In the Era of AI, Context Is Queen(and Content Is Still King!)

I attended GleanGo in San Francisco a couple of weeks ago, and if I can share ONE key takeaway, it's this:

We are spending far too much time talking about what AI can do and not nearly enough time talking about what it KNOWS.

And that distinction is everything. The enterprise AI race has largely been framed around models, copilots, agents, productivity, and increasingly sophisticated automation. Every few weeks, another capability arrives that moves the frontier forward.

But I don't believe the next phase of AI will be won by adding more tools. It will be won by context.

An AI system can have access to extraordinary intelligence and still produce remarkably mediocre work if it doesn't understand you, your organization, your goals, your standards, or the environment in which the work actually happens.

For years—and especially through the rise of the internet and social media—we've repeated the same adage:

Content is King.

And as someone who has built much of my career around storytelling and content, I am certainly not here to dethrone the King!

Content is still the knowledge, the evidence, the story, the intellectual property, the ideas. It gives AI something to work with.

But AI has made something else much more visible: Content tells AI what exists. Context tells AI what matters.

So in the era of AI, I think we need to make room on the throne. Context is Queen.

The Shift I Ask Leaders to Make: From Engineering to Architecting

In my work as an AI strategist, one of the biggest mindset shifts I encourage people to make is to stop thinking only about prompt engineering and start thinking about context architecture.

Prompt engineering asks:

What should I tell the AI to do?

Context architecture asks a much bigger question:

What does the AI need to understand in order to do this well?

There is a big difference.

Think about bringing a brilliant new person onto your team. You wouldn't hand them one instruction and expect them to immediately understand how you think, how the organization works, what good looks like, who the audience is, what has already been tried, or which decisions they can make independently.

You would onboard them. You would give them history, examples, expectations, language, goals, and boundaries.

You would introduce them to the culture and explain the things that aren't written down anywhere but that everyone who has been around long enough simply knows.

AI needs an onboarding experience too.

At GleanGo, this idea of context went far beyond connecting AI to documents and databases. Context can include activity, identity, intent, institutional knowledge, permissions, history, and even the human intuition embedded in how work gets done.

At Empressa, I have been instinctually building what I call the "Operating System (OS)" for AI agents: a stack of at least 10 markdown (.md) files that gives the agent the context it needs to understand its role, its environment, its knowledge, its boundaries, and how it should operate.

That is why I'm increasingly encouraging leaders and professionals to think like architects, not simply engineers. You are not just writing better instructions. You are designing the environment around the intelligence. And the more intentional that architecture becomes, the more useful, consistent, and trustworthy AI becomes inside it!

So how do you actually do that?

Here are three places to start.

1. Give Your AI an Identity

The first layer of context is identity.

Most people open an AI tool and immediately give it a task: "Write this." "Analyze this." "Create this." "Summarize this."

Then they wonder why the output sounds generic.

The AI may understand the assignment, but it doesn't necessarily understand who it is supposed to be while completing it.

Before assigning the work, establish its identity. What role is it playing? Who is it serving? What expertise should it draw upon? What principles should guide it? What does "good" look like? What should it never do?

If I'm building an AI strategist, for example, I don't only tell it to "provide strategic recommendations." I want it to understand the perspective from which those recommendations should be made, the audience receiving them, the evidence standards I expect, the language we use, the assumptions it should challenge, and when it should say, I don't have enough information.

Think of this as the AI's operating identity.

You can create a simple context document that answers questions such as:

• Who are you?
• Who are you helping?
• What are you here to accomplish?
• What principles guide your work?
• What does excellent work look like?
• What language do we use or avoid?
• What decisions can you make?
• When should you ask for help?
• What should you never assume?

This doesn't have to be complicated.

A one-page brief can dramatically change the quality and consistency of what you get back.

Because now the AI isn't simply responding to a prompt.

It knows who it is supposed to be.

2. Give It Your World, Not Just Your Files

The second layer is knowledge.

This is where content remains King.

Give AI the source material it needs: your research, strategy, presentations, reports, writing, customer insights, processes, examples, decisions, and other relevant information.

But don't stop at uploading files.

Organize the context around them.

A twenty-page strategy deck doesn't automatically tell an AI which idea matters most.

