Built But Unused Scores Zero
The hardest sentence in my job is "nobody uses it". Why good internal tools die, what actually makes them stick, and the one written confirmation that changed how my team ships.
Read Article →Insights
Notes on running enterprise AI in production.
I write about governing AI, measuring it honestly and getting business teams to use it. When a figure I have published turns out to be wrong, I correct it in the same place.
Audit-Ready AI: What ITGC and DPDP Actually Ask
Auditors do not need to understand transformers. They need to know who asked, what they were entitled to, and where it is written down. What an IT controls review actually asks, and the order DPDP work happens in.
Read Article →What Enterprise AI Actually Costs (And What It Delivers)
I removed the ROI figure this post used to lead with, because I could not reconstruct how it was calculated. The real cost lines of enterprise AI, the benefits you can actually prove, and how to build a case without inventing a number.
Read Article →What Actually Works When You Build AI Agents for Operations
Counting agents is the wrong scoreboard. What survived contact with production: a registry with named owners, autonomy in tiers, evaluations on your own data, and the interface where the work already happens.
Read Article →