Apple's annual developer conference (Apple WWDC) kicked off this morning on June 10, led by CEO Tim Cook and software director Craig Federighi, who unveiled Apple's latest system updates and technological innovations.
The press conference lasted more than an hour and had many highlights, such as a series of naming and system changes such as iOS 26, the dynamic sensing interface of "Liquid Glass", and the most anticipated "Apple Intelligence" device-side AI, which is not only deeply integrated into various devices, but also focuses on device-side computing and privacy protection.
Table of Contents
ToggleDevice-side AI "Apple Intelligence" officially debuts
Federighi was the first to introduce the company's self-developed "Apple Intelligence" device-side AI. Its biggest selling points are "local computing" and "privacy protection," which allows it to work without uploading data to the cloud.
All developers can use the new Foundation Models framework and use the large language model on the device to create apps, such as Kahoot, which can automatically generate test questions based on the user's notes. It also emphasizes that there is no need to pay extra cloud API fees to enjoy the functions.
Liquid Glass interface unveiled, capable of dynamically sensing the user's screen
Alan Dye, vice president of design, then introduced the new design interface, also known as " Liquid Glass ". He said that it will dynamically sense the content the user is currently using to automatically adjust the light and shadow and appearance to make the picture more lifelike.
Day said that new settings have also been added to the camera and photo apps, especially the novel function of converting 2D photos into 3D photos, which can make users feel more immersive when editing memory videos.
iOS 26 system unified naming, focusing on AI combined with daily applications
This time, Apple also unified the naming rules of its own iOS system, using "26" as the system version number to echo the year, in line with macOS and watchOS, in an effort to demonstrate a more consistent cross-platform integration strategy.
iOS 26 focuses on integrating AI with daily applications. For example, the Phone app will automatically generate summaries for voice messages and identify unfamiliar calls. The Messages app has added Apple Cash instant transfers, and AI instant translation has been integrated into Messages, FaceTime, and voice calls, allowing users to communicate more smoothly.
Other self-built apps also use AI, such as the wallet app, which supports digital IDs and integrates flight, ticket and luggage tracking. CarPlay adds widgets, quick message replies and vehicle control to help users handle daily life with AI.
Cross-platform evolution, comprehensive upgrade from watches, computers to VR
Apple Watch's watchOS 26 uses a new "liquid glass" interface to enhance the screen quality. The new feature Workout Buddy can provide encouragement and personalized suggestions based on the user's exercise records through Apple Intelligence. Users can quickly adjust the volume by simply shaking their wrists, and push notifications will automatically adjust according to the ambient sound.
The computer system macOS Tahoe 26 features a simple and transparent interface, and folders can be changed in color and emoji can be added to make the operation more personalized. After the Spotlight upgrade, it can also be sorted according to user habits and current activity status. The phone app has also officially landed on the Mac, which can be synchronized with the iPhone to automate the workflow.
VisionOS 26 enhances the user experience of VR devices, supports multi-person sharing of virtual space, game melee, and simultaneous video viewing, and also supports Sony PlayStation VR2 controller and Adobe space video editing. It can also create user-exclusive virtual characters for remote interaction.
WWDC did not spark market interest, with the stock price falling nearly 2% on the opening day
Although WWDC introduced many novel features that combined its own platform with AI, it did not drive the stock price up. When it opened, the stock price fell by 1.21% that day, closing at $201.75 per share.
Risk Warning
Cryptocurrency investment carries a high degree of risk. Its price may fluctuate drastically and you may lose all your capital. Please assess the risk carefully.
In an era where high TPS and low fees have become standard, simply building a new public chain is no longer enough to impress users and builders. Today, developers in the crypto world are turning to a new survival rule: " First build a super application, then turn it into a platform ." This "App-First" trend is quietly rewriting the development logic of blockchain.
Table of Contents
ToggleA chain without apps is just a deserted highway
Four Pillars researcher Ponyo recently published an article , using Hyperliquid and Abstract as an example, trying to prove that " a true console must have games that players love. "
He pointed out that in the past decade, many blockchain projects have focused on building "general infrastructure" such as L1/L2, hoping that developers and users will naturally flock in. But as former Microsoft CEO Steve Ballmer said: " If there is no application on the platform, it will eventually fall into the quagmire. "
Just as Office made Windows possible and Outlook supported Exchange Server, a chain would find it difficult to initiate a network effect if it did not have its own signature application.
