time-consuming tasks that every developer faces.

“You don’t start with a blank page,” David explains. “You, or your AI coding assistant, shouldn’t have to build authentication and think about the permission system. If you want your user to upload a file, you shouldn’t have to build that part. With Juno, all the small tools that you would have to build are already there.”

This frees developers to focus on what makes their application unique instead of on boilerplate infrastructure. The potential is already being demonstrated by the community, showcasing the diverse applications that can be built on the platform.

Community Showcase: What’s Being Built on Juno

The community is already building a diverse range of applications. Toolkit, for example, is a versatile suite for managing Service Nervous Systems (SNS) and other projects on the Internet Computer, acting as a key driver for governance and collaboration. Another project, Solutio, was born from the vision that a genius idea can come from anywhere. It provides a platform for crowdfunding and developing solutions, enabling communities to propose ideas and fund the features they need most.

For David, the most exciting prospect is seeing the unexpected emerge. A prime example is Gods Are Dead, a play-to-earn role-playing game where players fight enemies and craft powerful items. “A few months ago, there was a person who built a game, an RPG,” he recalls. “I never expected someone to build that… you’re like, ‘Oh yeah, that’s a really cool demo.’ This kind of thing… I want to be surprised.”

The Road Ahead

Juno is continuously evolving. David has just launched a deeper GitHub integration for automated deployments; he is working on improved TypeScript support for serverless functions; and he is exploring tools to support team collaboration. He is also focused on artificial intelligence. “I’m currently working on providing support for developers who want to work with AI,” David says, “whether it’s with vibe coding tools or getting better answers through AI assistants.” For example, he is working on providing llms.txt files, which help AI models understand documentation more effectively.

The mission remains the same: to grow the ecosystem by providing developers with the best possible tools while staying true to the principles of decentralization and ownership.

Juno represents a significant step forward, demonstrating the power of the Internet Computer to reshape the landscape of web development. It’s a testament to what a single developer can achieve when they identify a problem, envision a better way, and build a solution that puts control back where it belongs: in the hands of the creators.

Ready to take control of your development stack? Visit juno.build to get started and see what you can build.

How Juno Aims to Put Developers Back in Control was originally published in The Internet Computer Review on Medium, where people are continuing the conversation by highlighting and responding to this story.

We are excited to announce the successful completion of the Levitron milestone, a major step forward in transparency and insights for the Internet Computer Protocol (ICP). With this milestone, API boundary node (API BN) access logs become publicly available in real-time, offering unprecedented insight into how the Internet Computer and its hosted canisters are being used.

The edge of the Internet Computer: the API Boundary Nodes

API BNs act as the public edge of the Internet Computer. They are the first point of contact for every incoming request, forwarding each one to the appropriate subnet, replica node, and, ultimately, the target canister. This means they capture and log every single external request for the Internet Computer.

Unlocking new insights with public access logs

Public access logs are an invaluable resource for both canister developers and the broader ICP community, and their availability provides a wealth of information that was previously inaccessible. For example, they allow the following.

• Tracking canister usage: For the first time, developers can observe the full spectrum of canister interactions, including query calls, which were previously not visible.
• Gauging network activity: The broader community can track activity across popular dapps and the overall network.
• Providing data for analytics platforms: Ranking and analytics platforms can directly integrate these logs to provide accurate metrics directly from ICP.

Until now, these logs were not publicly accessible. However, in true Web3 spirit, the Levitron milestone changes this: each API BN now exposes an endpoint, allowing clients to subscribe to and stream these logs directly.

Decentralized, direct, and detailed

This new access is entirely decentralized, meaning there is no intermediary. Interested parties can stream information directly from the API boundary nodes of the Internet Computer itself.

Log streaming happens on a per-canister basis, providing detailed information such as:

• The method being called.
• The status code of the response.
• The client making the call.
• The cache status of the response (hit, miss, or bypass).
• The error information (if available).
• The request and response sizes.
• And more!

Prioritizing privacy

It is important to understand that these logs are designed with privacy in mind. They do not contain any information about the actual payload of the call, IP addresses, or the sender principal. The only user-related information provided is the client ID, which is a salted hash of the IP and the sender. This salt stems from the salt-sharing canister, is rotated every month, and is only accessible by the API BNs, meaning no other entity knows this salt. This ensures valuable usage data is accessible without compromising user privacy.

