AI Swarms, RSC Drama and The Bots Reading Your Site

…all packed and written for you in Malaga at 40+ degrees Celsius☀️

OpenAI and Anthropic are flexing new model updates, Cursor is building agent swarms, and developers are already finding ways to turn AI hallucinations into supply-chain hacks (because of course they are). From local AI projects built by PewDiePie to debate over open-source models and context engineering, these links have everything happening in AI right now before it all gets replaced by next week’s framework.

Devs, on the other hand, are still arguing over React Server Components so much so that TanStack threw in the towel on them entirely while CSS anchor positioning and subgrid are quietly making layout headaches a thing of the past. Meanwhile, AI agents are officially rendering human-centric GitHub pull requests obsolete, and Model Context Protocol is moving in to replace traditional APIs before we even finished writing the docs.

Finally, marketers are busy fighting AI-generated fluff and manufacturing fake hype on Reddit (I should know, I’ve been guilty of this for 15 years now). Also, LLMs are literally scraping your site in real-time, and ChatGPT turns out to be your biggest reader. Meanwhile, if you’re not pairing your e-commerce setup with structured PIM data and Schema markup, the AI crawlers won’t even know what product you’re trying to sell them.

Happy reading.

PLEASE🙏 Should you find emails from SWIPE unwanted or annoying, unsubscribe. If you’re interested in learning how and when you might end up in SWIPE and what you can expect in the future, check it out here.

one document, two hands - Sunil Pai argues that AI agents should operate alongside users inside applications rather than sitting as a chat barrier between the user and the software.

Agent swarms and the new model economics - Cursor explores how organizing AI agent swarms into tree-like planner/worker hierarchies maximizes context efficiency and drastically reduces execution costs.

The Arguments Against Open Source AI are Very Bad - Tom Bedor critiques common arguments against open-source AI, highlighting how open-source software underpins commercial technology and is virtually impossible to suppress effectively.

New Study Identifies 53 Slopsquatting Targets Across 5 Frontier LLMs - Socket analyzes research showing that multiple frontier LLMs hallucinate identical nonexistent package names, exposing developer ecosystems to potential supply-chain attacks.

The 2026 Generative AI Landscape Report - Similarweb presents data and insights on generative AI search adoption, evolving user habits, and changing web referral trends.

AI Halftime Report: H1 2026 - Kevin Indig reviews the state of AI in the first half of 2026, examining how AI’s influence on search and web traffic continues to grow while measurement capabilities lag behind. By Growth Memo

The new rules of context engineering for Claude 5 generation models - Anthropic shares lessons learned from drastically reducing Claude Code’s system prompt and designing simplified tool interfaces for advanced AI models.

What PewDiePie Is Building in AI Now: Odysseus Is Becoming a Serious Local AI Workspace - Jenuel Dev covers PewDiePie’s open-source local AI workspace project, Odysseus, detailing its features, hardware compatibility, and active community development.

16 AI Prompt Templates for Better AI Agent Outputs - Zapier provides a collection of structured prompt templates designed to improve output quality and consistency when working with AI chatbots and agents.

How GPT-5.6 fuses frontier intelligence with frontier efficiency - OpenAI outlines how GPT-5.6 improves efficiency across inference and agentic workflows to deliver higher intelligence at lower costs.

State of CSS 2026 - Devographics presents the annual State of CSS survey results, showcasing trends like anchor positioning as a favorite feature alongside developer insights on browser compatibility and AI adoption.

Building a Native LLM Chat App with RAG - Margelo details how to construct a fast, native mobile chat application using Retrieval-Augmented Generation (RAG) and local LLM execution.

ReactBench: Benchmarking AI Agents on Real React Code - The Million.js team introduces ReactBench, a specialized benchmark that tests coding agents on real-world React codebases and static analysis rules.

MCP vs API - Crystallize compares Model Context Protocol (MCP) against traditional REST and GraphQL APIs, explaining how MCP simplifies context sharing for AI agents.

How to Host a Static Site on AWS for the Price of a Domain - Atomic Object walks through setting up a low-cost static website on AWS using S3, CloudFront, and Route 53.

Rebuilding FIFA Standings Layout with CSS Subgrid - Ahmad Shadeed demonstrates how to recreate complex tournament bracket layouts using modern CSS Grid and Subgrid.

State Management - Alex Russell analyzes the evolution and pitfalls of web application state management patterns in modern frontend architecture.

