7*24

Hot topic

Market

Activities

According to monitoring by Dongcha Beating, Trajectory, a continuous learning platform founded by senior researchers from Google DeepMind, Apple, OpenAI, and Meta Superintelligence Labs, has officially launched. The company simultaneously released a manifesto titled 'Continual Learning: End of Frozen Software,' which clearly advocates that AI software should not remain in a static frozen state after deployment, but should convert every user correction, retry, and overwrite action into direct reward signals for online continuous updates of the model. The core mechanism of Trajectory lies in stitching together the execution traces of agents with the telemetry signals from end users. Execution traces accurately record the tool calls and intermediate reasoning steps of the model, while telemetry signals capture whether users ultimately accept, modify, or revert the results. The development team emphasizes that dynamic intervention data from users on outputs has long been discarded in traditional development, yet this is precisely the golden training signal for assessing whether an agent has truly mastered complex tasks. On the technical implementation front, Trajectory connects to mainstream observability platforms (such as LangSmith) via a standardized SDK. The toolkit can automatically import agent execution traces and convert conversation history, tool calls, reward feedback, custom metrics, and error messages into the standard Trajectory format for subsequent fine-tuning and evaluation. The accompanying web platform is still in development and will eventually provide visual retrieval and refined management of trace data. In terms of commercialization and ecosystem building, Trajectory has officially completed a $15 million seed round of financing, achieving a post-investment valuation of $115 million. This round was led by Conviction, with participation from Bessemer Venture Partners, Radical VC, and BoxGroup, and attracted investments from notable individuals such as Google Chief Scientist Jeff Dean and Stanford Professor Fei-Fei Li. The company has already established partnerships with leading AI-native agent service providers such as Decagon, Clay, and Harvey. However, due to the current system's reliance on periodic offline model fine-tuning, there remains a significant engineering gap before achieving true per-interaction real-time continuous learning.

[Trajectory, Founded by Former DeepMind Team, Achieves $115 Million Valuation to Address AI Feedback Loop]According to monitoring by Dongcha Beating, Trajectory, a continuous learning platform founded by senior researchers from Google DeepMind, Apple, OpenAI, and Meta Superintelligence Labs, has officially launched. The company simultaneously released a manifesto titled 'Continual Learning: End of Frozen Software,' which clearly advocates that AI software should not remain in a static frozen state after deployment, but should convert every user correction, retry, and overwrite action into direct reward signals for online continuous updates of the model. The core mechanism of Trajectory lies in stitching together the execution traces of agents with the telemetry signals from end users. Execution traces accurately record the tool calls and intermediate reasoning steps of the model, while telemetry signals capture whether users ultimately accept, modify, or revert the results. The development team emphasizes that dynamic intervention data from users on outputs has long been discarded in traditional development, yet this is precisely the golden training signal for assessing whether an agent has truly mastered complex tasks. On the technical implementation front, Trajectory connects to mainstream observability platforms (such as LangSmith) via a standardized SDK. The toolkit can automatically import agent execution traces and convert conversation history, tool calls, reward feedback, custom metrics, and error messages into the standard Trajectory format for subsequent fine-tuning and evaluation. The accompanying web platform is still in development and will eventually provide visual retrieval and refined management of trace data. In terms of commercialization and ecosystem building, Trajectory has officially completed a $15 million seed round of financing, achieving a post-investment valuation of $115 million. This round was led by Conviction, with participation from Bessemer Venture Partners, Radical VC, and BoxGroup, and attracted investments from notable individuals such as Google Chief Scientist Jeff Dean and Stanford Professor Fei-Fei Li. The company has already established partnerships with leading AI-native agent service providers such as Decagon, Clay, and Harvey. However, due to the current system's reliance on periodic offline model fine-tuning, there remains a significant engineering gap before achieving true per-interaction real-time continuous learning.

Trajectory, Founded by Former DeepMind Team, Achieves $115 Million Valuation to Address AI Feedback Loop

HTX data monitoring shows that XRP has dropped below 1.3 USDT, currently reported at 1.3 USDT, with a 24H decline of 2.33%.

【XRP drops below 1.3 USDT】HTX data monitoring shows that XRP has dropped below 1.3 USDT, currently reported at 1.3 USDT, with a 24H decline of 2.33%.

XRP drops below 1.3 USDT

On May 28, Josh Stevens, Vice President of Engineering at the prediction market platform Polymarket, responded to a report by The Information stating that "Polymarket plans to require users to complete KYC due to compliance pressures," saying, "This report is not true. Polymarket is currently testing a new beta product that only requires some trial users to complete KYC verification during the testing period; the existing Polymarket website has not introduced any new KYC requirements. He further stated that once the testing of this product concludes, users will no longer need KYC to use it.

