Archive - Agentic AI in Production

Correct by design: the video edition

Edition 016 in 24 minutes: how to trust code your agents write and nobody reads in full.

Nov 11 • Surendra Singh

24:31

Correct by design: software correctness and supply chain risk, from Sean Allen’s work

Agents write more code than anyone can read. Build correctness in before the code exists, and know which risks you cannot test away.

Nov 10 • Surendra Singh

How we repeated MIT’s weight-space experiments

Every step from prompt to score, including what went wrong.

Nov 5 • Surendra Singh

5:17

AI models just became something you can repair

MIT researchers changed 0.016% of a model with no retraining. Its math score went from 7.9% to 48.1%.

Nov 4 • Surendra Singh

2:25

How building software is changing: two eras of shipping

Coding agents can multiply what one engineer ships. Then the bottleneck moves: to review, to CI, to source control, and finally to the engineer.

Nov 3 • Surendra Singh

#nojs-banner { position: fixed; bottom: 0; left: 0; padding: 16px 16px 16px 32px; width: 100%; box-sizing: border-box; background: red; color: white; font-family: -apple-system, "Segoe UI", Roboto, Helvetica, Arial, sans-serif, "Apple Color Emoji", "Segoe UI Emoji", "Segoe UI Symbol"; font-size: 13px; line-height: 13px; } #nojs-banner a { color: inherit; text-decoration: underline; } This site requires JavaScript to run correctly. Please turn on JavaScript or unblock scripts