Real-time streaming and future enhancements

You can now stream logs in real-time, starting today! It is important to note that this direct streaming mechanism does not provide historical data. If you want to analyze past activity, you will need to build your own database to store the streamed logs.

To help you get up and running, we have provided instructions in our documentation, along with a sample client that is ready for use.

Looking ahead, once SEV-SNP is rolled out to the API boundary nodes, clients will be able to attest to these nodes. This will significantly enhance the authenticity and integrity of these logs, giving you even greater confidence in the data.

For more in-depth information and technical details, be sure to check out the official documentation and the sample client.

Real-time Transparency with Public API Boundary Node Access Logs was originally published in The Internet Computer Review on Medium, where people are continuing the conversation by highlighting and responding to this story.

The pace of development of AI has been unyielding. There’s been plenty of discussion around “vibe coding,” but this is clearly just the beginning of something. So what’s next?

The internet and our digital lives are about to change fundamentally with the Self-Writing Internet.

In the form of Caffeine, the DFINITY Foundation has created a fully end-to-end, decentralized AI development platform. This goes beyond the promise of vibe coding, which is limited more to developers. Caffeine will enable everyone, technical and non-technical, to define their digital world on their terms.

At Dogpatch Studios in San Francisco, the Hello, Self-Writing Internet event brought together 1,000 people in person and 50,000 on the livestream to feature the first-ever Caffeine hackathon and discussions on the future it promises. This article captures some of the highlights of the day, demonstrating the possibilities of Caffeine and the key questions raised by the new paradigm.

The Wait is Over

For those in attendance and watching online eager to get their hands on the new technology, a key announcement came when Dominic Williams, Founder and Chief Scientist of the DFINITY Foundation, took to the stage.

“Tomorrow, we’re starting to give people access to those who’ve signed up for the alpha,” he revealed. “It’s going to roll out slowly and carefully, but please make sure you’re in the line if you want to try it.”

This marked the official opening of Caffeine to the public. Invite codes are being released on a rolling basis, giving the first users the opportunity to start building apps just by talking.

While the announcement highlighted how Caffeine is being opened up to the world, a select group at the event was already putting this promise to the test.

The First Caffeine Hackathon

“If you can describe it, Caffeine can build it.”

— Bruno Calabretta, Head of Ecosystem Hubs, The DFINITY Foundation

At the Caffeine hackathon, the atmosphere was electric as a packed room of 80 creators, split evenly between technical and non-technical backgrounds, prepared to make history. The goal was clear: bring an idea to life.

The creators took this to heart, diving into their projects with such focus that some did not want to take breaks. The energy was infectious; soon, participants were answering each other’s questions with the suggestion to “Just ask Caffeine!”

“The feedback and the response were amazing,” said Bruno Calabretta, who coordinated the hackathon. “People were so ecstatic about Caffeine, how it works, what they can do with it.”

A dynamic emerged between the two tracks. Participants with technical backgrounds initially approached Caffeine as a traditional coding tool. They realized they did not have to write code from scratch, but could instead leverage their expertise to refine what Caffeine built.

The non-technical participants, however, came with a different mindset. “Non-technical people, they really came with the goal of exploring, of seeing what could come out of it,” Bruno observed. “They knew that they had an idea…and they tried for six hours to make it a reality. And the reality is that it happened pretty fast.” This allowed them to spend most of their time collecting feedback and improving the user experience, a crucial part of product development. The results were staggering, proving that a background in programming is no longer a prerequisite for application creation.

Hackathon Winners

After hours of intense creation, the numbers spoke for themselves. While participants could only make one final submission, over 240 apps were created from scratch throughout the day, resulting in over 75 final submissions for the judges. Some projects also went through 10 to 25 iterations.

As the deadline approached, the collective focus was so intense that when asked if they needed more time, the room answered “Yes” as one, earning them another fifteen minutes of creation.

From this incredible pool of creativity, the judges selected the winners from each track, with a prize pool of $25,000 for first place, $12,000 for second, and $7,500 for third in both categories.

The winning projects were announced on the main stage to close the event.

