Search the archive

APPLE VS. THE AI HYPE CYCLE

Apple's strategy focuses on its durable ecosystem, prioritizing hardware sales over AI innovations, despite the industry's AI investment focus.

CLAUDE'S NEW CONSTITUTION

Two of the fifteen external contributors who helped review Claude's new constitution are Catholic members of the clergy.

OPENAI'S FORMER SALES LEADER JOINS VC FIRM ACREW: OPENAI TAUGHT HER WHERE STARTUPS CAN BUILD A ‘MOAT'

Aliisa Rosenthal, OpenAI's former sales leader, emphasizes context engineering as a key advantage, particularly in enterprise applications, and advocates for developing lighter, cost-effective models.

today, you'll find guidance to simplify your AI agent journey.

DEEPMIND CEO SAYS CHINESE AI FIRMS ARE 6 MONTHS BEHIND THE WEST

Google DeepMind's CEO, Demis Hassabis, says that Chinese AI companies remain around six months behind the frontier AI of the leading Western labs. He says the response to DeepSeek's R1 model last year was a massive overreaction and that while Chinese firms are very good at catching up to the frontier, they have yet to show they can innovate beyond that. Some of the barriers to Chinese AI development are set to come down as the Trump administration is easing a ban on advanced AI chip exports to China. Sales of the most advanced AI processors will remain blocked on national security grounds.

GEMINI IN CHROME IS GETTING “SKILLS” AS IT MOVES TOWARD BECOMING A FULL AI AGENT

Gemini in Chrome is being upgraded into a proactive agent capable of performing complex tasks on users' behalf. A hidden internal page for 'skills' has been spotted in testing. The page features a dedicated interface where users can define specific capabilities for the AI. Skills will effectively allow users to 'teach' Gemini how to handle specific or repetitive workflows within the browser.

THE MESSY HUMAN DRAMA THAT DEALT A BLOW TO ONE OF AI'S HOTTEST STARTUPS

Mira Murati says that Barret Zoph was fired from Thinking Machines Lab after repeated concerns about his conduct and lack of productivity. Zoph claims he was fired for expressing an intent to take a job elsewhere. Within hours of Zoph's firing, Zoph, another co-founder, and a third employee had signed offers to rejoin OpenAI. The departures are a sign of how the AI race is as much a battle for talent as technology.

AI STARTUP HUMANS& RAISES $480 MILLION AT $4.5 BILLION VALUATION IN SEED ROUND

AI startup Humans& has secured $480 million in seed funding, reaching a $4.5 billion valuation. Key participants included Nvidia, Jeff Bezos, and GV. The company aims to develop human-centric AI for improved communication and collaboration. A product launch is expected early this year. The founding team consists of experts from OpenAI, Google DeepMind, and Meta.

WITHOUT BENCHMARKING LLMS, YOU'RE LIKELY OVERPAYING 5-10X

Those paying for LLM APIs and have never tested alternatives are likely overpaying. There are almost always cheaper options with comparable quality for every workflow. Benchmarking and finding an optimal model is a complex process, but it can result in significant savings. This post discusses how one team navigated the process to save thousands of dollars.

THE TRAGEDY OF THE AGENTIC COMMONS

AI agents can improve matching markets, like dating or job searches, by using LLMs to convert unstructured preferences into actionable data, outperforming standard questionnaires. However, in markets where everyone uses AI agents, congestion increases significantly, reducing efficiency without a pricing mechanism. Introducing prices and institutional design can alleviate congestion, streamline coordination, and enhance market efficiency.

MY AGENTS ARE WORKING. ARE YOURS?

AI agents are significantly increasing productivity by automating the analysis of thousands of research papers. A new anti-AI tool, Poison Fountain, corrupts AI training data to combat the perceived threat of machine intelligence.

Try TTS-1.5 [12] Learn More

INTRODUCING FASTMCP 3.0 🚀

FastMCP 3.0 was built to be as durable as it is future-proof. It moves beyond simple tool servers and enters the era of Context Applications: rich, adaptive systems that manage the information flow to agents. FastMCP 3.0 can source components from anywhere, compose and transform them freely, personalize what each user sees, track state across sessions, and more. The first beta is available now.

