In 1996, Bill Gates published an essay titled "Content is King". In it, he predicted how the internet would evolve and where value would ultimately be created. His thesis was simple and prescient: content would be the foundation of the internet economy, just as it had been for broadcast television.
“Content is where I expect much of the real money will be made on the Internet, just as it was in broadcasting." - Bill Gates
He was right.
Over the last 30 years, countless people and companies have built enormous businesses by creating, distributing, indexing, and monetizing content. YouTubers, TikTok creators, and media companies sit alongside some of the most valuable businesses in history:
...and many more.
Everyone is a content creator now. Every company is in the content business—whether they admit it or not.
But in 2026, content is no longer scarce.
Content is a commodity.
We can access nearly the entire body of human knowledge with a few keystrokes. Even more remarkably, AI can now summarize, remix, and explain that content for us in seconds—saving hours of reading, watching, and listening.
That doesn’t mean content isn’t important. It absolutely is, and creators deserve to be compensated for the value they produce.
But the competitive advantage has shifted.
In this new technological era, it’s not content that’s king—it’s context.
We here at Lytical expect the largest and most successful companies of the next 30 years to be the companies that master passing valuable context to AI models.
The more and more I use AI, the more I realize that context is everything.
Yes, you can craft better prompts. You can refine your questions. You can learn the quirks of different AI tools. But the real leap in usefulness happens when the AI already understands the situation you’re operating in—without you needing to explain it every time.
The most powerful AI experiences aren’t powered by better answers. They’re powered by better context.
Here are a few places where that’s become obvious to me.
When I first started teaching myself how to code, the process looked something like this:
Coding wasn’t hard because the information didn’t exist. All the content was there—I was able to ship real applications using it. What I lacked was context: the intuition, the architectural understanding, and the accumulated experience that senior engineers take for granted.
Then ChatGPT arrived in 2023.
Suddenly, I could ask the same questions I’d been typing into Google and get a direct answer—often with example code included. Development became dramatically faster and more enjoyable.
But it wasn't perfect; you were limited by how much you could copy and paste, and the AI model never had access to your full codebase. At best, it was able to alter one file and keep a few more in memory or conversation history, without knowing what other dozens of other file dependencies could break as a result of its changes.
But then in 2025, something I never know I needed came out; Claude Code. Not only was this new tool trained on virtually every programming language ever made, but it now had full context to your entire codebase. The hundreds of files, libraries, dependencies, and more that make up the entire ecosystem of a software application. This was a game changer, but really, only one thing changed; the content was the same (the vast breadth of coding knowledge the AI was trained on). What changed was, now the AI had the full context behind your increasingly angry questions of "WHY ISN'T THIS RENDERING CORRECTLY!?!".
Best of all, Claude Code didn't ask developers to change their workflow at all. It integrated natively into the tools they were already using; terminal, text editors like VS code, Github repos. This not only eliminated the switching costs (not just financial but time and effort switching costs), but it also enhanced developer workflows and enabled significantly more work to be done in the same amount of time; context switching costs are real.
I don't see Claude Code as a tool that will destroy developers jobs; I see it as a way to unleash a massive new wave of innovation. As Bill Gates said, software is content (which is why it's subject to copyright and licensing), and Claude Code will enable a massive amount of new software as people test new ideas and new ways of doing solving problems with software.
I've recently gotten big into personal finance, and on one of my favorite podcasts, I frequently hear the line "Personal finance is personal".
In personal finance, there's no one size fits all approach; what works well for one person could be disastrous for another, even with many things being equal like age, income, and location.
What I've come to understand this to mean is that context is extremely important in personal finance. Financial advice is everywhere. Articles, podcasts, calculators, “rules of thumb.” The content isn’t wrong—but it’s generic. Advice like “maximize your 401(k)” or “diversify your portfolio” only works when you ignore personal context: income volatility, risk tolerance, upcoming life events, tax exposure, and time horizon.
AI changes financial guidance only when it understands the full picture:
The same advice becomes either brilliant or dangerous depending on context.
The intelligence didn’t improve. The relevance did.
One of the best descriptions I’ve heard of MCP (Model Context Protocol) is that it’s the “USB-C port for AI applications".
Everyone is familiar with USB (and more recently, USB-C which, thanks to the EU, is the new standard for ALL phone charging ports). USB is a common connector that allows us to connect different devices to one another; phones, printers, laptops, tablets, game consoles, and countless other devices.
MCP is the same way; instead of AI models being limited to static training data, MCP allows tools and systems to surface live data in a standardized way. Any compatible AI model can access that data to gain context and produce better outputs.
MCP is already powering some pretty incredible integrations between different tools. LLMs are no longer limited to their static training data; they can now access context in a standardized way from a variety of systems, databases, and tools. This saves developers considerable time by eliminating the need to write custom integration code for every new AI model or tool that comes out in the future; simply plug it into your existing MCP server and it's good to go.
MCP allows for more capable and helpful assistants and agents, enabling them to work faster, more efficienctly, and with greater accuracy and reliability. I've found that with better context, AI's tend to hallucinate less because they have more concrete facts and data to base their responses on.
I’ve worked in digital marketing for over 15 years. I’ve used dozens of tools for analytics, SEO, advertising, and reporting. When I started building Lytical, my initial goal was simple: create the tool I always wished I had.
Part of it was built to avoid switching costs (having analytics, SEO, and ads data all in one centralized place), but as I continued to build, I realized just how much the way I've been working for the past 15 years will be changing.
The concept of "building a report" or "refreshing a dashboard" is not going to exist in 5 years. Going out and trying to seek data to understand why something is (or isn't) working in the marketing funnel will no longer be a challenge. All of the data is there, but stitching it together and drawing meaningful insights from it has always been the challenge, but that's exactly the type of challenge that AI is built for.
The real challenge has never been access to data. It’s been stitching that data together and understanding why something is happening—and what to do next.
Lytical is designed to provide AI models with the full context of your marketing ecosystem, including:
When that context is available, AI can do more than explain what is happening. It can explain why—and surface opportunities and risks you didn’t even think to ask about.
As I’ve tested Lytical on my own site, I’ve uncovered issues I missed despite 15 years of experience and a full manual audit. With complete context, the AI can monitor and analyze continuously. Even the best, most caffeine addicted growth-marketers I know need at least 4-5 hours of sleep per night.
Like Claude Code, Lytical is built to fit into existing workflows.
We use MCP to let users ask Claude and ChatGPT questions about their sites with full context from Lytical’s tracking and site scans. We built a Slack integration so teams can interact with Quincy directly—1:1 or in channels—receive automated alerts, and get scheduled updates.
Marketing is a team sport. Data and insights shouldn’t live in silos.
That’s why Lytical is organized around channels. Each channel becomes a living, shared context around a specific topic—effectively a dynamic dashboard that updates automatically and invites collaboration.
The most helpful AI systems are powerful not because of what you type into the prompt, but because of what you don’t have to include.
Conversation history. Stored knowledge. Live data from external systems. Ongoing monitoring as new information becomes available.
When AI has context, it doesn’t just answer questions—it surfaces insights, alerts, and opportunities you never asked for.
Five years from now, I expect every company to have a content strategy. I also expect the best companies to have a context strategy.
What context are we providing to our AI systems?
What context do we need from our own tools to do our best work?
How do we ensure the right content reaches customers in the right context?
Content is the message. Context is the meaning.
Without context, content is just noise.
To paraphrase Bill Gates’ conclusion from nearly 30 years ago: those who succeed will propel the internet forward as a marketplace—not just of content, ideas and products, but of context.
“The companies that win won’t create more content—they’ll ensure the right context is always present when decisions are made.”