Our podcast on markets, the economy and business. This week, how AI agents are redefining the shopping experience https://t.co/7WhqvaTtPm ...

The promise of AI agents is that they’ll act like human employees, deployed autonomously on tasks like navigating the web, sending emails and Slacks and working in Google Docs. But that also means they’ll need access to tons of applications, just like other staffers — creating additional work for IT teams managing that access, and increasing how vulnerable an organization is to identity-related hacks.Full story: https://t.co/9lp4v9MYTT📸: ConductorOne ...

AI Agent Challenges - Long-running AI agents face challenges working across many context windows [1] - Anthropic Engineering Blog explores creating a more effective agent harness inspired by human engineers [1]

Financial Data and Key Metrics Changes - In Q3 2025, Cheetah Mobile reported a total revenue of RMB 287 million, representing a 50% year-over-year increase, driven by a 151% growth in AI and other segments, which now account for 50% of total revenue compared to 30% in the same period last year [7][3] - The company achieved an operating profit of RMB 4 million, marking its first quarterly operating profit in six years, with a non-GAAP operating profit of RMB 50 million, compared to an operating loss of RMB 60 million in the same period last year [7][3] - Gross profit increased by 64% year-over-year, with gross margin improving to 75%, up from 68% in the year-over-year quarter [7] Business Line Data and Key Metrics Changes - The AI and other segments grew 151% year-over-year, contributing significantly to the overall revenue growth, while the internet business remained stable with a 6% year-over-year revenue increase [7][3] - The AI robotic business contributed about 15% of total revenue, growing approximately 100% year-over-year and 40% quarter-over-quarter [3][6] - The internet business delivered an adjusted operating profit of approximately RMB 21 million, up 55% year-over-year, while the adjusted operating loss for AI and other segments narrowed by 82% year-over-year [7] Market Data and Key Metrics Changes - The contract backlog for voice-enabled wheel robots in China increased by 32% from the previous quarter and has since doubled, indicating sustained demand [3][6] - The company noted strong demand for its voice-enabled robots, which are becoming more reliable and cost-effective for large-scale deployment [6] Company Strategy and Development Direction - Cheetah Mobile is focusing on building new growth engines through AI initiatives, particularly in AI robots and AI tools, which are seen as having enormous market potential [3][6] - The company aims to prioritize operating profit growth over revenue growth, particularly in its internet business, while managing costs and expenses in AI and other segments [7] - The strategy includes leveraging third-party and open-source models to enhance robot experiences and accelerate product updates [7] Management's Comments on Operating Environment and Future Outlook - Management expressed confidence in the company's turnaround, highlighting that 2025 has been a solid year with strong growth momentum [3] - The CEO emphasized the importance of disciplined execution, improved operational efficiency, and the potential of AI technologies to drive future growth [7] - The company is optimistic about the long-term potential of its AI robot business and AI-native tools, indicating a shift beyond the turnaround phase [6][7] Other Important Information - Cheetah Mobile holds minority investments in several companies within the AI space, which could enhance organic growth opportunities in the future [3] - The company maintains a strong financial position with cash and cash equivalents of about $224 million and long-term investments of $107 million as of September 30, 2025 [7] Q&A Session Summary - The Q&A session included questions from various analysts, but specific details of the questions and answers were not provided in the transcript [15][16][18][20][22][24][26][28][30][32]

AI Model & Strategy - Open AI models are essential for building intelligence and enabling customization for specific industry workflows [1] - Open source AI models offer increased visibility and customizability [2] - Synopsis' strategy heavily relies on AI to provide autonomous functionality in engineering workflows [4][5] - The industry anticipates AI agents to power transactions in the agentic economy [5] Agent & Technology - CrowdStrike aims to transform human expertise into AI agents for rapid response to cyber attacks [2] - Nemotron facilitates control over expertise and enables the creation of virtual analysts like Charlotte for security alert management [3] - Agent engineers are being developed with progressively higher levels of autonomy [3] - Open models deployed on GPUs, coupled with stack control, allow for fine-tuning, solving challenges in agentic commerce [6] - Nemotron enables the selection of the optimal balance between reasoning and performance, allowing for fine-tuning to achieve desired outcomes [7] Partnership - Nvidia and CrowdStrike collaborate to prevent AI-based attacks [6] - The industry relies on the partnership with Nvidia for components like Nemotron and Nemo agent kit to accelerate agent development [7]

