RT Avi Chawla (@_avichawla)I just built my own multi-agent deep researcher!It uses a 100% local LLM and MCP.Here's an overview of how it works:- User submits a query- Web agent searches with Bright Data MCP tool- Research agents generate insights using platform-specific tools- Response agent crafts a coherent answer with citationsTech stack:- Bright Data MCP for real-time web access- CrewAI for multi-agent orchestration- Ollama to locally serve GPT-OSSWhy Bright Data MCP?To build this workflow, we needed to ...

for today's tech check. Mackenzie Sagalos joins us now with data brick CEO Ali Gatsy in a CNBC exclusive. Mac, >> hey there Mike.So Ally, thanks so much for joining us. This is OpenAI's first major integration with a company like Data Bricks, but what does Data Bricks get out of this partnership. Are you seeing signals that enterprise adoption hasn't been scaling as fast as hoped.>> Yeah, so what we're seeing is that enterprises really really are excited to get agents actually working inside the enterprise. ...

Summary of Alibaba Group (BABA) Conference Call Company Overview - **Company**: Alibaba Group (BABA) - **Event**: Alibaba Cloud APSARA Conference, Day 1 - **Date**: September 24, 2025 - **Location**: Hangzhou Key Industry Insights - **AI Development**: Alibaba is currently at Stage 2 of 4 in its roadmap to achieve Artificial Superintelligence (ASI), with predictions that Large Language Models (LLMs) will become the next operating system and AI Cloud will be the next computer [1][14] - **Investment in AI**: The global AI industry has seen over US$400 billion in investments in the past year, with expectations to exceed US$4 trillion in the next five years [14] - **Data Center Expansion**: Alibaba anticipates a 10x increase in total data center power consumption by 2032 compared to 2022 [4][18] Financial Performance and Projections - **Stock Performance**: Following the conference, Alibaba's shares reacted positively, with a 10% intra-day increase. The stock has seen a year-to-date increase of over 110% [1] - **Valuation Scenarios**: The refreshed bull case valuation implies a potential upside of over 40%, with a target price of US$179/HK$174. The bear case valuation stands at US$106/HK$103 [1][5] - **Revenue Growth**: Alibaba Cloud's AI revenues have seen triple-digit growth for eight consecutive quarters, now accounting for 20% of total cloud revenues [5] Product and Service Developments - **AI Model Releases**: Alibaba has released over 300 AI models, achieving 6 million downloads and serving over 1 million customers with 170,000 derivative models [18] - **Flagship Model**: The Qwen3-Max model, with 1 trillion parameters, has surpassed competitors like GPT-5 and Claude Opus 4, ranking third on the Chatbot Arena leaderboard [18] - **Agent Development**: The upgraded low-code Agent development platform has seen over 200,000 developers build 800,000 agents, indicating strong adoption across various sectors [23] Strategic Focus Areas - **Global Expansion**: Alibaba Cloud is expanding its international presence, with new data centers planned in regions including Japan, South Korea, France, UAE, Brazil, Malaysia, Netherlands, and Mexico within the next 12 months [24] - **AI Infrastructure**: The company has significantly increased its AI computing power by 5x and storage capacity by 4x over the past year [23] Risks and Challenges - **Market Risks**: Key risks include lower-than-expected GMV growth due to macroeconomic factors and competition, as well as potential deceleration in cloud revenue growth [7] - **Profitability Concerns**: The quick commerce segment is expected to impact group EBITA negatively in the near term, although improvements in unit economics are anticipated [6] Conclusion - **Investment Recommendation**: The company is rated as a "Buy" with a focus on its growth potential in AI and cloud services, differentiating itself from other platforms through its unique offerings and strategic investments [5][6]

