Six months into working at Wadhwani AI Global, I’ve learned that the most important word in my job title isn’t “communications.” It’s not even “strategy.” It’s the invisible word that precedes both: context.
I came into this role with a fairly standard comms toolkit. Messaging frameworks, audience segmentation, campaign calendars, content pillars. The kind of structured approach that works well when your organisation has one geography, one cultural register, one kind of stakeholder. WAIG is none of those things: at any given moment, our team is working across multiple countries, sociopolitical systems, and with governments, civil society, founders, and funders alike. And the further I’ve gone into this work, the more I’ve come to understand that the framework I actually needed was hiding inside a concept we use to talk about AI, not marketing at all.
That concept is contextual engineering
In AI development, contextual engineering refers to something quite specific: the practice of building the right information environment for an AI system before you ask it to respond. A bare prompt gets a bare answer. But when you construct context, relevant background, precise framing, the right constraints, the system can actually do something useful with what you’ve given it. The quality of the output is a direct function of the quality of the context you build around it.
That idea stopped me when I first encountered it, because it describes something I’d been experiencing in communications without having the language for it. A message doesn’t land because it’s well-written. It lands because the context around it has been carefully built first. Strip that away and even a genuinely useful idea reads as noise. The parallel isn’t perfect, but the discipline is the same: you cannot bolt context on afterwards. It has to be the foundation.
If your model cannot navigate multilingual code switching and local linguistic norms, it breaks down in Lagos. If its recommendations are calibrated for mainstream crops rather than local food economies, it alters food systems, to the detriment of food security and smallholder farmers. If its diagnostic assumptions reflect Western healthcare infrastructures and disease prevalence rather than the clinical realities of East Africa, it risks failing physicians making life-or-death decisions in Nairobi. You cannot bolt context on afterwards it has to be the foundation. I’ve been sitting with this idea, turning it over. And slowly I’ve realised it describes exactly what’s been broken in how I’ve had to rethink communications.
When I joined WAIG, one of my first instincts was to build consistency. A unified voice, a clean global narrative, messaging that would travel. That instinct isn’t wrong, but I’ve learned it’s dangerously incomplete.
I kept asking myself: can the same event in Kigali work in Ethiopia? Can the same conversation held in Nairobi land the same way at a summit in India, where delegates have flown in from across the world? The answer, almost every time, was no. Not because the ideas aren’t relevant, but because the context that makes those ideas land is completely different.
In one workshop hosted alongside funder- led AI Scaling Hubs across Africa, I watched our senior leadership take the stage to present our tools and our work. But before they got anywhere near the products, they talked about India. They talked about where Wadhwani AI was built, the constraints that shaped it, the density of smallholder farmers, the fragmented healthcare infrastructure, the need to build for low-bandwidth, low-literacy environments. They drew a direct line between that context and Africa’s, not as a comparison of equals, but as a recognition of shared complexity. What struck me was how much work that context did on its own: suddenly, the audience could see their unique challenges reflected in WAIG’s approach to problem-solving.
Later that day, in every interview and conversation I had on the floor, I stopped leading with “what do you do” and started asking “why does this matter to you, here, now.” The answers were completely different. That’s marketing. Not the what, the why, and the where, and the who asking it. That’s the marketing version of contextual engineering. You cannot tack on relevance afterwards. You have to build with it from the beginning.
Six months in, this shapes nearly every decision I make. Before ideating any piece of content, any event, any post, the first question I ask isn’t “what do we want to say?” It’s “who is in the room, what do they already believe, what does this mean to them?”
Running communications for a global organisation sounds like it should be about finding the universal. In practice, it’s mostly about resisting that temptation. That instinct to smooth everything into one clean narrative, while professionally comfortable, is strategically dangerous. Especially when what reads as ambition in one geography reads as overreach in another. What signals partnership in one context signals dependency in another.
I’m also still learning, six months is not enough time to have earned certainty here. The organisations inspiring me aren’t usually those with the loudest global presence. They’re the ones who’ve sat with local partners long enough to understand the soil. At WAIG, that’s the ethos we lead with: MoUs with ministries, partnerships with institutions like Smart Africa, engagements that are slow, deliberate, and shaped by field learning because the stakes require that.
Context isn’t the last mile of a good idea. It’s where the idea actually begins. For someone who came into this role armed with frameworks, I’ve learned that the richest strategy doesn’t come from the toolkit, it comes from the room, the relationship, and everything that precedes the message. That’s been the most important thing I’ve learned. And I suspect I’m only six months into understanding what it really means and there’s a long way to go.