What Even Are Microservices? - var0 breaks down the fundamental definitions, trade-offs, and practical realities of microservice architectures versus monolithic setups.

We Stopped Using RSC on tanstack.com - The TanStack team outlines why they moved away from React Server Components on their main documentation platform in favor of client-side routing.

Experimenting with RSCs for Performance and UX in Next.js - Aurora Scharff explores practical React Server Component techniques in Next.js for URL-driven pagination, streaming search, and server-rendered draft previews.

State of Zero-Runtime CSS-in-JS - dx-styles provides an overview of zero-runtime CSS-in-JS tools and build-time styling solutions across the modern JavaScript ecosystem.

How to Think in Remix UI Instead of React - Sergio Xalambrí explains how shifting your mental model toward web standards, URL state, and isolated client entries improves application structure in Remix/React Router.

Soft Navigations and Interaction Performance Metrics - Chrome for Developers explains the Soft Navigations API for tracking route transitions and Core Web Vitals in Single Page Applications.

GitHub Is the Wrong Shape for This New World - Depot argues that human-centered PR workflows are becoming a bottleneck for agentic software delivery, requiring high-throughput infrastructure primitives instead.

New Research: The Core Web Vitals Thresholds You Trust Might Be Wrong for Your Site - Tammy Everts presents RUM research showing that Google’s 2.5-second LCP benchmark may not reflect optimal bounce and conversion rates for individual e-commerce sites.

Which AI actually reads your site? Two months of LLM traffic, measured - Evil Martians breaks down two months of server logs measuring LLM agent traffic, revealing that ChatGPT accounts for most reads while heavily preferring rendered HTML over raw Markdown.

5 Storytelling Frameworks to Attract B2B Buyers - Pierre Herubel outlines narrative structures designed to help B2B brands move beyond self-centered product updates and build engaging thought leadership. By Pierre’s Content Guides

In Defense of Brand Myth - Ana Andjelic explains why long-term brand heritage and foundational mythologies remain essential anchors for building desire in an era dominated by algorithmic content. By The Sociology of Business

Create a new Search profile - Google details the requirements and process for eligible creator accounts to set up integrated Search profiles across Google Search and Discover.

PIM vs. E-Commerce Platform - Crystallize clarifies the distinct roles of Product Information Management systems versus e-commerce platforms and how pairing them optimizes complex product catalog workflows.

ChatGPT Topic Authority Study - Semrush investigates how ChatGPT determines domain authority and selects source references when generating answers for topical user queries.

How to Build a Brand Presence on Reddit in 2026 - Social Media Today provides actionable strategies for businesses to authentically engage subreddits and navigate Reddit’s evolving community norms.

Fake fans, fake buzz: how your favourite band got big - The Guardian explores the underground world of manufactured online hype, streaming manipulation, and paid astroturfing within the modern music industry.

Why Your Content Sounds Like AI (and How to Fix It) - WordStream identifies common stylistic patterns that trigger “AI-written” detection in marketing copy and provides editing tips to restore a natural human voice.

How to Evaluate Google Ads Creative Before Testing - Search Engine Land outlines a pre-launch framework for assessing ad copy and visual assets to maximize performance before spending budget on live campaigns.

The Best Time to Post on LinkedIn - Buffer analyzes platform engagement data to pinpoint optimal posting windows and days for reaching professional audiences on LinkedIn.

How to Increase Facebook Engagement - Hootsuite shares practical tactics for improving organic reach and driving meaningful audience interactions under Facebook’s current feed algorithm.

Does Schema Markup Impact SERP Rank? - Moz examines data on structured data implementation to clarify its direct and indirect effects on search engine rankings and rich result visibility.

Boost YouTube Creative ROI - Think with Google presents data-backed creative guidelines for video ad formats designed to maximize campaign return on investment on YouTube.

15 Marketing Tasks You Should Automate Right Now - WordStream highlights key repetitive marketing workflows across email, social media, and lead management that can be streamlined with automation tools.

¯\_(ツ)_/¯

🛠TOOL you should check out: 15 GTM Prompts for Claude Opus 4.8 not a tool, but well worth downloading.

🎸LISTEN TO: Dire Straits - Six Blade Knife.

Happy reading /bookmarking!

If you enjoyed these, please SUBSCRIBE, recommend 👏 and share to help others find us! Let us know what you think in the comments section below.

🐦Follow SWIPE on TWITTER