AI Titus News — Wednesday, June 24, 2026CLAUDE NOW WRITES 65% OF ITS OWN CODE… ANTHROPIC SHIPS SLACK-NATIVE 'CLAUDE TAG' FOR ASYNC TEAM DELEGATION<br>★ Must-Read agentic engineering / build-better-agents Learn state of AI coding / useful technique<br>⚡ AI Frontier · smol.ai<br>★ Must-ReadThe 'harness era', beyond the modelIndustry thesis: the unit is no longer the model but model + harness + memory + config-management as one system. Core agentic-engineering read.
★ Must-ReadSakana 'Fugu' orchestration APIsLearns model selection + delegation across many models, multi-agent routing as infrastructure.
LearnDeepSWE coding benchmarkNew eval focused on realistic developer experience; a gauge for the state of AI coding.
LearnOpenAI GPT-5.5-CyberClosed-loop patch generation across 30M+ commits, where AI coding meets security.
GLM-5.2 open-weight modelZhipu's open model rivals Opus 4.8 and GPT-5.5, 1M-token inference, sparse attention.
Watch · latest videos<br>★ Must-ReadTag Claude in, right where you already workClaude works inside Slack with your channel's context and your org's tools, seeing multi-day work through alongside the team.<br>Claude / Anthropic · 67.8K views · 1.9K likes
★ Must-ReadPLANS for Fable 5: rebuilding my /Plan skillWhy planning beats vibe-coding, and how he rebuilt his /Plan skill for the new Fable 5 / Mythos-class models.<br>IndyDevDan · 11.6K views · 456 likes
★ Must-ReadMy AI memory now follows me across every toolGiving an agent persistent memory that travels across tools, so context isn't lost switching from Claude Code to others.<br>Cole Medin · 7.3K views · 262 likes
LearnI battle-tested Sakana Fugu's 'Fable killer'Putting Sakana's Fugu orchestration up against the frontier models in real agent workflows.<br>Nate Herk · 79.9K views · 1.9K likes
Learn12 open-source AI projects to try right nowA rundown of a dozen open-source AI projects worth a look this week.<br>Matthew Berman · 38.5K views · 2.4K likes
LearnShadow AI is terrifyingWhy unmanaged 'shadow AI' inside companies is a growing security risk, and how to find it.<br>NetworkChuck · 116.9K views · 4.5K likes
LearnHow to use Agent in Google FlowHow Google Flow's agent plans and reasons through complex creative tasks under your control, built on Gemini.<br>Google · 1.9K views · 165 likes
OpenClaw creator's new secret projectFirst look at what the OpenClaw creator is building next and why builders should care.<br>AI Jason · 1.8K views · 69 likes
Midjourney wants to delete 30% of all death...A fast take on Midjourney's latest move and where AI image generation is heading.<br>Fireship · 307.7K views · 12K likes
ChatGPT Futures, Class of 2026The ChatGPT Futures class on what they hope AI helps people do next: access, education, connection.<br>OpenAI · 2.9K views · 138 likes
Communities · Reddit<br>Unlimited-OCR is now on ModelScope! A 3.3B multilingual OCR model for one-shot parsing across single images, multi-page documents, and PDFs. License: MITr/LocalLLaMA
The Swiss Federal Supreme Court is evaluating Hereticr/LocalLLaMA
Seems this community might have missed it: Bill that would mandate AI chip location tracking gains industry support | Half a dozen companies have come out in support of the Chip Security Act, which would require location-tracking mechanisms for America’s most advanced computing chips.r/LocalLLaMA
State of AI Coding · leaderboards<br>Aider LLM leaderboardSWE-benchLMArena (Chatbot Arena)Aider polyglot
Builder Desk · Hacker News<br>We’re making Bunny DNS free693 pts · 223 comments
Founding a company in Germany: €9600, 152 days and I still can't send an invoice491 pts · 579 comments
There are a few things that I look back on as my mistakes in the early days345 pts · 169 comments
RubyLLM: A Ruby framework for all major AI providers236 pts · 34 comments
Raspberry Pi Pico W as USB Wi-Fi Adapter235 pts · 111 comments
Krea 2: SOTA open-weights 12B image model210 pts · 29 comments
A Practical Guide to SSH Tunnels: Local and Remote Port Forwarding179 pts · 36 comments
Voices & Blogs<br>★ Must-ReadSimon Willison: porting an image model to the browser with Claude CodeA real, blow-by-blow agentic-coding session, how he drove Claude Code end to end.
LearnSimon Willison: prompt injection as 'role confusion'New framing for the #1 agent security risk. Read before you ship tool-using agents.
★ Must-ReadAnthropic Engineering: building effective agentsThe canonical patterns doc, workflows vs. agents, when to use which.
LearnHamel Husain: evals skills for coding agentsHow to actually measure whether your coding agent is getting better.
Agentic infrastructure operations begin with accurate, reliable infrastructure dataThe New Stack
Sakana Fugu is more than a router. But it’s not the blueprint for AI sovereignty,...