Search the archive

TOKENSPEED — FEEL LLM TOKENS-PER-SECOND

Tokenspeed visualizes various LLM token-per-second rates across different content types to help users internalize the actual speed and feel of hardware benchmarks.

GITHUB CONFIRMS BREACH OF 3,800 REPOS VIA MALICIOUS VSCODE EXTENSION

GitHub confirmed that approximately 3,800 internal code repositories were compromised after an employee unknowingly installed a malicious Visual Studio Code extension.

ON GOOGLE DECLARING WAR ON THE WEB

Google's transition toward AI-generated search results threatens the open internet by replacing direct links to sources with a controlled, proprietary abstraction layer that devalues original human work.

See how video can elevate every stage of your funnel

SKECHERS MUST FACE CLASS ACTION OVER MISLEADING PROMOTIONAL EMAILS

Retailers now face real legal risk for using fake urgency in email subject lines, especially in Washington state, where more than 100 brands have already been sued over extended or repeated “limited time” offers. The lawsuit against Skechers shows that brands cannot treat phrases like “Today Only!” or “Ends Tonight” as harmless conversion tactics if the same promotion returns the next day. Brands should align promotional deadlines with actual campaign end dates and audit automated email flows that recycle urgency language across extensions. The court also rejected the argument that extending a sale helps consumers, which signals that intent matters less than whether the messaging creates false scarcity.

EXPEDIA IS PREPARING FOR A FUTURE BEYOND TRAVEL WEBSITES

AI is shifting travel planning from websites to chatbots and assistants, fragmenting how trips are discovered and planned. Expedia is positioning itself less as a consumer site and more as the infrastructure powering bookings. Travelers increasingly use AI for discovery but still prefer established platforms for transactions due to reliable fulfillment when things go wrong.

HOW THE COFOUNDER/CEO OF BEEHIIV RUNS HIS NEWSLETTERS IN MINUTES

This system automates five newsletter workflows. First, an AI editor turns rough drafts into publish-ready copy using a learned writing style and fact checks. Second, a subject line tool uses performance data to improve open rates through tested patterns. Third, an ad copy tool rewrites sponsor content to match voice and optimize clicks. Fourth, automated reports generate weekly branded performance PDFs for advertisers. Fifth, lead gen identifies subscriber employers, scores advertiser fit, and sends qualified leads in Slack daily.

DON'T ASK FOR REVIEWS ON WEEKENDS

Weekday timing matters more than convenience when asking for reviews. Researchers analyzed nearly 400 million reviews across 33 platforms and found weekend reviews scored .04 stars lower on average, while negative detractor reviews increased by 5.7%. Brands that rely on weekend traffic, like restaurants, bars, and hotels, face higher risk because service quality often drops during busy periods. Marketers should schedule review reminder emails during weekdays, lead with positive social proof, and test incentives like loyalty points or discounts since incentives increased review positivity by up to 83.4% in the study.

GOOGLE ASK MAPS: HOW TO OPTIMIZE FOR VISIBILITY

Google Ask Maps changes local SEO from a ranking problem to a recommendation problem, surfacing only 3-8 businesses per query and explaining why each is a fit rather than presenting a long list. The system draws on Google Business Profile data, review language patterns, website content, and external sources to build enough confidence to recommend a business for specific situations. Businesses that rely on generic service descriptions and broad profiles are at a disadvantage, while those with detailed reviews describing specific outcomes and content organized around customer decision-making scenarios are more likely to appear.

HOW TO MAKE CLAUDE BRUTALLY HONEST

This prompt turns Claude into a more rigorous, truth-first assistant that prioritizes accuracy over sounding confident. It avoids fabricating sources or numbers, flags unverified claims, and distinguishes facts from assumptions. It enforces transparency about what is known, unknown, or inferred, requires credible sourcing when possible, and pushes the model to avoid guessing, especially on recent or changing information.

THE FIGMA DESIGN AGENT IS HERE

Figma's new on-canvas design agent can generate multiple landing page or campaign design directions in parallel from a single prompt, then automate bulk edits like swapping components or updating typography across files. For marketing teams without dedicated design resources, the agent handles repetitive production work like converting assets to dark mode and applying brand design systems consistently. It also synthesizes scattered feedback from stakeholders into actionable next steps, reducing the back-and-forth that typically slows campaign delivery.

AI CITATIONS VS MENTIONS: WHAT'S THE DIFFERENCE AND WHAT'S MORE IMPORTANT?

AI visibility depends on two separate systems: mentions, which build brand awareness inside AI models, and citations, which drive referral traffic through linked sources. Mentions come from repeated associations in training data and retrieval, so brands should consistently tie themselves to target descriptors across PR, social, and on-site content. Citations depend on retrieval quality, which means content must rank well, stay fresh, and use tightly focused sections that AI can easily chunk and reuse in answers. Research shows only 13% of brands are both cited and mentioned, and 76.1% of cited URLs are unique to a single AI platform, so you need platform-specific visibility tracking instead of assuming one optimization strategy works everywhere.

