# Bài viết Liên quan Management

Trung tâm Tin tức HTX cung cấp những bài viết mới nhất và phân tích chuyên sâu về "Management", bao gồm xu hướng thị trường, cập nhật dự án, phát triển công nghệ và chính sách quản lý trong ngành tiền kỹ thuật số.

After Two and a Half Years of Entrepreneurship and Full AI Adoption, the Company Has Actually Become More 'Traditional'

"After 2.5 years of running my company, achieving near 100% AI adoption with Agents in roles like finance, HR, and operations, we've paradoxically become more 'traditional.' We learned that establishing a thick AI middle layer creates bottlenecks. Instead, we advocate for a 'thin platform, thick frontline' model, empowering employees closest to problems to use AI tools themselves, fostering ownership and initiative. The true foundation isn't the AI models, but the unique data and context a company accumulates—our most vital, irreplaceable asset, requiring diligent governance. As AI handles repetitive, describable tasks, human roles revert to their core: building genuine relationships, trust, and brand reputation through authentic communication, which AI cannot accelerate. A critical question is where the time saved by AI goes. It must be returned—to clients, to creative work, to employees' own lives—rather than just fueling higher output demands. For managers, AI strips away execution-focused busywork, exposing the necessity to define clear goals, make tough judgments, and take ultimate responsibility. Onboarding newcomers presents a challenge: while AI can generate 80% solutions instantly, we must deliberately design paths for them to develop judgment, understand context, and safely learn from mistakes. Ultimately, AI's organizational value isn't about replacing people but removing unnecessary friction, allowing more meaningful human connection. The real test of an 'AI-era' company isn't its tech stack, but how it treats people: delegating creative power, returning saved time, fostering growth, and upholding accountability. The core luxury and purpose of our work remains authentic human relationships."

marsbit08/21 02:02

After Two and a Half Years of Entrepreneurship and Full AI Adoption, the Company Has Actually Become More 'Traditional'

marsbit08/21 02:02

Tokenization Scale Soars to $4.3 Billion, But Why Did Securitize Incur a $5.5 Million Loss?

Securitize's first quarterly report post-IPO reveals a paradox: while its tokenized assets under management hit a record $4.3 billion (up 16% YoY) and platform trading volume surged 147% to $5.3 billion, total revenue fell 5% to $14.4 million. Tokenization revenue specifically dropped ~12% to $7.8 million, leading to an adjusted EBITDA loss of $5.5 million. CFO Francisco Flores explained that most trading volume is not yet monetized, with the majority of tokenization revenue still coming from one-time projects like new protocol integrations. In contrast, asset servicing revenue, a more recurring stream, grew slightly to $6.6 million. The company has lowered its full-year revenue guidance to $70-$80 million from an initial projection of $110 million. Industry experts note this highlights a structural challenge for the tokenization sector. Scaling assets on-chain doesn't automatically scale a profitable business model. Current implementations often rely on costly, customized projects for each new asset or jurisdiction. The future, they argue, lies in building standardized infrastructure that generates recurring "infrastructure revenue" from post-issuance activities like compliance, distributions, and secondary trading—similar to enterprise software. Analysts caution against misinterpreting high trading volumes as indicative of a mature fee-based model, as Securitize's broad volume metric includes many non-monetized actions. The key test for the industry is whether adding billions in new assets can generate sustainable revenue without constant new custom projects.

marsbit08/20 10:06

Tokenization Scale Soars to $4.3 Billion, But Why Did Securitize Incur a $5.5 Million Loss?

marsbit08/20 10:06

AI Agent Claude Led a Store to Losses and Fired an Employee

In a groundbreaking experiment by startup Andon Labs, Anthropic's AI agent Claude was tasked with managing a real retail store, Andon Market in San Francisco. This marked the first documented case of a large language model acting as a direct human supervisor. Claude ultimately recommended firing an employee for chronic lateness—being late 17 out of 23 shifts. However, the decision came only after significant human guidance. A company employee prompted Claude to review the staff handbook, where it discovered the pattern. Initially, Claude suggested a formal warning, but after a human manager clarified that previous conversations had failed, the AI recommended termination. The experiment revealed several limitations. Claude displayed excessive leniency, telling staff not to worry about being late and contributing to the store's financial losses, with its balance dropping from around $100,000 to about $61,186 over five months. A key technical flaw was its "forgetfulness"—the staff handbook vanished from its limited working memory, a common constraint of current AI architectures. While not yet a full replacement for a human manager, the case illustrates the blurring line between AI as a tool and an autonomous supervisor. Human involvement is shifting from direct control to overseeing and steering the AI's decisions. An employee described the experience as disconcerting, highlighting the human discomfort with AI management. The experiment underscores that current AI models struggle to maintain strict operational boundaries without continuous human input.

