The Age of Agent-Washing: Selling Step 10 to Companies Still on Step 2

A telling example: a World Economic Forum speaker relayed an “overheard” case study of a company building an entirely new business line with agents, working 24/7 on Slack. There was no mention of which company this was, how they got these results, or any critical questions about the role humans played in this.

AWS and others promised “up to 80% productivity gains.” Pressed on how that squares with the modest-to-poor ROI many report, the evidence was thin: a few high-level anecdotes limited to narrow functions like customer service, not the enterprise.

In short, agent-washing was in full effect, the idea that 10x revenue is a toggle away, when the “agents” most companies deploy today are often a souped-up deterministic workflow.

(Article continues after this break)

In our ​executive AI boot camp​, I quote Andy Wu, Professor, Harvard Business School, who said in HBR that leaders “need to chill out and panic at the same time.”

My build: feel the urgency, but be methodical.

Don’t rush into vendor commitments before you’ve nailed the use case.

As Booking.com’s head of AI, Pranav Pathak, told me, there’s a common tendency to believe AI is magical. Vendors promise immense results, and clients pursue deals before figuring out how to make the technology work.

Especially in agentic AI, where ​vendor-driven confusing terminology​ is making everything feel very overwhelming.

Yet, the recipe remains very simple:

1. Understand the pain point, use case, or opportunity
2. Build a proof of concept
3. Iterate, learn, and scale up what works

Today, we have solutions looking for problems.

Let’s reverse that.

As Character AI’s CEO, Karandeep Anand, said, the hardest part of thriving in the AI era is the speed at which it moves. I agree fully.

But as I said last week, ​focus beats frenzy​ whenever it comes to AI.

Rather than ‘hoping for magic’, let’s focus on what’s already possible today: start with a concrete pain, prove a small win, and show movement on at least one business-critical metric.

Then, we’ll talk truly agentic.

Until next week,

- ​Daan