Today, every new chain claims to be faster and have lower fees, but this kind of technical propaganda has long numb the market:
A chain without a killer app is like a game console without games. No one will buy it even if it has powerful performance.
Choose one of two routes: develop products first, or build a platform first?
Senior researcher Hasu divides the current development strategies of blockchain projects into two paths :
Route 1: Create “10x” innovations in technology, such as breakthrough architectures or consensus mechanisms, but such breakthroughs are becoming increasingly rare.
Route 2: First create a super successful application, then turn it into a blockchain platform, and then improve the ecosystem.
Hasu believes that the second path not only has a lower threshold, but can also effectively break the cold start dilemma:
Applications that have users and demand from the beginning can reversely drive the adoption and growth of the chain, and it is easier to focus on product-market fit (PMF) rather than just protocol-market fit.
Network effect enabler: Building your own application is the key to success
A popular application not only attracts users, but also ignites the blockchain's "network effect flywheel". Users will come to find applications, and developers will follow users, not the other way around. More importantly, building your own application allows the development team to focus on solving a single problem and avoid falling into the wasteful cycle of "building infrastructure and waiting for people to use it".
In addition, the coexistence of applications and chains can also bring about a " co-location effect" :
When your signature app attracts a large number of users and funds, other similar applications are more willing to be deployed on the same chain, forming an overall ecological expansion. For example, a strong DEX can become a reason for other lending or yield protocols to rush in.
When the development team takes the lead in creating first-hand applications, it will also make the community and other developers more confident to join the ecosystem, rather than relying solely on funding or rewards to attract external contributors.
Best Practices: Hyperliquid and Abstract Lead the App-First Model
In this App-First trend, Hyperliquid is one of the most representative success stories. This high-performance perpetual contract exchange (Perp DEX) uses self-built L1 to deeply integrate transaction matching with its own chain.
Since its launch, the platform's daily trading volume has reached 11.2% of Binance, 24.1% of Bybit, and 22.1% of OKX. Currently, DeFi TVL is at $1.54 billion, ranking ninth among all L1s.
On the other hand, Abstract , an Ethereum L2 extension solution that focuses on Web3 game portals, started with the "Portal App", integrating social account login, intuitive gas payment and a unified wallet system. So far, it has launched 33 mini games, attracted more than 2 million users, and completed 76 million transactions. Its strategy is clear: "Let users have fun first, and then slowly turn the game into a platform."
These projects not only prove that the App-First strategy is feasible, but also overturn the previous assumption of the development path: "You don't have to have a chain before you can have a product."
Web3’s Microsoft and Apple will start with products
The rise of Microsoft and Apple was based on the explosion of applications after the maturity of basic hardware. The same story will be repeated in the crypto world. Today's on-chain infrastructure is quite mature, with L2, DA layer, and ZK technology. The key to the future is no longer to create a better chain, but who can be the first to deliver a " truly easy-to-use and fun killer product ."
The real next-generation standard is no longer just infra or just application, but a powerful combination of the two: " First use the App to attract users, then use the chain to circle the ecosystem ."
Risk Warning
Cryptocurrency investment carries a high degree of risk. Its price may fluctuate drastically and you may lose all your capital. Please assess the risk carefully.
Recently, the Ethereum Foundation forwarded a post about Virtuals Protocol, the AI agent startup platform on the Base chain, hinting that the two sides will start cooperation. It is understood that the first cooperation project UndercoverIRIS is a collaboration with Nethermind led by Tomasz Stańczak, executive director of the Ethereum Foundation. UndercoverIRIS is an AI-driven Web3 smart contract audit agent. The technology provided by the Ethereum core team may break the dilemma that AI agents in the crypto are all hype and have no practical use scenarios. If the token reaches 100 million US dollars, the estimated profit will exceed 500 times.
