The Companies Winning With AI Aren’t Using Better Tools
Everyone has access to the same AI tools. The same models, the same interfaces, the same capabilities. So why are some people getting dramatically better results than others?
It’s not prompt engineering. It’s not which subscription they’re paying for. It’s context. The depth, quality, and structure of the information they feed into these tools is what separates useful output from generic noise. And most people aren’t building that context at all. They’re starting every AI interaction from zero, wondering why the results feel hollow.
This is about to become a serious competitive divide.
The Productivity Paradox Nobody Wants to Talk About
There’s an assumption baked into the AI hype cycle that more tools equals more output. Give your team AI, the thinking goes, and watch productivity soar.
The data tells a different story. Research from Harvard Business School, conducted with 758 consultants at Boston Consulting Group, found that AI created what the researchers called a “jagged technological frontier.” For tasks inside that frontier, consultants using AI completed 12.2% more tasks, 25.1% faster, with 40% producing higher quality results. Impressive. But for tasks outside the frontier, consultants using AI were 19% less likely to produce the correct answer. They were faster, but faster to the wrong conclusion.
A separate study from METR tracked experienced developers working on their own codebases and found something even more striking. When developers used AI tools, they took 19% longer to complete tasks. Not faster. Slower. Before starting, those same developers predicted AI would save them 24% of their time. After finishing, they still believed it had saved them 20%, despite the measured slowdown. The gap between perceived and actual value should concern everyone.
This isn’t an argument against AI. I use it daily and I’ve written about why it makes most people busier, not more productive. The point is that AI without context is a blunt instrument. It can generate, summarise, and analyse at extraordinary speed. But speed without direction produces volume, not value.
You Know More Than You Can Type
Here’s the fundamental problem. The knowledge that makes you good at your job, the instincts you’ve developed, the patterns you recognise, the relationships you understand, lives almost entirely inside your head. And AI can’t access any of it unless you give it to the tool explicitly.
Ikujiro Nonaka’s research on organisational knowledge, which has shaped how businesses think about intellectual capital for three decades, draws a distinction between tacit and explicit knowledge. Explicit knowledge is the stuff you can write down: processes, data, procedures. Tacit knowledge is everything else: experience, intuition, judgement, contextual understanding. Studies building on Nonaka’s work estimate that tacit knowledge accounts for roughly 80% of an organisation’s total knowledge base. The vast majority of what you know can’t be easily articulated, let alone typed into a chat window.
This is the context gap. Every time you open an AI tool and ask it to help with something, you’re working with, at best, 20% of the relevant information. The other 80% stays locked in your head, invisible to the model, irrelevant to the output. I’ve written before about how your information diet shapes your effectiveness. The same principle applies here, except now it’s not just about what you consume. It’s about what you capture and structure from what you consume.
The people getting exceptional results from AI aren’t smarter or luckier. They’ve found ways to bridge that gap. They’re capturing their thinking, structuring their knowledge, and feeding it to AI tools as persistent context. Not in a single prompt, but as an evolving, interconnected body of information the AI can draw from every time they interact with it.
Why Your Meeting Notes Are Probably Misleading You
If you’re already thinking “I record my meetings, that’s my context,” I’d push back on that.
AI transcription has improved enormously. Zoom now achieves 99% accuracy in ideal conditions, and most major platforms offer native transcription. But accuracy and usefulness are different things.
Every participant in a meeting brings context the AI doesn’t have. References to previous conversations, shared understanding of a client relationship, the subtext behind a cautious answer, the weight of a raised eyebrow when someone says “that’s interesting.” Research on AI transcription limitations confirms that most models struggle with pragmatic and cultural nuance, the kind of contextual signals humans process automatically but AI simply cannot see.
The result is that AI meeting summaries capture the words but miss the meaning. And if you’re feeding those summaries into other AI workflows, making decisions based on them, or using them as the foundation for follow-up actions, you’re building on incomplete information. It’s the same principle as garbage in, garbage out, just dressed up in a fancier interface.
This doesn’t mean meeting transcription is useless. It means it needs a human filter. Someone who was in the room, who had the context, who can distinguish between what was said and what was meant. That filtered version, the one with the human judgement baked in, is the context that’s actually valuable.
The Compounding Advantage of Structured Context
Think about this in terms of compounding returns. Every piece of context you capture and structure today makes every future AI interaction slightly better. Over weeks and months, that advantage compounds. You’re not just saving time on individual tasks. You’re building a foundation that makes all of your AI-assisted work more effective.
Research on executive productivity suggests that knowledge workers spend roughly 30% of their working day searching for information they know exists somewhere. That’s close to one full day per week lost to finding things. A structured personal knowledge base doesn’t just feed AI better. It makes you more effective with or without it.
The businesses and individuals who are building this advantage now will pull further ahead as AI tools improve. Better context windows, more sophisticated retrieval systems, and more capable models all amplify the value of the context you’ve already captured. The gap between someone who has been systematically recording and structuring their thinking for two years and someone who starts from scratch will only widen.
This is the context battle. And it’s already underway.
What a Practical Context System Actually Looks Like
This doesn’t require expensive software or a complex infrastructure. It requires discipline and a format that both you and AI can work with.
Markdown files (.md) have emerged as the format of choice for this kind of knowledge capture, and for good reason. They’re plain text, so they’ll never become unreadable due to a proprietary format change. They’re structured enough that AI can parse them efficiently. And they’re human-readable, which means you can style them, link them together, and browse them without any special tools.
The practical system looks something like this:
Daily capture. Record your thinking, decisions, observations, and filtered meeting takeaways every day. This isn’t journaling for the sake of it. It’s creating a searchable, structured record of the context that would otherwise evaporate. The emphasis is on filtering: not everything that happened, but everything that mattered and why.
Weekly synthesis. Summarise your daily notes into a weekly digest. What were the goals? What actually happened? What changed? This layer of synthesis is where raw notes become useful context, because it forces you to identify patterns and priorities that a daily view misses.
Interconnection. The real power comes when individual notes link to each other. A meeting note connects to a project brief. A decision log connects to the data that informed it. An observation connects to a previous observation that contradicts or reinforces it. Tools like Obsidian make this kind of interconnection native to the workflow, but the principle works regardless of the tool.
Proximity to your AI tools. Your context needs to be accessible to the AI tools you use. That might mean keeping files locally where desktop AI agents can read them, or structuring your notes so they can be easily loaded into a conversation. The closer your context sits to your AI workflow, the more useful it becomes.
The Human Filter Is the Competitive Advantage
There’s an irony here. In a world obsessed with automation and AI agents, the competitive advantage is profoundly human. It’s your ability to filter signal from noise, to recognise what matters, to add the context that no transcript or data feed can capture automatically.
AI is extraordinary at processing information. It’s terrible at knowing which information matters. That’s your job. And if you’re doing that job well, capturing the output, and structuring it in a way that compounds over time, you’re building something no tool can replicate and no competitor can shortcut. This is what the rise of the digital worker actually looks like in practice. Not a human replaced by AI, but a human whose accumulated context makes every AI interaction exponentially more valuable.
McKinsey’s research shows that nearly 90% of companies have invested in AI, but fewer than 40% report measurable gains. The gap isn’t about technology. It’s about the quality of the inputs. The businesses seeing real returns are the ones that have invested in the human systems that feed their AI tools, not just the tools themselves.
The context battle isn’t about who has the best AI subscription. It’s about who has the richest, most structured, most thoughtfully curated body of knowledge feeding into those tools. Start building yours now, because the gap is already opening.