REPO: CONTENT-AWARE ATTENTION REORDERING

RePo is a module that repositions tokens based on contextual relevance rather than fixed order, improving long-range attention and robustness to noisy or structured inputs. The method draws on Cognitive Load Theory to justify reorganizing input structure for more efficient model processing.

DIFFERENTIAL TRANSFORMER V2

DIFF V2 features key upgrades over DIFF V1: faster inference via compatibility with FlashAttention, better training stability by removing per-head RMSNorm, and a simplified λ parameterization that eliminates complex initialization schemes.

WAYPOINT-1: REAL-TIME INTERACTIVE VIDEO DIFFUSION FROM OVERWORLD

Waypoint-1 is a real-time-interactive video diffusion model. It can be controlled and prompted via text, mouse, and keyboard. Users can give the model some frames, run it, then have it create a world that they can interact with. They can move the camera freely with the mouse and input any key on the keyboard, all with zero latency. Each frame is generated with the controls as context. The model runs fast enough to provide a seamless experience, even on consumer hardware.

AI GAINS STARTING TO SHOW IN THE REAL ECONOMY

The introduction of AI tools like Claude Cowork and Alibaba's Qwen Assistant marks a shift towards AI's real-world applications, similar to ChatGPT's accessibility impact. Recent data from Q3 2025 shows a 4.9% productivity jump, with minimal change in hours worked, indicating potential initial AI-driven efficiency gains. Despite early-stage adoption, these developments could significantly boost economic productivity and eventually increase living standards.

ANTHROPIC CEO: SELLING H200S TO CHINA IS LIKE GIVING NUKES TO NORTH KOREA

Anthropic CEO Dario Amodei has likened the decision to allow Nvidia to sell GPUs to Chinese companies to giving nuclear weapons to an adversary. The Trump administration announced just over a month ago that it would allow shipments of Nvidia H200 accelerators to Chinese customers as long as the US received 25% of the revenues. Chinese authorities have yet to allow local buyers to acquire the GPUs. Anthropic wants stricter controls on AI exports, saying access to the chips would put Chinese model developers in a better position to compete with the West.

FULLY REMOTE)

As the first engineer on TLDR's new Applied AI team, you'll build AI agents and composable Claude Skills to let non-technical teammates create their own AI workflows. Learn more [21].

GOOGLE SAYS GEMINI WON'T HAVE ADS, AS CHATGPT PREPARES TO ADD THEM

While Google DeepMind CEO Demis Hassabis has confirmed that Gemini won't get ads for now, Google has not confirmed that it will never add ads to Gemini.

ELON MUSK SAYS TESLA'S RESTARTED DOJO3 WILL BE FOR ‘SPACE-BASED AI COMPUTE'

Tesla plans to restart its Dojo3 AI chip project focused on space-based AI compute.

CHATGPT SELF PORTRAIT

ChatGPT's response to a prompt about user treatment often reflects the user's behavior and framing.

GLM-4.7-FLASH

GLM-4.7-Flash is a 30B-class MoE model designed for local deployment in coding, agent tasks, translation, and creative writing.

OPENAI ROLLS OUT AGE PREDICTION IN CHATGPT

OpenAI introduced an age prediction model for ChatGPT consumer accounts to help apply teen-specific safety protections.

See first-hand the power of ELMs with the Velma Preview

ANTHROPIC WORKS ON KNOWLEDGE BASES FOR CLAUDE COWORK

Knowledge bases are persistent repositories that Claude can reference for relevant context and incrementally update with new information. They allow Claude to segment information into multiple, user-managed knowledge bases. Users will likely be able to select specific knowledge bases as context attachments when working on tasks in Cowork. The feature could be particularly relevant for workflows that involve automation and file management.

GOOGLE STITCH TO ADD API KEYS AND PRD GENERATION FOR PMS

Google Stitch will soon allow API key management and automatic PRD generation. The update aim to enhance design workflows by integrating Gemini models for high-resolution outputs and simplifying project documentation. This positions Stitch as a vital tool in Google's AI-driven development strategy, enhancing collaboration within product teams.

