Blog

    Till's Block 🏢

    Insights, trends and thoughts from the world of digital transformation.

    Nerd Level
    359 articles found, page 1 of 24
    Winged sandal-sneaker with headphones – abstract Hermes logo next to a red lobster figurine
    Till FreitagTill Freitag1. Juli 20265 min

    Hermes: TillClawn 🤡 Got a Brother – and It's Not a Parcel Service

    Next to TillClawn 🤡, there's a second building block in our stack as of today: Hermes. Not a parcel service, not another agent layer – a lean framework that ca…

    Read more
    Stylized number 5 made of orange ribbons and gears – cover for Claude Sonnet 5Deep Dive
    Till FreitagTill Freitag30. Juni 20269 min

    Claude Sonnet 5: Agentic AI Goes Mainstream

    Anthropic ships Claude Sonnet 5 – a Sonnet model that gets close to Opus 4.8 performance at a fraction of the price. August 2026 update: how Claude Opus 5 (July…

    Read more
    Map of Europe drawn in glowing circuit lines, overloaded GPU nodes
    Till FreitagTill Freitag30. Juni 20263 min

    Europe's Compute Crunch: Why There's Barely Any Capacity Left in 2026

    B200 clusters with months of lead time, sold-out EU regions, hyperscalers openly talking 'capacity constraints': Europe has a compute problem – and it will get …

    Read more
    Sakana Fugu – a conductor orchestrating multiple specialised AI models
    Till FreitagTill Freitag29. Juni 20265 min

    Sakana Fugu: Orchestrator, Not Monolith

    Sakana AI ships Fugu – not another foundation model, but an orchestrator that routes prompts across specialised LLMs. The strategic answer to vendor lock-in, ex…

    Read more
    Abstract isometric illustration of a glowing exoskeleton frame cradling multiple small AI cores inside sandbox chambers
    Philip SeekerPhilip Seeker26. Juni 20264 min

    Agent Harness as a Category: Why the Harness Is the New Product

    The harness — skills, sandbox, subagents, memory, channel routing — is the real product layer above the model. A map of who plays where: Claude Code, DeerFlow, …

    Read more
    AI Benchmarks Explained: Arena, SWE-Bench, AutomationBench & Co.
    Till FreitagTill Freitag26. Juni 20266 min

    AI Benchmarks Explained: Arena, SWE-Bench, AutomationBench & Co.

    How do AI benchmarks actually work – from LMArena to SWE-Bench to Zapier's AutomationBench? A tour of Elo rankings, static evals, agent benchmarks, and their li…

    Read more
    NotebookLM 3.5: From Preprocessor to Agentic Research Partner
    Philip SeekerPhilip Seeker26. Juni 20264 min

    NotebookLM 3.5: From Preprocessor to Agentic Research Partner

    Google upgraded NotebookLM to Gemini 3.5, added a secure cloud computer for autonomous web research, and plugged in Antigravity agents. The source reader has be…

    Read more
    AutoClaw by Z.ai – AI agent that turns a chat message into real executed work
    Till FreitagTill Freitag25. Juni 20265 min

    AutoClaw: Z.ai's Agent That Doesn't Chat – It Works

    AutoClaw is Z.ai's agent layer – not a chatbot, but an AI partner that actually executes tasks. What it does, how it teams up with our Agent Cluster, and where …

    Read more
    Serponado SEO Audit Report cover
    Till FreitagTill Freitag25. Juni 20262 min

    Serponado SEO Audit: What 8 Cluster Articles Reveal About Internal Links, Canonicals and Fragments

    Short SEO audit across all 8 Serponado cluster articles (DE+EN): link status, canonicals, fragment matches – with the full report as download.

    Read more
    Z.ai GLM-5.2 – performance comparison against Claude and Western frontier models
    Till FreitagTill Freitag25. Juni 20265 min

    Z.ai GLM-5.2 – why the next frontier leap comes from China

    Z.ai ships GLM-5.2 – more than doubling GLM-5.1, close to Claude Fable 5, cheaper, deployable on-prem. While US frontier models get shut down, China is quietly …

    Read more
    Local AI on a laptop – notetaker, LLM, privacy shield
    Till FreitagTill Freitag24. Juni 20264 min

    Is Local AI Killing the AI-SaaS Startups? An Honest View From the Engine Room

    Meetily takes meeting notes fully locally. Qwen runs on a laptop. RTX Spark hits notebooks in 2026. Are all the AI-wrapper startups about to die – or is this ju…

    Read more
    GLM-5.2 vs. Kimi K2.7 Code – split-screen illustration with Z-letter mark and crescent moon symbol
    Till FreitagTill Freitag21. Juni 20267 min

    GLM-5.2 vs. Kimi K2.7 Code: Two Open-Weight Releases in One Week – Two Very Different Bets

    Within four days in June 2026, Z.ai (GLM-5.2) and Moonshot AI (Kimi K2.7 Code) shipped their next-generation open-weight models. On the surface, similarly price…

    Read more
    Abstract UI cards with rocket, chat bubble, database and cursor – visual metaphor for the Lovable Feature Roundup May/June 2026Deep Dive
    Till FreitagTill Freitag21. Juni 20268 min

    Lovable Feature Roundup: What actually mattered in May and June 2026

    Subagents, native Claude MCP, the Preview Toolbar, Publish-from-chat, slow-query analysis in Lovable Cloud: in six weeks Lovable shipped more than some platform…

    Read more
    Split visual: classic Google search results on the left, an AI answer card with source citations on the rightDeep Dive
    Till FreitagTill Freitag18. Juni 20268 min

    SEO vs. AEO – aren't they the same thing?

    SEO and AEO sound like the same game with a different acronym. They're not. Where the two disciplines overlap, where they diverge — and what to do today to be v…

    Read more
    Isometric blueprint diagram: Antigravity orchestrator coordinating specialized worker agents in a pipeline
    Philip SeekerPhilip Seeker17. Juni 20265 min

    Antigravity in Practice: Multi-Agent Pipelines for Mid-Market Clients

    Antigravity moves multi-agent pipelines from lab toy to production tool. A hands-on architecture walkthrough from active mid-market projects — with setup, trade…

    Read more