By Francois van der Merwe, Founder & CEO, Otinga.io
If you’re leading a large organisation in 2026, you can feel the ground shifting under your feet. Not because the rules have changed, or because a new competitor launched a shiny product, but because the underlying economics of advantage are being rewritten in real time.
Intelligence is no longer scarce. Code is no longer scarce. Even labour, at least the “work” portion of it, is becoming increasingly abundant through AI agents and automation.
That combination changes more than productivity. It changes the terrain on which competitive advantage is built. It’s the difference between playing the same sport under different rules, and discovering that the field itself has been replaced overnight.
In that reality, the executive question is no longer, “How do we use AI?” but rather, “What do we uniquely know, and can we make that knowledge usable at machine speed?” Because when intelligence, code and labour are widely accessible, the last meaningful differentiator left is context: the proprietary understanding of your customers, your operations, your decisions, your history, and your constraints.
The rise of the “semantic layer”
A lot of the excitement around agentic AI platforms is tied to a simple promise: AI agents that don’t just answer questions, but do work across systems. They can pull information, execute tasks, coordinate workflows, and act with persistence.
But there’s an assumption underneath that promise that most organisations haven’t confronted yet: to deploy agents safely and effectively, an enterprise needs a machine-readable map of how the business actually works, a semantic layer that connects data, processes, roles, permissions, and decision logic. However, most enterprises do not have that layer. They have systems, dashboards, data lakes and reports and policies and workflow diagrams. But they don’t have a coherent, governed, connected representation of the organisation that a machine can interpret reliably.
This is why so many AI initiatives stall, fragment, or remain trapped in pilots. The problem isn’t that leaders don’t want AI, rather, the problem is that organisations are trying to pour high-powered capability into low-quality foundations.
Context engineering is the new operational excellence
Here’s the mental model that matters: the quality of decisions and outcomes across a business has always been shaped by the quality of context available to the people doing the work. A surgeon performs better with full patient history. A salesperson closes better when they understand the customer’s real situation. Teams execute better when they can see how their work connects to strategy and what “good” looks like.
What’s new is that this is even more true for AI agents. An agent with rich, accurate, current context will outperform an agent with generic context, even if the underlying model is identical. The gap is orders of magnitude. That’s why I believe context engineering is the new operational excellence. Not a technical upgrade or a software choice, but an executive mandate. Your most important job has shifted from being excellent inside your functional silo, be that in finance, operations, tech or marketing, to ensuring that every action across the organisation (human or machine) is operating with the highest quality context available.
It’s a profound change in what leadership means.
The fastest route to value is not “perfect data”
At this point, someone will say: “We don’t have perfect data.” And they’ll be right. No enterprise does. Data readiness is not a binary state. It’s an ongoing discipline in a living system that changes every day.
So the objective is not perfection. The objective is Pareto value: reach the level of readiness that unlocks meaningful outcomes quickly, then keep improving from there.
And in almost every enterprise, the best place to start is your customer model. Not because it’s easy, but because it’s the highest-leverage context you can create. If your organisation can build a canonical, governed, machine-readable understanding of who your customer is, what they need, how they behave, and what they value, you’ve created a foundation that can power everything from service and sales to product and risk.
This is also where executive alignment becomes real. Because when teams can operate from a shared, trusted view of the customer, the organisation stops debating opinions and starts executing against shared reality.
If you can’t value your data, you don’t know your moat
There’s another step most organisations skip, and it’s the one that separates organisations that will win from those that will merely “adopt AI.”
You need to infer the commercial value of your data assets: individually and collectively.
Most companies can’t answer the question: “What is your data worth?” with anything more concrete than “It’s important.” But in a world where context becomes the moat, not knowing the value of your data is like not knowing your market cap, except it’s way more dangerous, because it shapes how you allocate investment, how you prioritise governance, and how you negotiate vendor relationships.
Valuing data forces hard decisions:
- Which datasets are strategic, and which are just “digital clutter”?
- What must be governed with zero tolerance, and what can be improved over time?
- Where is the fastest ROI from enabling high-context decision-making?
- Which data should never leave your control, regardless of contract terms?
Without that clarity, AI becomes a series of experiments, but if you have that clarity, AI becomes a compounding capability.
The interface shift is already underway
While organisations debate AI roadmaps, the way people interact with enterprise systems is already changing. Think about how internet or simple Google searches have already been transformed. People are moving away from navigating complex UIs and toward querying information conversationally, expecting systems to respond with outcomes rather than menus.
This matters because it changes what “being digital” means. The next layer of enterprise advantage won’t be who has the most dashboards. It will be who can make their knowledge such as their policies, processes, product logic, customer understanding, accessible in machine-readable form so agents and people can act on it quickly and safely.
In other words: the interface shift is not cosmetic. It forces a deeper architectural and organisational change.
What leaders should do next
If you’re a CEO, CFO, COO, CIO or board member, the response does not need to be dramatic, but it must be decisive. Here are the practical moves that create high-context capability without waiting for a perfect enterprise-wide overhaul:
- Start with a canonical customer model. Treat it as a strategic asset, not a marketing database.
- Build a data catalogue and access model. Enumerate what you have before you try to leverage it.
- Define governance as an enabler, not a blocker. The objective is safe speed, not slow control.
- Measure context quality, not just data volume. High-context beats “more data” every time.
- Value your data commercially. Turn “data is important” into quantified priorities and investment cases.
These steps are not glamorous and they’re certainly not as exciting as a public AI launch or a viral demo, but they are the work that creates enduring advantage.
The organisations that will lead in 2030 will do one thing better
The winners won’t necessarily be the companies with the biggest AI budgets. They will be the companies that understood early that competition has changed, and built the foundations to operate in that new world. They will have leaders who are personally committed, and organisations that can turn what they know into high-context action at scale.
Because to innovate is human. But to innovate consistently, at enterprise scale, under pressure – that requires intention.