GOOGLE'S NEW UNIVERSAL CART WANTS TO FOLLOW YOUR ENTIRE SHOPPING JOURNEY ACROSS THE INTERNET

Google's new Universal Cart lets users save products from across Search, Gemini, YouTube, and Gmail into a single hub that tracks price drops, flags compatibility issues, and alerts on restocks. The more significant announcement may be Agent Payments Protocol (AP2), which allows AI agents to autonomously complete purchases within user-defined spending limits and brand preferences.

X IMPLEMENTS NEW POSTING RESTRICTIONS ON NONPAYING USERS

X slashed daily posting limits for non-paying users from 2,400 to 50 original posts and 200 replies.

GEMINI OMNI IS GOOGLE'S NEW WORLD MODEL, WITH ADVANCED AI VIDEO GENERATION CAPABILITIES

Gemini Omni Flash, capable of both creating and conversationally editing video, launches this week in YouTube Shorts and YouTube Create at no cost.

THE DEVIL WEARS PRADA - BUT FOR MARKETERS (3 MINUTE VIDEO)

A parody with Meryl Streep as CMO, Stanley Tucci as VP, and Anne Hathaway as the new hire who thinks all 4 P's look the same.

See it how it works and get started for free.

EVERYTHING GOOGLE ANNOUNCED AT I/O 2026: GEMINI, SEARCH, ANDROID XR, & MORE

Google announced tons of new Gemini-powered features across its biggest products and services at I/O 2026. This page compiles all of the consumer-facing and most notable developer announcements. Gemini 3.5 Flash, which surpasses 3.1 Pro in coding, agent, and multimodal benchmarks, is rolling out today in the Gemini app, Search, Antigravity 2.0, and Gemini API. Gemini 3.5 Pro, which is currently in testing, will be available next month.

SPACEX IS PLANNING TO BUY STARTUP CURSOR 30 DAYS AFTER IPO

SpaceX is expected to file for its IPO as soon as today and list its shares on June 12. It plans to acquire Cursor in July. If the deal doesn't go through, SpaceX will pay Cursor a $10 billion breakup fee in cash. The deal will likely still require regulatory review.

TESLA'S NEWEST ELECTRIC VEHICLE COULD JOLT THE TRUCKING INDUSTRY

Early reviews of the Tesla Semi suggest it could be a much-needed hit for the company. It is much less expensive than competing heavy-duty electric trucks and can travel farther. Cost and range are two of the main reasons that many logistics and delivery firms have been reluctant to buy electric trucks. Trucking firms in California have asked the state government for subsidies to help them buy more than 1,200 Tesla trucks, more applications than for all other electric trucks since the state's incentive program began in 2019.

TWO AI-BASED SCIENCE ASSISTANTS SUCCEED WITH DRUG-RETARGETING TASKS

Nature recently released two papers that describe AI systems intended to help scientists develop and test hypotheses. Google's Co-Scientist is designed to keep scientists in the loop, so researchers have to regularly apply their judgment to direct the system. FutureHouse's system can evaluate biological data from specific classes of experiments. Both models are designed to process massive amounts of information to create largely straightforward hypotheses, like whether a drug will work for a specific condition.

Save your spot today

ANTIGRAVITY 2.0

Google Antigravity is a dedicated platform for working with agents. It allows users to orchestrate multiple autonomous agents working in parallel across independent projects. Antigravity features an abstracted UI, dynamic subagents, scheduled tasks, artifacts, extended customization, live voice transcription, and more. It is available on macOS for both Apple Silicon and Intel.

GEMINI 3.5 FLASH: MORE EXPENSIVE, BUT GOOGLE PLANS TO USE IT FOR EVERYTHING

Google released Gemini 3.5 Flash at Google I/O, skipping the preview and going straight to general availability. It mostly has the same set of platform features as the previous Gemini 3.x series, but without computer use. The model's knowledge cut-off is January 2025, and it supports 1,048,576 input tokens and 65,536 maximum output tokens. Gemini 3.5 Flash is accompanied by a notable price bump.

ANDREJ KARPATHY, TESLA ALUM AND OPENAI CO-FOUNDER, JOINS ANTHROPIC

OpenAI co-founder and former Tesla executive Andrej Karpathy has joined Anthropic's pretraining team, which oversees the data training process for the company's AI models. He will lead a new group focused on using Claude to accelerate pretraining research. Karpathy previously led a team at Tesla that oversaw the functionality of the company's Autopilot advanced driver assistance system, and he was also a founding research scientist at OpenAI until 2017.

