AI in Business: An Assistant, Not a Decision-Maker

I am PM at 3 AM — Tue, 10 Nov 2026 19:37:44 +0000

Why delegating decisions to AI "Mr. B style" leads businesses to collapse - and where neural networks are actually useful

I’ve been using AI to optimize and automate workflows for over a year now. During this time, I’ve come to only one conclusion - and it’s not about a “wow, everything is automated” epiphany, but about the boundaries of its applicability.

What AI Does Perfectly

As a search engine, it’s simply flawless. As an analytical summary tool - flawless. As a calculator - it works, but only if the specific parameters are set with absolute precision: it doesn’t know how to guess.

That is where its strengths end. Beyond that point, business begins - and things get a lot more complicated.

Why AI Is Not Enough for Business

Let’s take a baseline human model: a 40-year-old brain with 20 years of management experience. Even they don’t always draw the exact same conclusions as another person on the ground, because AI doesn’t interact face-to-face with every single participant in the process. It doesn’t understand how each individual employee behaves under a manager who is making the calls.

And if the manager themselves doesn’t understand their subordinates - doesn’t know how they operate - and on top of that plugs in an AI, it only adds more chaos and token expenses rather than actual results.

Neural Networks (and LLMs, etc.) Are Not Business Management Tools

It’s important to establish this clearly: neural networks (including LLMs and other language models) are not business management tools. They are knowledge and Q&A systems bounded by uploaded data, its markup, and the configuration of agents and algorithms. This fact alone carries inherent risks for a business in terms of the accuracy of the conclusions generated.

The system only works if the entire setup consists of LLMs simply running your business without your input. But you cannot execute a human vision or concept through that approach.

Who AI Is Actually Good For

It’s great for those who don’t quite know what they want, but need a direction to execute. For learning, researching, and getting quick prompts like: “How do I do this in that program - oh, like this, like this, and like this, but fast.” It will pull the info, organize it neatly, and allow you to learn effectively.

Where the Line is Drawn

AI is a helpful tool for a sales manager, marketer, project manager, operations manager, or middle manager. But it is not for a CEO or C-level executive who decides where the business goes next.

It is a process optimizer and organizational assistant, not a decision-maker.

However, if you want to grow a business and maintain your own strategic vision - AI is not your helper.

The “Mr. B Style” Collapse

“Mr. B” is a conditional archetype for a somewhat smart leader who ambitiously uses AI to make decisions. Many leaders following the “Mr. B style” have decided to delegate not just analytics to AI, but actual decision-making based on nothing other than the AI’s subjective opinion. It has reached an absurd level.

Today, when qualified specialists are being laid off left and right, AI has stopped meeting real business needs. Yet it keeps doing things fast, without weekends, without getting sick, or taking vacations - and when this is implemented without extra analysis, research, and experience, the business slowly starts to wither away.

The Other Side of the Coin: AI Degradation

AI itself is degrading because it lacks a fresh, foundational base to learn from efficiently. Most people have stopped creating original research, deep analytics, and new methodologies. Everything is becoming flooded with AI-slop used to train more AI-slop.

As one blogger put it: every fifth person suddenly wanted to “vibe-code” some new app that they don’t need in hell. They vibe-coded it, burned money on tokens, and now have no idea what to do with it. Worse still, they don’t even understand what they built or how it works.

It’s like performing open-heart surgery after watching a YouTube video.

Bottom Line

From an MVP perspective - it’s efficient. From a proof-of-concept perspective - it’s garbage. For learning - it works well. For physical execution and combat deployment in place of real human performers - only as an assistant.

AI is an assistant, not a replacement. The decision-maker remains human.

If you’re tired of empty hype and ready to build products with real human logic rather than algorithmic guesswork, hit Subscribe below and drop your thoughts in the comments - let’s discuss how we can keep common sense alive in modern management.

DEV Community: I am PM at 3 AM