Your organization prioritized AI adoption, but you actually need AI fluency.
The New Stack Manu Narayan
Opinion — commentary, not a factual news event.
Your teams may have AI, but not the know-how to use it well. That gap is slowing results, even where the tools are already in place.
Based on reporting by The New Stack, Manu Narayan — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Some teams are already using AI to move faster, but the rest of the organization is still stuck waiting for approvals, rollout plans, or someone to explain where to begin. That split is becoming the real problem. The issue is no longer just getting access to the tools. It’s getting people to use them well enough that the work actually changes.
Harvard Business School found workers using these tools finished tasks 25% faster and produced results rated more than 40% better in quality. But there’s a catch: performance got worse when people used the tools without understanding where they fit and where they don’t. That’s the article’s central point, and it’s a useful one. Access helps, but fluency is what turns a shiny license into a better workflow.
A lot of companies respond to uneven adoption by handing out the same platform, offering broad training, and hiring a few specialists for IT. That sounds tidy. It also misses the bigger opportunity. The real work is often not about dropping AI into an existing process. It’s about stepping back and redesigning the process itself. The example given is sales lead routing: a team asks IT to improve it, but a broader look shows the data pipeline is more complex than it should be, which opens the door to overhauling the whole thing and even moving toward fully agentic lead follow-ups.
The proposed operating model reflects that shift. Instead of a central team taking requests and delivering them months later, the article argues for a central hub that handles platform strategy, governance, and reusable patterns, plus AI engineers embedded inside departments as spokes. Those engineers would scout use cases, build prototypes, and share patterns across functions like finance and operations. As departments mature, they’d rely less on those engineers for direct building and more for enablement. The goal is simple: make AI fluency part of how the business works, not a side project with a Slack channel.
McKinsey’s finding in the piece pushes the same direction: workflow redesign is the single biggest factor in whether a company sees meaningful bottom-line impact. That means the first question isn’t which model to buy. It’s which parts of the organization can actually think with AI instead of just around it.
My take — AI-written commentary, not fact-checked reporting
This is the part companies keep missing: AI adoption without AI fluency is just expensive clutter with better branding. The central-hub-plus-embedded-engineers model makes sense because the old command-and-control setup is too slow for tools that change every week. The future belongs to the teams that can redesign work, not the ones that collect licenses like office plants.
Read more about this at: The New Stack