Are We Ready?
Let’s go back to stage zero.
Before you deploy AI, what do we need to know?
What decisions actually get made, by whom, using what information?
Where does tribal knowledge live that’s never been documented?
What validation steps happen informally that would need to become explicit?
Which outputs require human judgment that AI can’t replicate?
I bet only a small number of companies actually start at stage zero with this reality check before they reduce headcount.
Two days ago, Amazon announced it’s cutting 16,000 corporate jobs. That brings the total to roughly 30,000 job cuts in the last three months. At the same time, entry-level hiring across tech has collapsed. Recent graduates are finding fewer doors open than any cohort in recent memory. Tech employment itself is shrinking—not slowing, shrinking.
This syncs with what Anthropic CEO Dario Amodei revealed recently: “AI is now writing the vast majority of Anthropic’s code... and may be only 1–2 years away from a point where the current generation of AI autonomously builds the next.”
The machines are building the machines.
Do we actually have a plan for ourselves—human?
What We Already Know
Set aside the predictions. Look at what’s already happening.
If you ask me what AI at Work really means? From what we’ve seen, it means cost-cutting. Eliminating jobs. Full stop.
The hiring freeze is real. Entry-level positions are disappearing. The first rungs of the career ladder are being pulled up.
The cuts are accelerating. AI-attributed layoffs made headlines throughout 2025. And that’s just the companies willing to say it out loud.
CEOs are stating it plainly. Salesforce cut thousands from customer support because AI handles the volume now. Ford’s CEO warned AI will replace half of white-collar workers. IBM’s CEO said he could easily see a third of back-office roles automated away.
These aren’t speculation. But are these strategic roadmaps? Or are executives just pressing the panic button?
Either way, the layoffs tell us what companies are doing. They don’t tell us whether companies know what they’re doing.
Cutting headcount is easy. Replacing what those people actually did—the judgment calls, the context, the informal validation—that’s the part nobody’s planned for. We don’t hear people talking about how they cooperate with AI. Just how they chat with it.
Which brings us to the real problem.
The Missing Pipeline In Your Workflow
I could tell you to prepare yourself. Learn to use AI. Augment your value. You’ve heard it before.
But here’s what I’m not seeing: actual workflow optimization. Plans that use AI to streamline how work gets done. Instead, AI shows up as an output tool. Generate this. Summarize that. Write me a draft.
AI is a powerful discovery tool. But it cannot discover without a data pipeline. And most organizations don’t have one.
Conversing with AI is not discovery. You’re not feeding it relevant data. You’re prompting it. A prompt is not data. Information embedded in prompt context is not data. It’s a request dressed up as input.
Discovery requires structure. It requires knowing what data exists, where it lives, and how it flows through your organization. It requires the stage zero work most companies skip.
We’ve Been Here Before
Most organizations failed at digital transformation. The reason? Fragmented data. Fragmented responsibility for managing the data pipeline. No one owned it. Everyone touched it. Nothing connected.
Managing data is difficult. It requires tools. Analytical skills. Governance. Patience.
These are the same ingredients we need to transform workflows in the AI era.
If you couldn’t build a coherent data pipeline for digital, what makes you think AI will be different? The technology changed. The problem didn’t.
This is spot-on about the missing data pipeline issue. Companies rush to deploy AI without maping the informal knowledge flows that actually make things work. I saw this at my last job where they automated a process but nobody documented the edge cases that the team just "knew" how to handle. The result was chaos untill someone manually intervened every time.