Core Viewpoint - The article discusses the recent talent shifts within Meta and the implications for its organizational structure and strategy in the AI sector, highlighting the challenges and opportunities faced by the company in the competitive landscape of AI development [4][6][21]. Group 1: Meta's Talent Acquisition and Loss - In June 2025, Meta acquired a 49% stake in Scale AI for $14.3 billion and recruited Alexander Wang, the 28-year-old founder of Scale AI, to lead the newly formed Meta Superintelligence Labs [4]. - Following the acquisition, Meta experienced a wave of talent departures, including long-term employees and new recruits returning to OpenAI, indicating dissatisfaction with the company's environment [4][8]. - The rapid turnover of talent is attributed to an increasingly bureaucratic structure and internal political struggles, which have made the work environment less appealing for top-tier AI talent [8][9]. Group 2: Organizational Structure and Culture - Meta's organizational structure has become more cumbersome, with an increase in VP levels leading to slower decision-making processes, which contrasts with the company's previously agile culture [8][9]. - The lack of clear ownership in model training and the presence of overlapping responsibilities among teams have created inefficiencies and internal competition, hindering productivity [10][11]. - The article suggests that a smaller, more focused team of 150 to 250 individuals would be more effective for achieving breakthroughs in AI models compared to a larger team of 5,000 [9][10]. Group 3: Comparison with Other AI Companies - Other AI companies like OpenAI and Anthropic have a more mission-driven approach, which helps align their teams towards common goals, reducing internal conflicts and enhancing productivity [12][21]. - Google employs a top-down approach with clear authority figures guiding research, which contrasts with Meta's bottom-up culture that can lead to disorganization [10][12]. - The article highlights that while Meta has a strong social network, its organizational inefficiencies may hinder its ability to compete effectively with companies like OpenAI and Anthropic, which are currently attracting top talent [23][24]. Group 4: Future of AI Organizations - The article discusses the potential for new organizational structures in AI startups, emphasizing the importance of decentralization and trust within teams to enhance efficiency [26][27]. - It suggests that AI can significantly improve organizational productivity, allowing for a shift away from traditional hierarchical structures towards more agile, networked teams [26][27]. - The future of talent competition in Silicon Valley is expected to cool down as market expectations are reassessed, impacting the recruitment of top AI talent [34][35].

If you found it insightful, reshare it with your network.Find me → @_avichawlaEvery day, I share tutorials and insights on DS, ML, LLMs, and RAGs.Avi Chawla (@_avichawla):Pytest for LLM Apps is finally here!DeepEval turns LLM evals into a two-line test suite to help you identify the best models, prompts, and architecture for AI workflows (including MCPs).Works with all frameworks like LlamaIndex, CrewAI, etc.100% open-source with 11k stars! https://t.co/Xayu1aFGFV ...

AI is about to be a staple across all content.And Doodles is already ahead of the curve.3 years of LLM training powering future content on-demand ✅Iconic, recognizable characters ✅A strong foothold in culture with collabs with massive brands like McDonald's and most recently Kellogg's... ✅Stop fading this team. It’s Doodles season.Jack (@Jack55750):Burnt Toast explains how Doodles is utilising AI to become the next big global IP“We’re training an LLM on the last 3 years of our work we’ve made with GoldenWol ...

The White House is definitely keen to get a deal done. And uh as you said, there are multiple parties involved. Oracle is the publicly traded entity that's uh front and center in terms of being the lead company that will be owning Tik Tok along with these other private investors.And look, at the end of the day, it's clear the algorithm would be rewritten or they would be licensing it. that is that was a sticking point all along. So, uh Oracle already has the data for Tik Tok in their data centers.So, that p ...

K2 Think, the LLM newly launched by the UAE, is an efficient AI system reasoning model. It works its way through problems step by step, and is particularly effective at mathematical and coding tasks https://t.co/S3Fae5DB8T ...

“The LLM is a kind of an overachieving intern who never sleeps, eats lots of data, and is always super excited.”What happens when students become dependent on AI? Listen to “The Weekend Intelligence” https://t.co/74PFlrR5hW ...

The UAE’s choice to build its latest LLM on top of a Chinese model, rather than an American open-source alternative, is part of a careful calibration of the country’s positioning amid the geopolitical push and pull of AI https://t.co/juPGoJSpXW ...