Get a New Map

Everyone in the software industry is doing AI wrong. Myself included.

We're treating AI like a faster compiler for the same ideas we've always had: ship more features, write more CRUD, reduce headcount, and squeeze efficiency out of familiar systems. The result? Marginal gains and a ceiling on ambition.

If you're an engineering leader, you've probably felt this tension already. Leadership asks, "Can we do this with fewer engineers now?" Meanwhile, your backlog is full of systems that should be rethought, but never quite justify the risk. AI becomes a cost-cutting tool instead of a force multiplier.

That's backwards.

The Pattern of Paradigm Shifts

Every real paradigm shift in software starts this way. When cloud computing arrived, we lifted-and-shifted servers before we invented entirely new architectures. When CMSs took off, we rebuilt brochure sites faster—then eventually changed how content teams and engineers collaborated altogether.

The first phase is always efficiency. The second phase is reinvention.

We're still stuck in phase one with AI.

The truly interesting work—the category-defining, ambitious projects—doesn't look like "doing the same roadmap faster." It looks like changing what the roadmap is.

What Category-Defining Looks Like

If you've ever said, "We'd never have the time or staff to build that," you're probably looking at the right kind of problem.

• They collapse entire workflows, not steps. Instead of "AI-assisted ticket triage," think "tickets no longer exist as a primary interface." AI is embedded into the system of record, not bolted onto it.
• They unlock previously uneconomical capabilities. Real-time compliance instead of periodic audits. Continuous migration instead of one-off rewrites. Personalized experiences at enterprise scale without bespoke teams.
• They start from constraints, not features. Regulated data. Legacy systems. Low tolerance for downtime. These aren't blockers—they're the raw material for defensible systems competitors can't copy quickly.

How to Find These Projects in Your Org

Here's practical advice you can act on:

• Audit the "heroics." Where do your best engineers save the business through manual effort, tribal knowledge, or one-off scripts? That's a signal of a system begging to exist.
• Look for long-running pain, not loud pain. The most ambitious projects are often quiet annoyances that have survived multiple planning cycles because they're cross-functional, risky, or hard to scope.
• Ask: "What would we build if we weren't afraid of touching that system?" Legacy Rails monoliths. Core financial pipelines. Healthcare integrations. These are exactly where AI-enabled leverage compounds.
• Fund experiments, not transformations. The goal isn't a 12-month rewrite. It's a focused, production-grade wedge that proves a new way of working—then earns the right to expand.

Where This Leads

If you're watching AI reshape the industry and thinking, "There's a bigger move we should be making but we can't afford to get it wrong," that's exactly the moment to act. The companies that use this window for ambition—not just efficiency—will define the next era of software.