AI Model Performance - Grok 4.1 Fast ranks 1 in ²-Bench for Telecom Agentic Tool Use with 93% accuracy, outperforming Claude Opus 4.5 & Gemini 3 Pro [1] - Grok 4.1 Fast costs up to 50x less than Claude Opus 4.5 [1] AI Agent Tooling - Tool calling is where the whole game is for AI agents, and Grok 4.1 Fast excels in this area [1]

Overview of X42 Protocol - X42 is an open payment protocol developed by Coinbase that enables websites and APIs to charge per request using stablecoins, facilitating microtransactions for both humans and AI agents [3][12] - The protocol addresses the web's original omission of a native way to pay for small things online without friction, aiming to bring pay-per-use back to the web in a programmable and instant manner [13] - Cloudflare and Coinbase established the X42 Foundation in September 2025 to promote a shared open approach to agent payments across the web, indicating a goal to scale beyond crypto-native circles [14] - Gartner analysts predict that machine customers, such as AI agents acting as buyers, will influence or participate in $30 trillion worth of purchases by 2030, highlighting the potential impact of X42 [15] Technical Aspects and Ecosystem - X42 functions by allowing a client (user or AI agent) to request something from a server, which responds with a 402 payment required message including the price and payment details; upon payment, the server provides the requested goods [17][18] - Coinbase operates a hosted facilitator that verifies and settles payments on-chain, processing USDC payments on Base without a facilitator fee; sellers can also self-host or use community facilitators [19] - The X42 discovery layer, or Bazaar, serves as a machine-readable index of servers that accept X42 payments, enabling AI agents and apps to discover available services and their costs [20][21] - Ethereum and Solana are closely involved, with X42 being implemented on Ethereum layer 2 Base and Solana, positioning these ecosystems at the forefront of real payment flows [24][25] Market Traction and Future Outlook - In October, X42 activity surged by over 10,000% in a 4-week period, reaching a record daily high of over 680,000 transactions, indicating substantial initial interest and usage [37] - Google's Agent Payments Protocol (AP2) has added support for an X42 extension via Coinbase's facilitator, linking X42 to an emerging agent standard from a major tech company [41] - Following the initial surge, trading volume decreased by over 90% from its peak, and daily transactions fell by more than half, suggesting that the initial hype outpaced market readiness [44] - The industry is monitoring ERC8004, a proposed Ethereum standard that would provide AI agents with a portable on-chain identity and shared registries for reputation and validation, which could significantly enhance the capabilities of Ethereum-based AI agents when paired with X42 [49][50]

Product Launch - Llama Cloud launches Llama Sheets API, enabling conversion of spreadsheets into structured parquet files for AI agent processing [1] - Llama Sheets is positioned as a generic document processing tool within Llama Cloud's existing suite of tools like Llama Parse, Llama Extract, and Llama Classify [2] - The API allows processing spreadsheets and using them directly in document workflows [3] - The API extracts each table as an individual object, along with a title and description [5][6] Technical Details - Parquet files are described as a structured format, similar to dataframes, that AI agents and coding agents can readily interact with [1][10] - The API can be integrated with coding agents like Cloud Code or frameworks like Llama Index [7] - The API includes examples and documentation for using it with coding agents [7] - The extracted data can be downloaded as a parquet file [10] Use Case & Feedback - The API can be used to analyze spreadsheets and automate workflows [14] - User feedback on the API's inputs, outputs, and features is encouraged via a feedback button in the UI [14]

Have you ever had your agent working for almost one hour only to understand that he went in the wrong direction or in the middle of something very important he ran out of context window. Me too. That's why in the last months I developed a workflow that consists in dividing a big feature into smaller markdown tasks.Hi, I'm Alex Gavesco and I'm going to present backlog MD, a tool for project management for AI agents and humans. Okay, let's start. So, have you ever seen a terminal cand board.Well, when I start ...

Right. Hello everyone. Uh today I will present meta adaptive context engineering or meta AC for short which is a new framework designed to optimize AI agents beyond single dimension approaches.We will explore how orchestrating multiple adaptation strategies can overcome the limitations of existing context engineering methods. Now a little introduction about myself. Uh so I'm Alberto Romero.I'm the co-founder and CEO at jointly. And for context at jointly we build the main specialized agents for regulated in ...