Technical Track Winners

• First Place: Road Patrol, which enables citizens to improve their cities through map-based reporting, including things such as potholes, broken street lights and more.
• Second Place: K, which scans your AI generated apps and projects for security gaps and other vulnerabilities.
• Third Place: SkillSprout, which makes learning new skills more engaging by enabling you to collect points and achievements.

Non-Technical Track Winners

• First Place: BlueLens, which enables you to monitor the safety of your drinking water with real-time data and analytics, including an integrated AI chat tool.
• Second Place: Will Maker, which allows you to generate legal documentation such as a will, power of attorney, and healthcare directives for California.
• Third Place: DentalTracks, which enables you to accumulate all your dental records in one place and keep them private and secure.

The diversity of the winning applications highlights Caffeine’s broad appeal. Projects like BlueLens, DentalTracks, and Will Maker demonstrate how the general public and entrepreneurs can build sophisticated, real-world solutions for specific needs without writing a single line of code. Road Patrol showcases the platform’s potential for enterprise and civic tech, enabling the creation of robust management tools. Meanwhile, K, the security scanner, is a prime example of how developers can leverage Caffeine to build specialized tools for their own workflows, moving beyond vibe coding to create complete, self-writing solutions.

The success of the non-technical participants was a powerful validation of Caffeine’s mission. But this event was more than just a showcase of technology. As Bruno Calabretta emphasized on the day, “We are making history today. The people who were part of the hackathon were the first people in the world who were able to use Caffeine and to leverage its potential.”

Caffeine Experience Pods

The excitement was not confined to the hackathon. At the back of the main stage, a dedicated space allowed attendees to test Caffeine for themselves.

These Experience Pods were an instant draw. As Pierre Samaties, Chief Business Officer, noted, “People were piling up at the experience pods to try out Caffeine.” Attendees rushed to the screens the moment they opened, building real tools on the spot, from managing invoicing workflows to tracking job applications, turning their ideas into reality in minutes.

The Next Hackathons

Those gathered at the event were the first to experience Caffeine, but this is just the beginning.

The DFINITY Foundation plans to take this experience on the road with a series of in-person hackathons in multiple cities. For those inspired to participate in the future, Bruno offers this advice:

“Start thinking about something that you want to build, that you’ve been thinking about for a long time but you’ve never been able to implement. Go and look for that idea, that you know is going to be massive because now you can really build it.”

The message is clear: the Self-Writing Internet is here, and the next wave of creation is open to everyone.

Keynotes and Demos

Dominic Williams, Founder and Chief Scientist of the DFINITY Foundation, took the stage for a series of keynotes and live demos that moved from the theoretical to the tangible, describing the vision for the Self-Writing Internet and the groundbreaking technology that powers it.

A New Paradigm

In his first keynote, Dominic framed the current moment as one marked by a fundamental paradigm shift. He noted that “in the future, and really now with this technology, app development is a conversation.” He outlined the evolution of the internet, describing its development from craft coding to vibe coding, and now to self-writing. He described how in this new era, the ability to build applications through natural language democratizes creation for potentially 5 billion smartphone users worldwide. This shift, he explained, is enabled by a new, specialized tech stack built for AI, one which results in sovereign and tamperproof apps that are fully owned by their creators.

Data Persistence and Live Demos

In his next session, Dominic delved into one of the key technologies that makes this new era possible: enhanced orthogonal persistence. He explained that this tech, which allows data to live directly in the code without a traditional database, is the key to ensuring data is never lost during the rapid, conversational upgrade cycles of self-writing. To prove it, he built and deployed three applications live on stage: a blog, a user photo gallery, and a task manager. He then upgraded the task manager with new features, demonstrating how the app’s existing data was migrated safely and seamlessly.

From Conversation to Commerce

Before his final demos of the day, Dominic unveiled the upcoming Caffeine App Market, a new, open model for app distribution where users can “clone, not install” applications, allowing them to customize and build upon the work of others in the Caffeine ecosystem. This will enable Caffeine users to monetize their creations. To showcase the App Market, he cloned two existing applications, a task management app and a jewelry e-commerce store.

