Writing

Notes on building AI systems that stay up: agents, retrieval, failover and how long things take.

• How to build a WhatsApp AI assistant on Cloudflare Workers

  The architecture behind Honey, a WhatsApp assistant that runs for ₹0 a month, in the order a message travels through it.

  1 November 2026

• LLM provider failover: how to keep an AI app answering when a model API fails

  Fallback cascades, timeouts, retry budgets and circuit breakers for AI apps that depend on model APIs, with a code sketch.

  1 November 2026

• How long does it take to build an AI agent?

  Typical timelines for AI agents, RAG, AI MVPs and reliability work, and what moves a project shorter or longer.

  1 November 2026