[Polymarket Denies Implementing KYC, Only Requires Verification for Test Users]On May 28, Josh Stevens, Vice President of Engineering at the prediction market platform Polymarket, responded to a report by The Information stating that "Polymarket plans to require users to complete KYC due to compliance pressures," saying, "This report is not true. Polymarket is currently testing a new beta product that only requires some trial users to complete KYC verification during the testing period; the existing Polymarket website has not introduced any new KYC requirements. He further stated that once the testing of this product concludes, users will no longer need KYC to use it.

Polymarket Denies Implementing KYC, Only Requires Verification for Test Users

On May 28, Iranian Tasnim News Agency reported, citing a military source, that US forces opened fire in the vicinity of the port of Hormuz near the Strait of Hormuz, following a standoff between the Iranian Revolutionary Guards and a US tanker attempting to pass through the strait.

[Iranian Media Confirms US Military Fired Shots Amidst Standoff with Revolutionary Guards]On May 28, Iranian Tasnim News Agency reported, citing a military source, that US forces opened fire in the vicinity of the port of Hormuz near the Strait of Hormuz, following a standoff between the Iranian Revolutionary Guards and a US tanker attempting to pass through the strait.

Iranian Media Confirms US Military Fired Shots Amidst Standoff with Revolutionary Guards

On May 28, the central parity rate of the RMB against the USD increased by 51 points from the previous day to 6.8240, marking the highest appreciation since February 15, 2023.

[RMB Central Parity Against USD Rises 51 Points to 6.8240]On May 28, the central parity rate of the RMB against the USD increased by 51 points from the previous day to 6.8240, marking the highest appreciation since February 15, 2023.

RMB Central Parity Against USD Rises 51 Points to 6.8240

According to monitoring by Dongcha Beating, AI programming lab Cognition has announced the completion of a new funding round exceeding $1 billion, with a post-funding valuation soaring to $26 billion. This round was co-led by Lux Capital, General Catalyst, and 8VC, with new investors Ribbit Capital, Atreides, and Layer Global joining in. The company's valuation has jumped 1.5 times in just 8 months, confirming the capital's strong confidence in the independent agent route despite the competitive pressure from major underlying model giants. Rapid commercialization and enterprise penetration provide strong factual support for Cognition. Since early 2026, the enterprise usage of AI programmer Devin has increased more than tenfold, with an annual recurring revenue (ARR) rate climbing to $492 million. Currently, Cognition's enterprise clients include top global institutions such as Mercedes-Benz, Goldman Sachs, Citibank, and Dell, and are further expanding to government and enterprise clients like the U.S. Army and Navy. In terms of technical depth, Cognition is forming a dual-driven approach of agent autonomous iteration and self-developed large models. Currently, 89% of the code merge requests (PR) in the Cognition team are independently written and delivered by Devin. At the same time, the company is accelerating its model's autonomous training, with its self-developed model SWE-1.6 becoming the most active model in the Windsurf editor, achieving an inference speed of up to 950 tokens per second. Cognition emphasizes that as an independent agent laboratory, its core advantage lies in the ability to flexibly schedule the optimal underlying model for different tasks, ensuring excellent cost-effectiveness and multi-cloud neutrality amidst the native competition among underlying model vendors.

[Valued at $26 Billion, AI Programmer Devin Handles 89% of Company’s PR with Nearly $500 Million Annual Revenue]According to monitoring by Dongcha Beating, AI programming lab Cognition has announced the completion of a new funding round exceeding $1 billion, with a post-funding valuation soaring to $26 billion. This round was co-led by Lux Capital, General Catalyst, and 8VC, with new investors Ribbit Capital, Atreides, and Layer Global joining in. The company's valuation has jumped 1.5 times in just 8 months, confirming the capital's strong confidence in the independent agent route despite the competitive pressure from major underlying model giants. Rapid commercialization and enterprise penetration provide strong factual support for Cognition. Since early 2026, the enterprise usage of AI programmer Devin has increased more than tenfold, with an annual recurring revenue (ARR) rate climbing to $492 million. Currently, Cognition's enterprise clients include top global institutions such as Mercedes-Benz, Goldman Sachs, Citibank, and Dell, and are further expanding to government and enterprise clients like the U.S. Army and Navy. In terms of technical depth, Cognition is forming a dual-driven approach of agent autonomous iteration and self-developed large models. Currently, 89% of the code merge requests (PR) in the Cognition team are independently written and delivered by Devin. At the same time, the company is accelerating its model's autonomous training, with its self-developed model SWE-1.6 becoming the most active model in the Windsurf editor, achieving an inference speed of up to 950 tokens per second. Cognition emphasizes that as an independent agent laboratory, its core advantage lies in the ability to flexibly schedule the optimal underlying model for different tasks, ensuring excellent cost-effectiveness and multi-cloud neutrality amidst the native competition among underlying model vendors.