A folder full of brand documents doesn't tell it which document is authoritative.

Ten examples of your writing don't necessarily explain why one sounds like you and another doesn't.

This is where architecture matters.

Tell the AI what your current strategy is, which document supersedes another, who your primary audiences are, what you believe, which proof points you trust, and which examples represent excellent work. Tell it what is outdated, still being tested, confidential, and most important right now.

You're not simply giving AI a library.

You're giving the library a librarian.

And that layer of meaning becomes even more important as we move toward agents.

An agent may increasingly be able to search, prioritize, recommend, coordinate, and execute. If we expect it to take action, it cannot simply have more information.

It needs to know which information should shape the action.

That is context.

3. Teach It Your Judgment

This may be the most important layer of all.

Give AI examples of how you think.

Because the most valuable context you possess may not live in a document anywhere.

It lives in experience.

Think about how an experienced professional makes a decision. They don't simply retrieve information. They understand what happened before. They recognize patterns. They know which stakeholder needs to be involved. They notice when something feels inconsistent. They remember the mistake that was made three years ago and quietly avoid making it again.

Much of that knowledge has become so natural that we don't even recognize it as knowledge anymore.

We call it judgment.

And while I don't believe we should—or can—outsource human judgment entirely to AI, we can give AI much more context about how we exercise it.

One of the best ways to do this is through examples and corrections.

Don't simply fix an AI output and move on.

Tell it why you changed it.

"This recommendation is technically correct, but it ignores how this audience will perceive it."

"I chose option B because protecting trust matters more here than optimizing for speed."

"This sounds like me, but I would never use that phrase because it conflicts with how I talk about our customers."

"These two sources disagree, but this one should take precedence because it is newer."

"This is the kind of decision I want you to escalate rather than make independently."

Those corrections are incredibly valuable. They reveal the invisible criteria behind your decisions. Over time, you are creating a stack of context around the AI:

That is context architecture.

Your Context Stack Becomes an Asset

This is where the conversation becomes much bigger than better prompting.

Organizations are beginning to recognize that institutional knowledge is one of the most valuable inputs they can give AI.

I believe professionals should recognize the same thing.

You have institutional knowledge too.

Years of experience have created a context layer around your work: the frameworks you developed, the patterns you recognize, the stories you know how to tell, the relationships you've built, the mistakes you no longer make, and the judgment that looks effortless today because you spent twenty years earning it.

That context is part of your professional value.

And this connects directly to something I've been writing and speaking about extensively: the transition from the Career Economy to the Portfolio Economy.

The Career Economy taught us to think about professional value through the container of a job.

The Portfolio Economy asks us to recognize the assets underneath it: our expertise, relationships, reputation, intellectual property, community, and increasingly, our ability to activate all of those things with AI.

AI dramatically lowers the cost of turning what we know into something we can build.

But there is a prerequisite:

You have to know what you know.

And then you have to make enough of that context explicit that AI can work alongside you without stripping away the judgment that made the expertise valuable in the first place.

This is why I believe becoming an operator in the AI economy is going to matter far more than becoming an expert in any individual AI tool.

Tools will change. Models will change. Interfaces will change.

What remains is your ability to architect the intelligence around you—to decide what AI should know, what it should do, what requires your judgment, and how all of it works together toward an outcome.

Don't Just Prompt AI. Onboard It.

So the next time you open ChatGPT, Claude, Copilot, Gemini, or whatever tool comes next, resist the urge to immediately start engineering the perfect prompt.

Ask yourself three questions first:

Then give it the task.

That is the difference between engineering an instruction and architecting a relationship with intelligence.

And I believe that's where the next era of AI becomes much more interesting.

For decades, we have said Content is King. And it still is.

Content gives AI the substance—the knowledge, evidence, ideas, and intellectual property from which it can work.

But context gives all of that content meaning.

It tells AI who it is, what matters, what to trust, how to interpret what it sees, and when human judgment needs to enter the room.

So I'm not here to dethrone the King. I'm simply suggesting that, in the era of AI, we finally recognize the power sitting beside him.

Content is still King. Context is Queen.

And when the King provides the substance and the Queen understands the whole kingdom, intelligence finally has what it needs to rule wisely.

Long live the Queen.

Miri Rodriguez