/pilot purgatory: why your ai initiative isn't showing roi

/pilot purgatory: why your ai initiative isn't showing roi

Leadership mandated AI. Now they want the return — and your pilots have nothing to show for it. James sits down with Bob White, a 20-year ed-tech and SaaS operator (Cengage, Houghton Mifflin Harcourt, VPG) who now advises companies on AI adoption, for an honest teardown of why most enterprise AI never pays off: escaping "pilot purgatory" with a problem-first funnel, why nobody being on the hook is the real reason ROI never lands, the financial giant forcing employees to hit AI-usage quotas (and why it backfires), and why bolting AI onto a broken process just makes it faster and more expensive.

/hotfix: the week ai got real

Five AI stories from the same week. One thesis: AI just moved from pilot to production. The first /hotfix from The Forward Slash — a new short-form read from James Carman on the AI news engineering leaders actually need to track. ~20 minutes covering EY and Microsoft's billion-dollar forward-deployed engineer concession, the 83/29 agentic AI security gap, Anthropic overtaking OpenAI on enterprise revenue, Kubernetes winning the AI runtime war, and the VS Code extension that walked off with 3,800 GitHub repos.

/apocaloptimist: ai leadership in 2026

Is your AI strategy replacing people, or amplifying them? James sits down with Mike Hruska — founder of Baryons, former NIST material scientist, and TEDx speaker on what magic teaches leaders — for a human-first tour of AI leadership in 2026: the two kinds of executives this market is creating, why high-performing teams shrink to two or three when AI joins the room, what AIQ — Authentic Intelligence Quotient — actually measures, and what "apocaloptimist" leadership looks like when the calculus has changed under your feet.

/ai agents: more steak, less sizzle

* Is your enterprise AI actively shipping, or just sitting pretty in a deck? James sits down with John Capobianco, head of AI and Developer Relations at Itential and former senior network architect for Canada's House of Commons, for an engineering leader's tour of what's actually shipping in 2026: spec-driven development with Ralph loops, MCP as the next generational protocol, agents built like org charts instead of Swiss Army knives, and the open-source agent security stack. 21 minutes for engineering leaders ready to ship.

/agent guardrails: you can't blame the bot

The hardest question in enterprise AI isn't "can we build agents?" — it's "how do we let them act on our behalf without losing the company?" In Part 2 of our conversation with John Willis, co-author of The DevOps Handbook, James gets concrete about what agent guardrails actually look like in regulated environments: delegated authority, read/write/execute permissions, kill switches, and the Andon Cord for AI. Twenty-one minutes for engineering leaders who've been told to "go faster on AI" without a plan for what happens when an agent misbehaves.