This area is owner-only.

Sign in with the badge in the corner to continue, or head back to the site.

Writing

Essays and articles on enterprise architecture, AI, and the future of practice. RSS.

Strategy

  • Sajiv Francis editorial card

    August 25, 2026 6 min read

    The Next Leaders Won't Ask for Status

    AI can watch a program the way instruments watch an aircraft. When status becomes telemetry instead of testimony, decision-making stops being reactive.

    Status reporting is a human-powered telemetry system with lossy sensors and political filters. AI-generated live program status changes the physics: leaders see variance as it forms and decide while options are still cheap.

    • AI
    • Leadership
    • Program Delivery
  • Sajiv Francis editorial card

    June 7, 2026 7 min read

    AI Transformation Is a Leadership Problem First

    An executive brief from Stanford's AI-Driven Leadership program

    Generative, analytical, and agentic AI each fail differently in the enterprise, and the failures are organizational before they are technical. Notes and frameworks from Stanford's AI-Driven Leadership program.

    • AI
    • Leadership
  • Sajiv Francis editorial card

    May 23, 2026 2 min read

    The AI Economic Trap

    How Corporate Efficiency Could Bankrupt the State

    The AI economy isn't short on wealth, it's short on the architecture to distribute it before the systems built to fund society quietly break.

    • AI
  • Sajiv Francis editorial card

    May 22, 2026 1 min read

    Your Agents Are Only as Good as Your Data Map

    What a viral consulting ad gets right about the AI conversation inside every enterprise

    Agentic AI is only as effective as the data it reasons over, and in most organizations, nobody holds the map of which process produces which data. AI won't fix that, but it's a great forcing function.

    • AI

AI

  • Word-cloud banner of the post's themes: video, crew, hand-off, Librarian, Filmmaker

    June 15, 2026 5 min read

    MCP Is a Librarian, A2A Is a Relay Team

    What building a five-agent video pipeline taught me about how AI systems coordinate, and why the pacing gap between the two protocols is the whole point.

    A small crew of AI agents fetches a page's words, writes a scene, films and publishes it. Here's how it works, why one step is slow and the mistakes that cost me a day.

    • AI
    • Agents
    • A2A
    • MCP
    • Architecture
    • Veo
  • Sajiv Francis editorial card

    June 13, 2026 6 min read

    A Grounded Personal Chat: The Two-Layer RAG Architecture

    Why I moved away from full-document ingestion and built a retrieval system with a library for readers and a worker for the LLM.

    Build of my personal chat interface

    • rag
    • architecture
    • personal-chat
    • pgvector
    • cloud-flare

Software Engineering

  • Word art from the voice mode post: voice, avatar, chat, Whisper, TTS

    August 17, 2026 5 min read

    Voice Is a Mode, Not a Bot

    Adding speech and a live avatar to a grounded chat without changing the retrieval stack

    A speech-to-text & text-to-speech roundtrip on my personal RAG chat: Grok-first audio with Cloudflare Workers AI and self-hosted fallbacks.

    • AI
    • Voice
    • Cloudflare Workers AI
    • xAI Grok