POLYMARKET LAUNCHES PRIVATE COMPANY TRADING SO INVESTORS CAN SPECULATE ON ANTHROPIC, OPENAI

Polymarket is launching prediction markets tied to private company milestones. This will allow traders to bet on valuations, IPO timing, and secondary-market activity for companies like OpenAI and Anthropic. The system will use the Nasdaq Private Market as the exclusive resolution data provider. Traders can take a position on whether specific private-market events happen, though without equity ownership, shares, or voting rights. Individuals will be able to engage with private-market value creation earlier.

Make your agents better with Zero.

OPENAI HAS THE SMARTER MODEL. ANTHROPIC IS WINNING ANYWAY

OpenAI's frontier models beat Anthropic's in benchmarks, but Anthropic is still in the lead for enterprise spending.

FORTNITE RETURNS TO THE APP STORE WORLDWIDE AS EPIC SIGNALS 'FINAL BATTLE' WITH APPLE

Epic is confident that governments around the world will not allow Apple junk fees to stand now that the company is forced to show its costs.

SOFTWARE'S CENTAUR ERA

There was a period where the best chess performance was achieved through a human wielding a chess machine.

BE TLDR'S FIRST VP MARKETING ($300K-$400K, FULLY REMOTE)

TLDR is hiring a VP of Marketing to scale growth, pipeline, and brand across 8M+ subscribers, 13 newsletters, and advertisers like AWS, Google Cloud, Microsoft, and Anthropic. Apply here [21].

DISNEY ERASED FIVETHIRTYEIGHT

Disney erased the FiveThirtyEight site - roughly 200,000 human-hours of work.

GITHUB IS INVESTIGATING UNAUTHORIZED ACCESS TO ITS INTERNAL REPOSITORIES

There is no evidence of impact to customer information stored outside of GitHub's internal repositories.

DUMB WAYS FOR AN OPEN SOURCE PROJECT TO DIE

The maintainer left or burned out, the project was sabotaged or captured, the release pipeline broke, force majeure, or the world moved on.

GEMINI 3.5 FLASH

Google has introduced Gemini 3.5 Flash, a new model focused on agentic workflows, coding, and long-horizon task execution. The release also expanded Gemini access across Search, enterprise tools, Android Studio, and Google's developer platforms.

OPENAI ANNOUNCES NEW GUARANTEED CAPACITY OFFERING FOR CUSTOMERS TO SECURE COMPUTE

OpenAI's new Guaranteed Capacity offering allows customers to secure long-term access to compute to power AI products, agents, and workflows. Customers can choose between one-, two-, and three-year-long commitments, with discounts based on how long the commitment is. The company will offer Guaranteed Capacity until it sells out of its current allocation. It plans to offer it again in the future.

KARPATHY JOINS ANTHROPIC

Andrej Karpathy announced he has joined Anthropic, citing the next few years at the LLM frontier as especially formative for his return to R&D. Karpathy noted he remains passionate about education and plans to resume that work later, signaling the move is research-focused rather than a permanent pivot away from teaching.

GOOGLE DETAILED THE SHIFT TOWARD AGENTIC GEMINI PRODUCTS

At I/O 2026, Google outlined how Gemini models were being integrated across consumer products, creative tools, and developer platforms. The company also shared that monthly token usage across its AI systems had grown to more than 3.2 quadrillion.

MODEL HALF-LIFE

Model half-life - the idea that model releases will keep getting faster and faster - doesn't really make sense under consideration. Model releases have definitely increased in pace, but the release time isn't halving every six months. This post looks at model release dates for several of the most well-known models and lays out predictions for when the next models might come out.

SEE HOW DEVELOPERS BUILD AGENT MEMORY →

USING CLAUDE CODE: THE UNREASONABLE EFFECTIVENESS OF HTML

HTML's richness allows it to convey complex information more effectively than Markdown, including layouts, data tables, and interactive elements. It enhances readability by organizing specs into well-structured, easily navigable documents and offers better sharing and interaction capabilities. Claude Code uses HTML to efficiently ingest context from various sources, aiding in specs, design prototyping, and creating custom editing interfaces with improved engagement and clarity.

OLMOEARTH V1.1: A MORE EFFICIENT FAMILY OF MODELS

OlmoEarth v1.1, a new model family, reduces compute costs by up to 3X while maintaining performance, making planet-scale mapping more affordable. The models efficiently process remote sensing data by optimizing token sequence lengths, crucial for reducing computational costs. Methodological improvements allow similar performance to the original version with significantly less compute, benefiting developers and enhancing scientific research in remote sensing.

REAL-TIME LONG VIDEO GENERATION

NVIDIA's LongLive 1.0 is a framework for interactive long-form video generation that supports sequential prompting and real-time user-guided editing using streaming attention and KV-cache optimization techniques.