Notes from
the workshop.
Essays, post-mortems, and field notes from consulting engagements and product work. Long-form when the topic deserves it. Subscribe by email — see the footer.
Your team has AI tools. They're not using them.
Most organizations have paid for Copilot or given the team Claude access. Adoption is shallow because nobody has done the work that turns a model into a workflow. Four examples of the depth gap, and what to do about it.
The automation gap in Indian manufacturing
Tier-1 and tier-2 plants are sitting on years of sensor data, running CNCs and SCADA systems that emit hundreds of signals a second, while quality engineers inspect castings with a magnifier. The gap is not capital. It is analytical workflow.
How to evaluate whether your business problem needs ML or just better SQL
Most projects pitched to us as AI engagements are really SQL problems with a logistic regression on top. A four-question framework for telling the difference, with worked examples — and the discipline of giving the smaller answer when it is the right one.