But there was more to Caffeine’s e-commerce potential. To demonstrate the platform’s power to build from the ground up, Dominic also created a golf store from scratch with a single prompt, which included a Stripe integration for payments.

As he concluded, “This system is designed to enable individuals and organizations to create production apps that are secure, resilient, and sovereign entirely through conversation.”

The Conversations

Beyond the live demos and hackathon creations, the day was also filled with insightful discussions with industry leaders, exploring the future of AI, investment, and the internet itself.

Interview: UTOPIA for Enterprise and Government

In an exclusive interview for the Hello, Self-Writing Internet livestream, Pierre Samaties, Chief Business Officer at the DFINITY Foundation, discussed the significance of UTOPIA for enterprise and government clients. He described UTOPIA as a private version of the Internet Computer, designed for organizations that require full control over their hardware for digital sovereignty.

This directly addresses a major trend of governments and enterprises seeking to build their own sovereign software, free from the data privacy concerns associated with traditional Big Tech.

“Caffeine will probably play an even bigger role on the enterprise and government market,” Pierre noted, explaining that the combination of Caffeine and UTOPIA provides a powerful tool for building and customizing fit-for-purpose software, bypassing lengthy and restrictive procurement processes.

Panel: From Seed to Singularity: Charting the Next Frontier in AI Investment

Speakers:

• Dr. Sue Xu, Managing Partner, Amino Capital
• Kumar Dandapani, Managing Partner, Cadenza
• Jing Xiong, AI Investment Director, YZi Labs
• Moderator: Jason Gao, DFINITY Foundation, Ecosystem & Strategic Partnerships

Three Key Takeaways:

• Investment is shifting from data-dependent AI to self-evolving AI: The panel noted a clear pivot in investment strategy away from AI that relies on human-labeled data and towards self-learning, agentic AI that can automate complex, domain-specific tasks.
• The evolution from “co-pilot” to “co-creator”: The consensus was that AI development tools are moving beyond simply assisting developers to becoming autonomous “co-creators.” This shift empowers a new wave of “solo unicorns” and will fundamentally change the startup landscape.
• Compute is the new capital: A recurring theme was that access to compute power is a critical bottleneck and a key investment thesis. Platforms like Caffeine that democratize or lower the cost of compute are seen as highly valuable and currently underrated.

Fireside Chat: AI Scaling as an Engine of Innovation and Growth

Panelists:

• William Stranzl, GTM, Anthropic
• Pierre Samaties, Chief Business Officer, DFINITY Foundation
• Moderator: Colm Woods, Managing Director, CW8 Communications

Three Key Takeaways:

• Enterprise AI is shifting from experimentation to implementation: Businesses are now actively using AI to unlock new revenue streams, driven by a desire to innovate and not get left behind. This is moving the conversation from simple chatbots to sophisticated agentic AI that can act as virtual collaborators.
• Sovereignty and customization are critical for adoption: For enterprise and government clients, the ability to build sovereign software on a controlled infrastructure like UTOPIA is a major driver. This allows for hyper-customized applications that fit specific business processes, rather than forcing processes to fit the software.
• A new, open app economy is emerging: The combination of powerful AI models and platforms like Caffeine is fundamentally changing how software is created and distributed, leading to a new trend of hyper-local, decentralized applications.

Fireside Chat: From Protocols to Possibilities: The Internet Yesterday, Today, and Tomorrow

• Speaker: Dominic Williams, Founder & Chief Scientist, DFINITY Foundation
• Moderator: Michael Casey, Chairman, Advanced AI Society

Three Key Takeaways:

• Self-writing is a new paradigm, not just better vibe coding: The discussion distinguished between vibe coding (an engineer using AI as a tool) and self-writing (where the entire app lifecycle is a conversation). This shift moves development from a technical task to a creative one, accessible to anyone.
• The future of the internet is fluid and democratized: The chat explored a future where creating an app is as common as a web search. This leads to a “fluid web,” where AI can generate user experiences on the fly by combining functionality from various apps, which act like microservices.
• A new paradigm requires a new, sovereign tech stack: The speakers emphasized that, like mobile before it, the self-writing era requires a purpose-built tech stack. The Internet Computer provides this, offering tamperproof, resilient, and sovereign apps that run without traditional cybersecurity overhead, a crucial safety guardrail for a world of AI-driven development.