THE ASSISTANT AXIS: SITUATING AND STABILIZING THE CHARACTER OF LARGE LANGUAGE MODELS

Two components are important to shaping model character: persona construction and persona stabilization. Without the proper construction, assistants can easily inherit counterproductive associations from the wrong sources. Even when assistants are well-constructed, they easily drift away from their roles, which makes stabilizing and preserving models' personas particularly important. Ensuring models are aligned will become increasingly important as they become more capable and are deployed in increasingly sensitive environments.

SLONK: CHARACTER.AI INTERNAL SYSTEM

Character.ai uses an internal system called "Slonk" to manage GPU research clusters. Slonk merges the flexibility of Kubernetes with the productivity of HPC environments to optimize ML research workflows.

TRAINING CLI AGENTS WITH SYNTHETIC DATA AND RL

Nvidia showcases how a reasoning model with no prior knowledge can be taught to safely use the LangGraph CLI for complex tasks like spinning up servers and building Docker containers, using synthetic data, and Reinforcement Learning with Verifiable Rewards.

securing agents in production. Register now →

META'S 3D SHAPE GENERATION

ShapeR reconstructs full-scene 3D meshes from image sequences by extracting multimodal features and feeding them into a flow transformer that generates object-specific shape codes.

NATURAL LANGUAGE TO SOLVER-READY CODE

Microsoft's OptiMind translates natural language optimization problems into mathematical formulations. It was designed to streamline the traditionally slow and expert-heavy modeling step in optimization workflows.

SKYPILOT

skypilot is a unified open-source framework for running, managing, and scaling AI/ML workloads across clouds, Kubernetes, and on-prem infrastructure with a simple interface and intelligent scheduling.

KAGGLE COMMUNITY BENCHMARKS

Kaggle has launched Community Benchmarks, a feature that allows users to collaboratively build and share custom model evaluations that go beyond static accuracy metrics to better reflect real-world AI performance.

AI ENGINEERING HAS A RUNTIME PROBLEM

AI teams struggle to deploy agents because no standardized runtime exists to handle state, streaming, isolation, recovery, and scaling across sessions and users. Existing frameworks, observability, and eval tools only cover building and testing agents, leaving each team to reinvent complex infrastructure from scratch. A proper runtime would serve agents as APIs, persist state, manage streams, recover from failures, and scale safely, turning notebooks into production-ready AI applications.

FULLY REMOTE)

As the first engineer on TLDR's new Applied AI team, you'll build AI agents and composable Claude Skills to let non-technical teammates create their own AI workflows. Learn more [18].

THE COMMODITIZATION OF SERVICES

AI will make services ubiquitous and embedded in everything we do, but the businesses providing them will collapse.

LLMS AND YOUR CAREER

Technically interesting companies seek those with deep fundamental knowledge.

Most AI agents never reach production. Agent Bricks by Databricks

fixes that by helping you build agents that actually work — accurate, reliable and grounded in your data.

See why it's worth your time

A BUSINESS THAT SCALES WITH THE VALUE OF INTELLIGENCE

OpenAI believes that its business model should scale with the value intelligence delivers. It introduced consumer subscriptions as people demanded more capability and reliability, created workplace subscriptions, and added usage-based pricing so costs scale with real work getting done. It has more recently applied the principle to commerce, helping people choose what to purchase next and explore products. Advertising follows the same arc.

ADS IN CHATGPT

OpenAI plans to introduce ads in the free and ChatGPT Go tiers to support broader access to AI. Ads will be tested first in the US, while ensuring user trust, data privacy, and ad quality.

ELON MUSK'S XAI BRINGS 1GW COLOSSUS 2 AI TRAINING CLUSTER ONLINE

xAI's Colossus 2 supercomputer is now online. It is the first gigawatt-scale AI training cluster in the world, with upgrades still to come. xAI has been building infrastructure at a speed that far exceeds its primary rivals. The company's Colossus 1 and 2 systems now represent more than one million H100 GPU equivalents.

BUGBOT'S CODE REVIEW EVOLUTION

Cursor's Bugbot has evolved into a capable AI code reviewer that detects logic, performance, and security issues in pull requests.