Blog

    Notes on AI product management, n8n automation, LLM systems, and building useful tools.

    March 24, 2026

    The Great U-Turn: Why AI Walled Gardens Will Break the Legacy Web

    As major players like Anthropic and OpenAI build walled gardens and ban open-source clients, heavily subsidized Chinese models are poised to dominate the API market. Meanwhile, the internet faces a massive architectural U-turn. After years of building defenses to block bots, companies must now completely redesign their data layers to serve AI agents. The shift from SEO to Answer Engine Optimization (AEO) will leave legacy platforms behind, creating a massive opportunity for new players to build agent-first interfaces from the ground up.

    mcp
    open source ai
    walled gardens
    aeo
    geo
    agentic web
    March 4, 2026

    Code Is a Liability, Not an Asset

    A breakdown of Cory Doctorow's essay on why more code, especially AI-generated, is a threat. It covers the hidden dangers of treating code as an asset, the myth of stable software, and the critical difference between writing code and software engineering.

    cory doctorow
    code liability
    software engineering
    software
    February 26, 2026

    The Cold Truth of Static Intelligence: Why Intent Engineering is Failing

    A critical assessment of current AI limitations, focusing on the failure of 'intent engineering'. Despite scaling, LLMs remain static and lack the dynamic, deeply human value systems derived from experience. Using the Klarna AI case, this post argues that centralized LLMs are light-years away from understanding true, adaptable intent.

    intent engineering
    llm limitations
    ai ethics
    static models
    klarna
    agi
    February 10, 2026

    System Failure: The Great Convergence and the End of the Junior Role

    The traditional tech hierarchy is collapsing. AI agents like Claude Code and Lovable are merging the roles of PMs, designers, and engineers, effectively eliminating the entry point for juniors. We are entering a brief window where professionals must transition into AI operators before token costs lock the gates. Adapt now, or risk obsolescence.

    ai agents
    future of work
    claude code
    lovable
    tech convergence
    January 23, 2026

    The Human API: Your Only Defense Against AI's Corporate Takeover

    AI is eliminating pointless corporate meetings, but it demands a new skill: clear communication for AI transcription. Product managers must now become 'human APIs,' connecting business fundamentals with technical execution. As LLM costs rise and teams shrink, the non-technical PM is becoming obsolete. It's time to adapt by mastering the tools and the economics, or risk being replaced. The future of product management is technical, concise, and relentlessly focused on value.

    future of work
    product management
    ai adaptation
    technical skills
    llms
    January 19, 2026

    Digital Sovereignty Breach: Escaping the Telemetry Trap

    A journey into replacing Google-centric devices with Linux alternatives like the PinePhone reveals a harsh reality: we are losing ownership of our technology. This is a look at the forced telemetry in our phones, cars, and homes, and the near-impossible challenge of reclaiming our digital sovereignty in a world that profits from our data.

    self-hosting
    slinux phone
    pinephone
    privacy
    telemetry
    device ownership
    smart tech
    January 18, 2026

    The Content Creation Paradox: Why LLMs Need Us to Keep Writing

    We've reached a point in history where LLMs are consuming everything on the web. These new products-whether it's Google, Bing, or the various AI assistants-consume content that doesn't really belong to them. They act as middlemen, delivering results without you ever accessing the original source. They eliminate the need to visit the actual page, to read the original author's work, to see other people's contributions on platforms like Stack Overflow.

    llm content
    stackoverflow