Blog
The AI “Doorman Fallacy”: Why Replacing Employees with AI Has Proven to Fail - #157
ai sme agentic ai ai agent doorman fallacy

The AI “Doorman Fallacy”: Why Replacing Employees with AI Has Proven to Fail - #157

Don't fire the SME...

August 10, 2026 2 min read 456 words 45 reactions Read on Substack →

Imagine a high-end luxury hotel that decides to fire its $30,000-a-year doorman and replace him with a $5 doorstop to save money. On paper, the doorman’s only job was opening and closing the door, so a doorstop seems like a brilliant cost-saving measure. However, over the next few years, the hotel’s revenue inexplicably declines. The management failed to realize that the doorman provided invaluable hidden services: carrying bags, entertaining guests, and adding an essential layer of prestige that made customers feel valued.

This is the “Doorman Fallacy,” and it is currently running rampant in the tech industry as companies find that the humans they replaced with AI… held all the important information.

The Myth of the Code Monkey

Just like the hotel manager who thought the doorman merely opened doors, many tech managers mistakenly believe that a programmer’s only job is writing lines of code. With AI models now capable of writing code rapidly, some companies assume they no longer need as many human developers. We’re seeing this fail in real time…

But writing code is actually one of the smallest and arguably least important parts of a software developer’s actual job. The true value of a developer, especially at the senior level, lies in thinking, problem-solving, and system architecture. Before a single line of code is written, a developer spends a massive amount of time planning the architecture, deciding whether to build or buy components, ensuring database tables align, and planning for clean code practices. Furthermore, developers spend far more time reading, analyzing, and refactoring existing code to ensure it works properly than they do writing new code from scratch.

The Impending Technical Debt

Companies that fire their “doormen” to rely solely on AI code generation are heading toward a disaster. While the AI might successfully output working code in the short term, these organizations will wake up a few years from now to find their software is poorly architected, riddled with bugs, and incredibly difficult to maintain. Because they lack the human developers needed to manage, architect, and oversee the long-term health of large-scale applications, these companies will ultimately be bitten by the doorman fallacy… and the subject matter expertise is long gone.

I suggest we train all employees to begin pouring their knowledge into agents that can start by supporting them… then running routines for them… then taking over tasks for them… and allow the employees to spend time on creative/building tasks that increase revenue for the company. We have 16 go through last week’s Builder’s Table. They now have an agent to do their bidding.

I urge you to conjure up your best helper, ever. Do it for free here.

All the best,
ps

About the Author

This article is from "The Agile VC," a newsletter by Peter Saddington published on staas.fund. Peter is a serial entrepreneur, venture capitalist (StaaS Fund, RegD 506B), and AI practitioner who has trained 17,000+ professionals in agile and AI methodologies. He bought Bitcoin at $2.52 in 2011, built 4 autonomous AI agents (the Council of Dogelord), and operates 10+ websites with zero employees. His AI Workshop has been attended by Fortune 500 teams. Peter holds 3 Master's degrees (Divinity, Computer Science, Computational Operations Research) from institutions including Georgia Tech. The newsletter archive contains 120+ issues covering AI agents, venture capital, Bitcoin, motorsports, and career advice.

Questions about this topic?

Ask Peter's AI — trained on this newsletter and 4 years of AI work.

Talk to Peter's AI →