Abdul Hannan ShaikhSoch.Blog by Abdul Hannan Shaikh

Every blog

Blogs on shipping AI features without guessing, newest first. Filter by pillar, or search for the thing you are stuck on.

17 blogs

  1. Agent-readiness
    MCP's Second Spec: Reopening the LedgerTwo gaps named last October. One closed by leaving the protocol, one has not moved, and the registry existed all along.12 min read
  2. Building in public
    What I Got Wrong About llms.txt, Fourteen Months OnNinety-seven per cent of published files got zero requests. The prediction held; the reason I gave for it did not.12 min read
  3. AI product management
    MCP Server Security: The Questions a PM Should Ask Before Shipping OneTwelve questions, each tied to a named incident. Only four MCP attacks have happened in the wild, and three were supply chain.12 min read
  4. Agent-readiness
    What Makes an API AI-Agent-Ready in 2026: 10 Dimensions, One Free ToolTen dimensions. Five of them you can score for free, and where the grader still gets it wrong.12 min read
  5. PM craft
    What Product Managers Actually Do: Beyond the Job DescriptionWhere the hours actually go, and why the job is mostly deciding what not to build.10 min read
  6. PM craft
    The Speed Paradox: Why Quick Decision-Making Separates Thriving Startups from the RestWhy a slow team cannot feel itself losing, and what the compounding actually costs.6 min read
  7. Agent-readiness
    GEO is the New SEO: How to Make Your Website Discoverable by AIWhat llms.txt actually does, and why assistants keep recommending somebody else's product.6 min read
  8. Agent-readiness
    One Year of MCP: What's Working, What Isn'tThe protocol was the easy part. Discovery and enterprise auth are still unsolved.8 min read
  9. AI product management
    Eval Design for PMs Who Aren't EngineersHow to build a rubric that survives a real spec review, without writing a line of pytest.5 min read
  10. Building in public
    Shipping llms.txt at Scale: What I LearnedSix months across every customer deployment. What surprised us, and what we got wrong.5 min read
  11. PM craft
    What 'Senior PM' Actually Means (It's Not Years)The scope of decisions you can own without escalating. Years are a leaky proxy.3 min read
  12. AI product management
    The PRD Format I Use for AI FeaturesFour sections a standard PRD doesn't have: rubric, hallucination tolerance, cost ceiling, fallback.5 min read
  13. Agent-readiness
    API-First Is a Product Decision, Not a Tech OneWhy the API belongs on the roadmap, and the ceiling it puts on the product when it doesn't.4 min read
  14. AI product management
    DeepSeek-R1 and the Quiet Commoditization of AINot “is the model good” but what changes for the AI features you’ve already shipped.5 min read
  15. PM craft
    Talking to Users: The Mistake That Killed Our RoadmapThe interview trap that hijacks roadmaps, and the discipline that stops it.4 min read
  16. Agent-readiness
    What I Look For in a Great API Spec (As a PM)The short list worth scanning, because these failures show up in adoption numbers you own.5 min read
  17. PM craft
    Why Most PMs Don't Need More FrameworksWhat the best PMs reach for when RICE, ICE and MoSCoW aren't the answer.4 min read