Writing
Notes on answer-engine distribution, structured agent pipelines, founder unit math, and distributed systems.
Answer Engines Read Tables, Not Taglines
A prose hero section tells a human buyer why your product feels modern. ChatGPT, Claude, and Perplexity skip the adjectives and cite whichever competitor published a pricing table and a plain spec sheet.
Why Structured Diffs Beat Raw Context for Production Agents
Point an agent at 120 KB of raw HTML, stack traces, or filings and it spends half its budget guessing which line matters. Hand it a typed 12 KB diff and reasoning becomes the easy part.
The Fixed Floor Problem: Buy vs. Build for AI Data Pipelines
Rolling your own ingestion or evaluation runner looks like a weekend script on day one. The real bill arrives in month three — as a fixed engineering floor you owe whether you pull 5,000 rows or 500,000.
The 280ms Ocean Tax: Running Realtime AI Workflows from Bengaluru
Most AI voice and browser demos are benchmarked five miles from `us-west-2`. Run the exact same stack from Bengaluru with mixed Hinglish and code identifiers, and physics exposes every lazy architectural assumption.
Stop Mocking Your Own Boundaries: Hermetic Tests and Personal APIs
Function mocks only prove your code calls your mocks; shared staging clusters make you wait 25 minutes to learn a queue flaked. Keeping real handlers intact behind deterministic state boundaries solves both — at work and on this site.