Panel: Guardrails for Intelligence: Governance and Security in the Age of AI

Speakers:

• Jan Camenisch, CTO, DFINITY Foundation
• Alex Fowler, Chief Strategy Officer, Nexus
• Adnan Dakhwe, CISO, DelphinusCyber
• Moderator: Colm Woods, Managing Director, CW8 Communications

Three Key Takeaways:

• Security and privacy are often afterthoughts in the current AI boom: The panel highlighted a “crisis in leadership” where commercial pressures for speed and monetization frequently overshadow fundamental security and privacy concerns, leaving users to fend for themselves.
• Decentralized infrastructure offers a more secure foundation: The panel contrasted the “broken” traditional IT stack with secure-by-design protocols like the Internet Computer. Platforms like Caffeine can help users migrate to this more secure, sovereign environment where they have greater control over their data and applications.
• True AI governance requires end-to-end decentralization: To effectively govern powerful AI, control must be decentralized across the entire stack, from the training data and models to the execution environment. This removes single points of failure and is the only way to ensure long-term safety and alignment.

In the Future…

The Hello, Self-Writing Internet event was a landmark occasion. It brought together the tangible creativity of the hackathon winners, visionary keynotes, and insightful panel discussions to mark the arrival of a new paradigm.

The day showcased a future where the ability to create and deploy complete applications is no longer limited by technical skill, because as Dominic Williams explained, “in the future, natural language alone will unlock tech team superpowers.”

However, the results that emerged from the energy and creativity on display in San Francisco proved that the Self-Writing Internet is not really a distant concept. It is a reality today.

We would like to extend our thanks to all who attended in San Francisco and watched around the world, including the incredible hackathon participants and the insightful speakers who shared their expertise.

Keep Up to Date with Caffeine

You can find out about the latest developments related to Caffeine on X: https://x.com/caffeineai

Relive the Day

You can watch the full livestream of the event here:

Experience Caffeine for Yourself

Be among the first to build apps just by talking. Apply for early access to Caffeine. Invite codes are now being released on a rolling basis.

Apply here: https://caffeine.ai

Hello, Self-Writing Internet: Creations and Conversations was originally published in The Internet Computer Review on Medium, where people are continuing the conversation by highlighting and responding to this story.

By Andrea Cerulli

With the Niobium milestone, the Internet Computer Protocol takes a major leap forward in its mission to bring new levels of privacy to the blockchain. This milestone sees the implementation of the vetKeys feature, which is now live and available for all developers building on ICP. This upgrade opens the door to decentralized key management and onchain data privacy, marking the beginning of the Internet Computer’s privacy era.

Blockchains are transparent by default, which is ideal for integrity and availability. However, this is a limitation for many real-world applications that rely on confidentiality, like messaging, identity, healthcare, finance, governance, and gaming.

With vetKeys, developers can seamlessly integrate end-to-end encryption into their applications, ensuring sensitive data stays protected even on a public blockchain. By leveraging a protocol called verifiably encrypted threshold key derivation, vetKeys brings strong confidentiality and security even in a decentralized environment.

What Does the Niobium Milestone Introduce?

At the core of vetKeys is a newly integrated verifiably encrypted threshold key derivation protocol (vetKD). This enables smart contracts running on ICP, which are called canisters, to request strong cryptographic keys that are securely derived and managed by subnet nodes.

The derived keys are encrypted under a user-provided public key, ensuring that neither the subnet nodes nor the canister can access the derived keys. Furthermore, users can verify that the key was correctly generated according to the protocol, that it was correctly encrypted using their public key, and that the key was not tampered with.

Canisters can now access this new functionality through a newly introduced system API, giving developers direct access to decentralized key management from within their smart contracts, without relying on off-chain components. To make integration seamless, the Niobium milestone introduces user-facing and developer libraries that simplify incorporating vetKeys into both frontend workflows and backend canister logic, making it easier to build privacy-preserving dapps on the Internet Computer.

Additionally, the vetKD protocol enables canisters to access threshold decryption, where decryption operations are securely distributed across the nodes of a subnet. This distributed approach enhances security by eliminating single points of failure: no single party holds enough information to decrypt data on their own. Instead, decryption requires participation from a threshold of nodes, and the result is cryptographically verifiable.