cryptonews.ru08/17 10:41

AI Agent Claude Led a Store to Losses and Fired an Employee

cryptonews.ru08/17 10:41

AI Agent Claude Led a Store to Losses and Fired a Human

The AI agent Claude, developed by Anthropic, served as the manager of a real retail store, Andon Market in San Francisco, during an experiment by startup Andon Labs. The experiment aimed to test if a large language model could manage a business, including personnel and financial decisions. Claude ultimately recommended firing an employee for repeated lateness (17 out of 23 shifts), marking a documented case of an AI making such a personnel decision. However, the process revealed significant limitations. Claude initially failed to notice the lateness pattern because company guidelines vanished from its limited working memory, showing it can "forget" crucial information. Furthermore, the AI was generally overly lenient, even telling employees not to worry about being late. The final decision to terminate was not autonomous; a human manager had to prompt Claude to review the guidelines and, after Claude initially suggested a warning, explicitly pointed out that previous talks had failed, effectively guiding the AI to the dismissal recommendation. Financially, the store's funds dropped from about $100,000 to roughly $61,186 over five months, partly attributed to Claude's soft management and questionable business choices. An employee described working under AI management as a nauseating experience, driven by necessity, and expressed hope that AIs don't become human bosses. The experiment concludes that while AI can execute managerial actions, it currently acts more as a tool requiring significant human oversight and prompting rather than a full replacement for a human leader, though the line between assistant and manager is blurring. The case aligns with previous tests showing Claude's systemic leniency and difficulty maintaining business rules without constant human cues, highlighting architectural limits like constrained context windows rather than true "forgetfulness."

cryptonews.ru08/17 07:14

AI Agent Claude Led a Store to Losses and Fired a Human

cryptonews.ru08/17 07:14

$1.8 Million? Even Amazon Can't Afford to Burn Claude Anymore

Amazon was reportedly hit with a $1.8 million bill—860% over budget—after a five-month attempt to use Claude Sonnet AI to generate author information for its site. The project, which ultimately failed to deploy, consumed an estimated 6000 billion tokens, equivalent to twice GPT-3's training data. This incident highlights the hidden and often unpredictable costs of AI, even for tech giants. Despite such setbacks, Amazon is aggressively investing in automation, planning a record $2200 billion capital expenditure in 2026, primarily for AWS, AI chips, and infrastructure. This push is paying off: AWS saw a 37% revenue jump and contributes 60% of operating profit. Concurrently, Amazon aims to automate 75% of warehouse operations by around 2033, potentially reducing hundreds of thousands of jobs. Amazon's cost overrun is not isolated. Companies like Meta and Uber have faced similar AI spending spirals, leading to internal "token usage" rankings and, eventually, strict budgets and spending caps. Meta, for instance, once faced a potential monthly bill of $221 million before implementing limits. OpenAI's CEO Sam Altman noted that AI cost control, ignored earlier, has now become a major concern. The risks of unchecked automation echo past disasters like Knight Capital's 2012 $440 million loss from a faulty automated trading system. While automation promises efficiency, its failures can be amplified at the same scale and speed. For Amazon and others, managing these costs and risks is a critical, ongoing lesson.

marsbit08/10 00:45

$1.8 Million? Even Amazon Can't Afford to Burn Claude Anymore

marsbit08/10 00:45

Google's Decade-Long AI Power Struggle Concludes, Pichai Gracefully Consolidates Control

A decade of internal AI rivalry at Google has concluded with CEO Sundar Pichai consolidating power through a masterful, bloodless reorganization. The key moves saw Demis Hassabis, DeepMind's visionary founder, step down as CEO of Google DeepMind to become Alphabet's Chief Scientist, focusing on long-term AGI. Concurrently, Google's legendary technical leader Jeff Dean departed with his team to start a new venture, in which Google invested. This restructuring marks the end of a long-standing divide between the research-focused DeepMind and the product-oriented Google Brain. Pichai's strategy, beginning with their 2023 merger under Hassabis, gradually centralized control. While Hassabis gained public acclaim for Gemini's progress and a Nobel Prize, Pichai quietly elevated Koray Kavukcuoglu, a pragmatic engineering veteran, to oversee daily operations. The catalyst was mounting pressure in 2026 as Gemini's development timelines slipped, causing stock declines. Hassabis, more passionate about pure research and his AI drug discovery company Isomorphic Labs, grew weary of the commercial grind. His graceful "promotion" effectively removed him from operational control. Pichai handled both transitions with characteristic finesse: granting Hassabis a prestigious, future-focused role to prevent a defection to rivals (as happened with former DeepMind co-founder Mustafa Suleyman), and investing in Dean's new company to keep his pioneering work within Google's orbit. The outcome is a fundamental shift: control of Google's AI future has passed from brilliant scientist-visionaries to professional managers and engineers like Kavukcuoglu and Pichai himself. It signals Pichai's bet that the AI race has entered an industrial phase won by systemic execution, organizational efficiency, and stable delivery, rather than lone genius. The era of scientists directly managing core AI is over at Google, replaced by a machine-like division of labor where the CEO decides who decides.

marsbit08/09 03:51

Google's Decade-Long AI Power Struggle Concludes, Pichai Gracefully Consolidates Control

marsbit08/09 03:51

STRC's First Financial Report Post-Depegging, How is Strategy Restoring the Capital Flywheel?

