Strategy

What Actually Matters When Everyone Has AI

AI generates 10 drafts in seconds. But it lacks the two things that actually matter: Taste and Context. Here is how to build them.

By Essam Shamim · 9 February 2026 · 5 min read

The Tool Matters Less Than You Think

Everyone talks about AI tools. Which one is better. Which one is faster. Which one gives the best output.

I have been using AI heavily for the last two years. What I keep noticing is the tool matters way less than people think.

The real difference is taste and context.

Taste is knowing what looks right and what feels off. What to ship. What to hold back. What to ignore completely.

Context is understanding the situation around the work. The client. The history. The constraints. The risks. The timing.

AI is incredibly good at producing options. Ten drafts in thirty seconds. Five variations on a strategy. Twenty different ways to phrase something.

It is terrible at knowing which option fits this specific situation right now.

Where This Shows Up

The tool cannot make that call. You have to.

What Changed for Me

I use AI for almost everything now. Research. First drafts. Data analysis. Idea generation. But I never let the tool decide what is right.

I am slowly building my own system around it. A kind of internal context operating system that carries how I think, how I decide, and what I avoid.

Example: AI will suggest elaborate onboarding sequences for new clients. Looks great. Twelve touchpoints. Automated everything.

My context says that does not work for the agencies I advise. Clients get overwhelmed. They stop engaging by touchpoint four. Simple three-step onboarding works better even though it feels too basic.

AI does not know that unless I tell it. And even when I tell it, I still have to review the output to make sure it actually followed that rule.

The Shift in What You Actually Do

Because AI handles production, your job shifts to curation and judgment. You spend more time reviewing than producing.

Taste Is Expensive to Build

Here is what most people miss. Taste takes years to develop.

You cannot teach someone taste in a workshop. You build it by doing the work hundreds of times. Seeing what works. Seeing what fails. Developing instinct for what fits and what does not.

AI does not have that. It has patterns from training data. But it does not have your specific taste developed from your specific experiences.

This is why I keep seeing agencies hire people who can use AI tools and then get frustrated. The person can generate content fast. But the content does not match the brand. Misses the client context. Needs heavy editing.

The bottleneck is not the tool. The bottleneck is taste and context.

Context Is Even Harder

Context is everything AI misses. Why did we make this decision six months ago? What happened the last time we tried this?

I keep a running document of context for every major project. Things AI cannot know because they never got written down formally.

AI will give me a formal email with data-driven recommendations. Wrong approach for this client. They want a quick Slack message with a straightforward suggestion.

Building Your Own System

Here is what I am doing to make AI actually useful instead of just fast. I am building a living document of my taste and context.

When I use AI now, I feed it relevant parts of this document first. Then it generates output that is closer to what I actually need.

But even then, I still review everything. Because the document is never complete. New situations come up. Rules need exceptions. Context changes.