If you live in AI conference rooms and tech Twitter, you hear one narrative on repeat: AI is about to wipe out huge swaths of jobs. In my conversation with David Cost, Chief Digital Officer at Rainbow Apparel, he offered a very different, data‑driven view from the operator’s seat.
David doesn’t see an AI‑driven job apocalypse. What he does see is an extremely powerful augmentation tool that, when used well, can transform organizations without gutting headcount.
Replace vs. Augment: Two Very Different Outcomes
David splits real‑world AI projects into two buckets:
-
- Replace people: Build a robot to do what a human does and reduce headcount.
- Augment people: Use AI to give humans superpowers so they can do more, better, faster.
His track record so far is stark:
-
- Replacement projects have “failed and failed miserably.”
- Augmentation projects have been “huge successes.”
Take customer service. Vendors promised that voice AI could handle ~40% of calls to completion. With 12 years of human call‑center data, Rainbow could measure impact precisely.
What happened? After rolling out AI:
-
- The hours worked by human agents and
- The number of cases they touched
barely moved. If you saw the before‑and‑after graphs, David says, you’d have no idea any automation had been deployed. The likely story: customers weren’t satisfied, didn’t know how to reach a human, hung up, and the AI still marked those calls as “completed.”
In product data, Rainbow hoped AI could auto‑create most attributes for the 1,500 SKUs they launch daily. The AI could draft 70–80% of a record, but humans still had to review, correct, and fill gaps. Net effect: maybe 10–20% productivity gain, not enough to credibly cut staff once AI costs are included.
The pattern: AI helps, often meaningfully, but not enough to remove people from the loop.
What CEOs Are Actually Planning
The broader leadership data backs this up. David cites a KPMG survey of 500 CEOs on AI’s impact on headcount:
-
- Only about 10% planned to decrease headcount.
- Roughly 45% expected to increase it.
- Around 55% planned to keep it the same.
If AI were truly triggering a near‑term labor apocalypse, you’d expect that mix to look very different. David’s challenge to the hype is simple: if you believe AI is wiping out jobs, show concrete examples where headcount was reduced, and prove AI was the driver.
Even in Coding, It’s “Build More,” Not “Hire Less”
Coding is the one area where nearly everyone agrees AI is spectacular. I’ve personally seen massive productivity gains in web development. David agrees: AI produces and refactors code at remarkable speed.
But his observation is telling:
“Most organizations that use it hire more developers.”
Once you can build faster, you don’t decide to build less: you decide to build more. You still need skilled engineers to decide what to build, architect systems, prompt and guide the AI, and review and maintain what it generates. Again, AI is augmentation, not replacement.
The Real Risk: Data, IP, and Governance
Where David does see serious near‑term risk is in data, privacy, and IP exposure, especially through cheap, consumer‑grade AI tools.
He shared a striking example: a CEO of a financial firm took attorney work product, pasted it into a $20/month chat account, and broke attorney–client privilege. Regulators were able to subpoena:
-
- Everything he’d entered, and
- His attorneys’ underlying work product.
Many people assume their $20/month chat account is private. It isn’t.
The same goes for trade secrets. Paste a formula, recipe, or confidential design into a non‑enterprise AI, and you may have effectively publicly disclosed it, undermining your ability to secure patents or copyrights.
David’s conclusion: companies will need to block free and $20/month AI tools on corporate networks and move to enterprise‑grade AI platforms with clear privacy, security, and legal protections.
So What Should Leaders Do?
For all his skepticism about job cuts and hype, David is emphatic: this is not the time to sit out AI.
He likens AI to spreadsheets:
-
- Spreadsheets didn’t eliminate accountants;
- They made the people who mastered them early, and their companies, far more competitive.
AI, he argues, is “spreadsheets times ten.” But success is less about which model you choose and more about organizational design:
-
- Rethinking roles, incentives, workflows, and decision‑making;
- Treating the company as an information‑processing system and redeploying AI where it meaningfully improves decisions.
Across past tech waves, personal computing, the internet, mobile, social, David has seen one pattern: late adopters never fully catch up.
So if there’s a myth to retire, it’s not that AI is transformative. It’s that transformation must equal a job apocalypse. The more useful (and actionable) story is that AI will amplify the organizations and people who lean in thoughtfully, starting now.
Discussion points include:
- Why most real‑world AI wins come from augmenting people, not replacing them
- What actually happened when Rainbow tried to automate customer service and product data with AI
- How CEOs really expect AI to affect headcount over the next few years
- Why even in coding, AI leads companies to build more, not hire fewer developers
- The overlooked risks around data privacy, legal exposure, and IP with $20 AI tools
- How to think about AI as “spreadsheets times ten” for organizational design and strategy
- Why early, thoughtful adopters of AI may gain a lasting competitive advantage