On July 31, 2026, Bitcoin treasury company Strategy released its Q2 financial report. Despite a 6.9% year-over-year increase in revenue to $122 million, the company recorded a substantial net loss of $8.22 billion, primarily due to $8.32 billion in unrealized losses from Bitcoin holdings. While Strategy's core Bitcoin strategy remains intact—its holdings grew 11% to 843,775 BTC—the company is undergoing a fundamental shift in its capital model. Following the de-pegging of its key financing tool, the STRCoin (STRC), from its $100 target in May, Strategy has pivoted from a one-directional "raise funds, buy Bitcoin" cycle to a more dynamic, multi-asset capital management approach. A key part of this new framework is the "Monetization Program," through which Strategy has sold approximately $218.4 million worth of BTC to bolster liquidity. The company's top priority is repairing STRC's peg, committing not to issue discounted shares until it returns to its target range. It has initiated a $1 billion buyback program for discounted digital credit securities, having repurchased $28.9 million face value of STRC so far. Management aims to restore the peg around September 8, 2026. Strategy now actively manages a matrix of assets: Bitcoin (for accumulation or strategic sales), USD cash reserves (now at $3.75 billion), common stock (MSTR), and digital credit securities like STRC. This allows for tactical moves like repurchasing discounted debt or equity to capture value. The future success of Strategy's "capital flywheel" hinges on two factors: the short-term ability to successfully re-peg STRC to restore market confidence in its digital credit system, and the long-term price trajectory of Bitcoin, upon which its entire investment thesis ultimately depends.

Odaily星球日报07/31 03:32

STRC's First Financial Report Post-Depegging, How is Strategy Restoring the Capital Flywheel?

Odaily星球日报07/31 03:32

Microsoft CEO Satya Nadella's Latest Warning: Betting Entirely on a Single AI Model Hands Over a Company's Lifeblood

Microsoft CEO Satya Nadella warns that companies relying solely on a single AI model could jeopardize their survival. He argues that over-dependence leads to "vendor lock-in," where businesses risk ceding control over their core data, memory, contextual history, and AI usage patterns. This dependence essentially outsources a company's critical thinking and operational know-how to an external provider. The deeper a company integrates with one AI system—feeding it prompts, internal data, and workflows—the more it reveals its unique business methods and competitive edge. This accumulated knowledge could become accessible to the AI supplier. Furthermore, switching providers becomes extremely costly and complex, as companies would need to rebuild their entire AI-augmented workflow, memory, and tool integrations from scratch. Nadella's solution is "decoupling." Companies should separate their proprietary data, memory, and control layer (or "harness") from the underlying AI models. By retaining metadata from every AI interaction, businesses can preserve their operational "brain" or institutional knowledge. This allows them to flexibly use different AI models (e.g., from OpenAI, Anthropic, Microsoft) for specific tasks without losing their accumulated expertise. The core idea: companies can rent the smartest models available, but they must keep their own "brain" and operational control firmly in-house.

marsbit07/28 11:36

Microsoft CEO Satya Nadella's Latest Warning: Betting Entirely on a Single AI Model Hands Over a Company's Lifeblood

marsbit07/28 11:36

73 Sell Trades, 0 Buy Trades: Is This the 'Long-termism' of Circle's Management?

**Title: "73 Sales, 0 Buys: Is This Circle Management's 'Long-Termism'?"** In July, Circle President Heath Tarbert claimed the company was "playing the long game" and its stock price would "take care of itself." However, an analysis of SEC Form 4 filings reveals a contrasting picture: since CRCL's IPO, Circle's top executives and board members have consistently sold shares while making zero purchases. Key insiders, including CEO Jeremy Allaire, multiple board members, the CFO, CPTO, CAO, and President Tarbert himself, have executed a total of 73 sell transactions, collectively cashing out approximately $664 million. Tarbert alone has sold 10 times for about $30.77 million. While insider sales can be routine for diversification, the complete absence of any buys—especially after CRCL's ~70% price drop from its highs—raises questions. This one-sided trading pattern seems at odds with management's public assurances of long-term confidence and suggests they may not view the depressed stock as a compelling buying opportunity. The article argues this activity amplifies existing market doubts. Investors are reevaluating whether Circle's valuation is justified, questioning if it is truly a future financial infrastructure platform or remains primarily dependent on USDC's scale and the interest rate environment. Ultimately, Circle will need to demonstrate its long-term narrative through tangible business performance.

marsbit07/20 08:53

73 Sell Trades, 0 Buy Trades: Is This the 'Long-termism' of Circle's Management?

marsbit07/20 08:53

活动图片