Blog
Notes from running AI in real businesses.
Case notes from live work, how we grade what we build, and plainly worded opinion. No launch posts.
Latest · Case notesA photo of the shelf is worth a morning's walkNotes from the shelf-intelligence system running at a supermarket chain in Thailand. The hard part wasn't the vision model. · 1 min readRead Bench · What a bench is, and why every tool gets oneA fixed set of real cases with a written right answer, re-run every month. It's not clever. That's the point.
Opinion · 8,212 projects missed one deadlineMahaRERA flagged thousands of projects for a missed quarterly report. Most of that report is paperwork that already exists.
Opinion · Pick the boring workflow firstThe workflow that annoys the founder most is usually the worst first project. The one nobody talks about is usually the best.
Research · Evals that survive a model swapMost eval suites test the prompt, not the job. When the model changes underneath you, the difference shows up the same week.
Opinion · 8,212 projects missed one deadlineMahaRERA flagged thousands of projects for a missed quarterly report. Most of that report is paperwork that already exists.Opinion · Pick the boring workflow firstThe workflow that annoys the founder most is usually the worst first project. The one nobody talks about is usually the best.Research · Evals that survive a model swapMost eval suites test the prompt, not the job. When the model changes underneath you, the difference shows up the same week.Tell us the job your team hates. We'll tell you if AI can do it.
A 30-minute call with the people who would build it. No technical knowledge needed: just describe the job, how often it happens, and how you check it's done right.