Table of Contents
ToggleNethermind, the company behind IRIS, is led by Ethereum’s executive director
It is understood that the first cooperation project between Ethereum and Virtuals Protocol will be IRIS ( UndercoverIRIS ), which is backed by Nethermind, a community of developers and researchers on Ethereum. IRIS is an AI-driven Web3 smart contract audit agent that scans smart contracts, identifies vulnerabilities, and provides feasible remediation measures.
Nethermind behind IRIS may be unfamiliar to non-developers, but the founder of the team, Tomasz Stańczak, is a core member of Ethereum. In March, the Ethereum Foundation announced that Taiwanese blockchain expert Hsiao-Wei Wang and Tomasz Stańczak will serve as co-executive directors.
If IRIS reaches $100 million, the estimated profit will exceed 500 times
It is understood that the IRIS token is launched by Virtuals' Virgen points system, and the subscription process can be referred to in previous reports. The token will be launched on the Ethereum mainnet, with 50% of the tokens given to the Ethereum community, and no team token share. Pre-sale participants will also subscribe to $IRIS at a valuation of approximately $200,000. If the market value of $IRIS reaches $100 million, there will be a 500-fold return.
In addition, according to @ethermage’s post, the Virtuals team is trying to support AI agents on Ethereum L1. He said that although it is the same EVM, the community and capital styles of Layer 1 and Layer 2 are completely different.
Risk Warning
Cryptocurrency investment carries a high degree of risk. Its price may fluctuate drastically and you may lose all your capital. Please assess the risk carefully.
The InfoFi track has entered the Warring States period, and the competition between Kaito, Cookie, and BubbleMap has become fierce. Mirra AI has received relatively little attention. Perhaps this project that combines InfoFi with AI model training also has room for ambush. Mirra AI is a project that uses social platforms (such as X) to train AI. Quality content creators and scouts looking for high-quality content can get points from it. Mirra AI's scoring rules emphasize real content creation, rather than turning white papers into other words or AI-generated content.
Table of Contents
ToggleMirra AI uses social platforms to train AI agents
Mirra AI is a project that uses social platforms (such as X) to train AI. In simple terms, it can be divided into two roles: Creator (content creator) and Scout (scout). If you find excellent creative content on Twitter, you can tag @MirraTerminal in the comment area. At this time, Mirra will verify the content, which is equivalent to submitting AI training data, contributing data, and making AI smarter. After completing the verification, both the creator and the scout will receive points. Please note that you need to link your X account to receive the points.
As a content creator, the content you create becomes part of the training AI. When your insights are verified, you can also earn points. When Mirra points are popular enough, the points ranking also represents who is the recognized thought leader. For the role of a scout, you decide what content the artificial intelligence learns and you can also earn points from it. Mirra AI builds decentralized AI in this way. The first training case is Mirra Agent , and it seems that there is a possibility of training other agents.
Mirra likes content that provides food for thought and is written by real people
Mirra founder Lester shared the process of earning points on Twitter. Generally speaking, content detection is divided into two times. The first time is when a scout tags Mirra in a post, the scout's content and account will be scanned. If it is detected (suspected of being a robot, ChatGPT content, spam, blacklisted keywords), the scout's request will be ignored directly. When the first detection is passed, it will display: "Content has been detected. Being evaluated."
Mirra will give a score to the content during the second detection. Both parties can get points in the system, but the scores of both parties are different. Mirra may also give a message that "your content has not passed the detection".
Lester also explained the scoring criteria. Generally speaking, to get a score you need to:
- Related topics.
- Useful information.
- Operability.
- Based on facts.
- Share your personal opinions.
- Critical thinking.
- Funny, sharp humor (Mirra likes black humor and witty expressions).
- Original content.
Also try to avoid:
- Similar to ChatGPT writing.
- No opinion.
- Lack of critical thinking.
- Low quality lip service (most kaito yappers just rewrite the project's whitepaper).
- General comments.
- Overuse of emojis.
- Hashtag.
- Mentioning too many accounts in a post.
- Overly technical content.
In short, Mirra likes content that makes her think while reading and is written by real people.
However, on the on-chain reputation platform ethos, some users also expressed negative views on Mirra AI.
Risk Warning
Cryptocurrency investment carries a high degree of risk. Its price may fluctuate drastically and you may lose all your capital. Please assess the risk carefully.