Important Note: vetKeys guarantees strong confidentiality up to the point of decryption. Once the plaintext is handed off, for example, to a canister, it should no longer be assumed private. At that stage, it’s the developer’s responsibility to handle the data securely within the application logic. Furthermore, there are several applications that can benefit from canister-side decryption, where the decrypted content is intentionally revealed to all users at a specific time or event.

What Can You Build with vetKeys?

By introducing threshold key derivation to canisters, vetKeys opens the door to a broad set of privacy-preserving applications that need confidentiality, flexible access control, and stronger trust guarantees. Beyond privacy, vetKeys extends the chain-key capabilities of canisters with new advanced cryptographic functionalities, enabling more powerful and secure dapps.

Decentralized Key Management Services (DKMS)

Managing user keys, especially in multi-device and multi-user settings, is notoriously complex and error-prone. vetKeys enables the development of decentralized key management services that let users easily generate, retrieve, and share cryptographic keys across devices and with other users. By offloading key derivation to subnet nodes, without ever exposing the keys, vetKeys simplifies key distribution. Moreover, because threshold key derivation is deterministic, users can reliably retrieve their keys on demand, without having to explicitly store them anywhere.

Onchain Encrypted Storage

A core application of DKMS is the generation of encryption keys for securing data, whether stored in a canister, on another blockchain, or off-chain entirely. With vetKeys, these keys can be securely shared across devices and between users, enabling powerful privacy-preserving use cases. This includes private storage solutions, end-to-end encrypted messaging, password managers, and collaborative applications operating on confidential data.

Identity-Based Encryption (IBE)

vetKeys enable identity-based encryption, allowing data to be encrypted directly to an identity, such as a principal, an Internet Identity, an email address, or even an Ethereum address. This makes it possible to encrypt data for a specific user or account, even if that user has never previously interacted with the dapp. By authenticating with the dapp using their identity, the user can securely retrieve their decryption key and access the data.

Timelock Encryption

Another variant of advanced encryption enabled by vetKeys is timelock encryption, which allows a sender to encrypt to a specific timestamp, ensuring that the recipient can only decrypt after the specific time has passed. Canisters can enforce this time-based access control by requesting threshold decryption of a ciphertext only after a predetermined expiry time, keeping sensitive information sealed until the appropriate moment. This makes it possible to implement time-sensitive applications, like sealed-bid auctions, time-locked documents, dead-man switches, and delayed reveal NFTs.

Timelock encryption also serves as a key building block for protecting against Maximal Extractable Value (MEV): by keeping transaction details confidential until after block inclusion, it prevents adversaries from doing front-running and reordering of transactions.

Threshold BLS Signatures

Chain Fusion technology allows canisters to natively interact with other blockchains, such as Bitcoin or Ethereum, without relying on external bridges or trusted intermediaries. This is made possible through threshold signature schemes, which enable canisters to instruct subnet nodes to collectively compute ECDSA, Schnorr, and EdDSA signatures. vetKeys extends this capability by introducing a new threshold signature scheme to canisters: threshold BLS signatures. BLS signatures are particularly well-suited for multichain applications due to their compact size and efficient aggregation properties. By supporting threshold BLS, vetKeys further enhances ICP’s interoperability, empowering canisters to participate in more advanced multichain dapps and protocols.

Verifiable Randomness

vetKeys can also function as a Verifiable Random Function (VRF). This means that canisters can generate randomness that is not only unpredictable and tamper-proof but also publicly verifiable. Trusted randomness is a critical building block for a wide range of decentralized applications, including trustless online lotteries, fair casino games, and GameFi experiences where outcomes must be demonstrably fair. It also plays a key role in NFT ecosystems — for example, in assigning randomized traits or rarities during minting, or enabling dynamic, chance-based in-game interactions.

Start Building with vetKeys

Developers can start using vetKeys today using the official documentation:

• vetKeys developer docs
• Libraries and examples
• Reach out and interact with the community

The Internet Computer’s Privacy Era: vetKeys Unlocked was originally published in The Internet Computer Review on Medium, where people are continuing the conversation by highlighting and responding to this story.