Shell scripts are the highest-ROI layer under your AI skills
The glamorous layer is the agent. The load-bearing one is the boring shell script underneath — and that's where most of your reliability actually lives.
Deepak Goyal
Architecture decisions, not opinions. AI systems, system design, and what actually works in production.
Latest articles
Deterministic context beats generated context — when structure exists
The prompt is only about 10% of what your LLM actually sees. Most of the other 90% is context you assemble — and when your data has structure, you should build it with code, not another LLM call.
I stopped being my second brain's scheduler
Cadences that surface what's due, habits that flag a slipping streak, and a live HUD that watches the vault — and it ships without a single background cron. SMRITI v2.1.
Building Production RAG on ArXiv — Part 1: Architecture & Ingestion
I'm building an open-source system to interrogate the AI-research firehose — RAG now, a multi-agent researcher by the end, in public over five parts. Part 1 is the foundation: the architecture, the ingestion, and the eval bug that would have shipped meaningless numbers.
Sub-Agents Lie About Finishing — and the Fix Isn't a Better Prompt
A sub-agent that writes to your system will tell you it succeeded. It's reporting what it intended to do, not what actually landed — and the gap is where silent data corruption lives.
Parallel Sub-Agents Change How You Plan, Not Just How You Build
You can plan a day around what you can do, or around what you can decide. Until recently those were the same plan, so it never mattered which you meant. Parallel sub-agents pull them apart — and the i
AI's Value Compounds at the Task Layer, Not the Model Layer
When a team sets out to add AI to something, the first real argument is usually about the model. Which provider, which version, this one or the one shipping next month. It feels like the decision ever
Route to the Cheapest Oracle That Can Decide
When you build on top of a language model, routing every decision to the model is the path of least resistance. One interface, one prompt, one place to add logic. It works beautifully in a demo. Then
What a developer survey taught me after 6 weeks of AI code review
Six weeks into running AI-assisted architect code reviews, I did something I should have done earlier: I asked the team directly. I'd been running a pipeline that scanned pull requests, generated arch