schristoph.online

Weekly

The Sunday review

Every Sunday: the week's posts in one place, what connects them, and a few public reads I found worth your time. By email or RSS. Get it by email or via RSS.

2026

  1. Weekly Review ā€” Aug 3ā€“9, 2026

    Gall's Law and building from small bricks, self-consistency as probability amplification, and why a read-only AI assistant beats the ones racing to act.

  2. Weekly Review ā€” July 27-August 2, 2026

    Tesler's Law and who owns complexity, generate-and-verify through the lens of P vs NP, and why understanding is the new bottleneck when agents write the code.

  3. Weekly Review ā€” July 20-26, 2026

    Hyrum's Law and who pays for a broken API in the age of agents, the algorithmic race to beat O(nĀ²) attention, and benchmarking coding agents on your own tasks.

  4. Weekly Review ā€” July 13-19, 2026

    Conway's Law as your real architecture, why AI tokens cost what they do, and multi-agent lessons from robot soccer to a Bedrock world cup.

  5. Weekly Review ā€” July 6-12, 2026

    From the whiteboard to the AWS building block that turns the idea on: compression as intelligence, the agent-memory loop, and post-quantum crypto on KMS.

  6. Weekly Review ā€” June 29-July 5, 2026

    Where knowledge lives and who owns it: the agent-memory spectrum, vector search, vaults as graphs, and owning your work.

  7. Weekly Review ā€” June 22-28, 2026

    Five builds that each treat an AI agent as a first-class user, from scoped tokens and edge payments to per-agent microVMs and grounded search.

  8. Weekly Review ā€” June 15-21, 2026

    A three-part MCP series that turns AWS guidance into runnable, measured code and a real AWS deploy, plus running agents like a team and scoring voice-clone reference clips.

  9. Weekly Review ā€” June 8-14, 2026

    Cheap models writing agent harnesses, the pipeline's writing and editing stages, a cloned voice on SageMaker, Claude Fable 5, and a closed MCP gap.

  10. Weekly Review ā€” June 1-7, 2026

    Bounded-context skills, the ingestion and research stages of the AI content pipeline, an agent that pays, two models meeting on Bedrock, and why code makes skills reliable.