Valued at $26 Billion, AI Programmer Devin Handles 89% of Company’s PR with Nearly $500 Million Annual Revenue

On May 28, traditional Bitcoin mining companies continued to attract market interest after transitioning into artificial intelligence, large-scale computing, and data center infrastructure. A series of related stocks recorded double-digit gains, with IREN leading the sector on Wednesday, rising over 13% and approaching its historical high; Cipher Digital and Hut 8 even reached all-time highs. This trend continues the upward momentum seen over the past few months: companies that once focused on Bitcoin mining are shutting down some mining operations and reallocating resources and capital towards AI power and computing infrastructure. Many investors now believe these new ventures are more attractive than traditional mining operations.

[Bitcoin Miners Shift to AI Business, Cipher and Hut 8 Share Prices Hit New Highs]On May 28, traditional Bitcoin mining companies continued to attract market interest after transitioning into artificial intelligence, large-scale computing, and data center infrastructure. A series of related stocks recorded double-digit gains, with IREN leading the sector on Wednesday, rising over 13% and approaching its historical high; Cipher Digital and Hut 8 even reached all-time highs. This trend continues the upward momentum seen over the past few months: companies that once focused on Bitcoin mining are shutting down some mining operations and reallocating resources and capital towards AI power and computing infrastructure. Many investors now believe these new ventures are more attractive than traditional mining operations.

Bitcoin Miners Shift to AI Business, Cipher and Hut 8 Share Prices Hit New Highs

HTX data monitoring shows that DOGE has fallen below 0.1 USDT, currently reported at 0.1 USDT, with a 24H decline of 1.70%.

【DOGE falls below 0.1 USDT】HTX data monitoring shows that DOGE has fallen below 0.1 USDT, currently reported at 0.1 USDT, with a 24H decline of 1.70%.

DOGE falls below 0.1 USDT

On May 28, Trump posted on social media stating, 'Former SEC Chairman Gary Gensler and the so-called 'anti-crypto brigade' nearly destroyed the U.S. cryptocurrency industry, driving Bitcoin, crypto perpetual contracts, and innovation overseas, but I saved it all. Today, the U.S. has become the 'crypto capital of the world,' and builders and entrepreneurs are returning to their America. Under my leadership, we will create a 'future-oriented' digital asset market structure framework that will prevent cryptocurrency opponents from easily overturning it. The new frontier of finance is being established in America, and I will never let the crypto industry down.'

[Trump: I Saved the U.S. Cryptocurrency Industry and Will Establish a 'Future-Oriented' Market Structure Framework]On May 28, Trump posted on social media stating, 'Former SEC Chairman Gary Gensler and the so-called 'anti-crypto brigade' nearly destroyed the U.S. cryptocurrency industry, driving Bitcoin, crypto perpetual contracts, and innovation overseas, but I saved it all. Today, the U.S. has become the 'crypto capital of the world,' and builders and entrepreneurs are returning to their America. Under my leadership, we will create a 'future-oriented' digital asset market structure framework that will prevent cryptocurrency opponents from easily overturning it. The new frontier of finance is being established in America, and I will never let the crypto industry down.'

Trump: I Saved the U.S. Cryptocurrency Industry and Will Establish a 'Future-Oriented' Market Structure Framework

Президент Соединенных Штатов Америки Дональд Трамп заявил, что рынок блокчейн-прогнозов должна контролировать Комиссия по торговле товарными фьючерсами (CFTC). Ведомство ранее заподозрили в слишком высокой лояльности к криптобизнесу семьи Трампа.

$50 Million Funding Ignites Airdrop Anticipation, Variational Becomes New Focus of Perp DEX

Musk Posted a Recruitment Ad for SpaceX, and After Reading the Comments Section, I Understood

XRP Has Overtaken Solana And Ethereum In This Major Metric

Cloudflare CEO: How I Decided Which Employees to Replace with AI?

Secures Over $60 Million in Funding from Dragonfly, Sequoia, and Others. An In-Depth Look at the On-Chain Derivatives Protocol Variational | CryptoSeed

XRP Ledger Hits No. 4 In RWA Rankings In Less Than A Year

General Tensor and Talisman Joining Forces To Build Financial Rails for Decentralized AI

Why Did Zhipu Surge Nearly 30% in a Single Day?

Warsh's First Day in Office, Markets Deliver a 'Wake-up Call': Rate Hike Expected This Year

Microsoft to Discontinue Using Claude: Too Expensive, or a Calculated Move?

Article

Economist: Warsh's Nomination is a 'Malicious Suppression' of Gold and Silver, but Hawks Cannot Return

Micron Technology Stock Price Boosted by Multiple Investment Banks, Latest Average Target Price Raised to $1,008

Polymarket Sees Temporary Decline in Trading Activity After v2 Upgrade

Isenção de responsabilidade：O conteúdo acima não representa as posições da HTX.，A HTX não faz quaisquer recomendações de negociação.。

Todos os comentários0Mais recentePopular

Tópicos em destaque