One Fighting Unit: What the Mongols Knew About AI That Your Organisation Doesn’t

I delivered 241% sales growth in one year. KPMG audited the numbers. The fastest growing water filter brand in the country. Then the Chairman of Electrolux looked at me across the table and said five words that changed the next twenty years of my career.

“You are not ready for management.”

I was fired.

Not because the results were wrong. The results were spectacular. I was fired because I had achieved those results by destroying everything around them. The sales team had collapsed. Morale was gone. Key people had resigned. I was twenty-nine years old and I had just learned the most expensive lesson of my professional life.

Results without relationships is destruction.

I didn’t understand what that meant. Not really. Not for years. But I understand it now, and I see the same pattern playing out in AI implementations across Asia-Pacific every single week.

The Horse Everyone Is Buying

Here is what most organisations are doing with AI right now.

They buy the tool. They run a training session. Maybe two. They measure license utilisation. They declare transformation.

MIT’s Project NANDA research, published in July 2025 and tracking over 300 implementations, put a number on the outcome: 95% of organisations are getting zero return from their AI investments. Not low returns. Zero.

The finding that should reframe every AI conversation in your boardroom: the divide between the 5% who succeed and the 95% who fail is not driven by model quality or regulation. It is determined by approach.

The technology works. The approach doesn’t.

This is Electrolux all over again. The strategy was right. The execution destroyed the organisation that was supposed to carry it.

But I didn’t come here to talk about what’s failing. I came here to tell you about the Mongols.

Why the Mongols, Not Napoleon

When most people think of cavalry, they think of the Napoleonic model. Heavy horses. Armoured riders. Rigid formations designed for a single devastating charge. It looks impressive on parade. It dominates in the short term.

It’s also exactly the wrong analogy for AI.

Napoleon’s cavalry was expensive to maintain, vulnerable to changing conditions, and catastrophically fragile when the terrain didn’t cooperate. His invasion of Russia is the textbook example: the largest army in European history, built for shock and overwhelm, collapsed because the supply lines broke, the horses died in conditions they weren’t bred for, and the rigid doctrine couldn’t adapt.

Sound familiar? It should. This is the enterprise AI playbook. Massive investment. Rigid implementation. Impressive on the quarterly earnings call. Then the organisational terrain fights back and the whole thing stalls.

The Mongols did something completely different. And what they did maps almost precisely to what the 5% of AI winners are doing today.

The Mongol Advantage

Genghis Khan built the largest contiguous land empire in human history. From Hungary to Korea. His forces were routinely outnumbered. His warriors didn’t have the biggest horses, the heaviest armour, or the most sophisticated weapons.

What they had was integration.

Mongol children learned to ride by age three to five. Not a weekend course. Not a six-hour certification. A lifetime of building the instinctive connection between rider and horse that made them function as a single organism in battle.

Their horses were small, tough, and resilient. They could travel over 80 miles a day and forage under snow for food. They survived in conditions where European warhorses would have died within weeks. Each warrior carried three to five remounts, rotating between them based on terrain and conditions.

The horses were trained to respond to subtle cues. The line between rider decision and horse response was invisible. In battle, they didn’t look like a person controlling an animal. They looked like one entity: thinking, moving, adapting as a single unit.

One fighting unit.

This is not a quaint historical parallel. This is the most precise analogy I have found for what AI readiness actually requires.

What the Mongols Actually Teach Us About AI

Integration, not implementation. The Mongols didn’t “implement” horses. They integrated with them over years of shared experience. The 95% of organisations failing at AI are implementing tools. The 5% succeeding are integrating AI into the human systems that make those tools useful: leadership, culture, capability, process, data, governance. I call this the Human Layer. It’s the deliberately designed system of human judgment, governance, and capability that makes AI safe to scale.

At Electrolux, I implemented a strategy. I did not integrate with the people who had to carry it. The results were mathematically perfect and organisationally catastrophic.

Lifelong rider training, not a weekend workshop. The AI skills training industry has a structural problem. It teaches the 2024 skill in a 2026 world. Most programmes spend a day teaching people to type better prompts, as though “think step by step” were a universal fix.

The real skill isn’t prompting. It’s judgment. Can your people look at what AI produces and know whether it’s right? Can they catch the errors? Can they apply twenty years of domain expertise to supervise a system that moves fast but doesn’t always move right?

I call this the Auditor Mindset. The Mongol rider didn’t just sit on the horse. The rider read the horse’s body language, felt the terrain through the horse’s movement, knew when to give the horse its head and when to assert direction. That instinctive judgment came from thousands of hours of shared experience, not from a training manual.

Multiple remounts for different terrain. Mongol warriors didn’t ride one horse into every situation. They maintained a string of three to five mounts and chose the right one for the conditions. A fast horse for scouting. A strong horse for carrying supplies. A battle-trained horse for combat.

Asking “which AI is best?” is like asking “which horse is best?” Best for what?

The fluent leader maintains a stable: different tools for creative work, analysis, research, communication. The organisations I see struggling hardest are the ones that bought one tool (usually Copilot) and expected it to do everything. That’s taking a draft horse into a cavalry charge.

Small, prepared forces defeating massive armies. This is the insight I want mid-market leaders in Asia-Pacific to hear most clearly.

The Mongols were routinely outnumbered. They won through speed, coordination, intelligence, and adaptability. Not despite their smaller numbers. Because of them.

MIT’s research confirmed this advantage with hard data. Mid-market companies move from pilot to deployment in 90 days. Enterprises take nine months or longer. The difference isn’t resources. The difference is organisational complexity.

A twelve-person Special Forces team can accomplish what would require a hundred conventional soldiers. Not because they’re superhuman, but because they’ve mastered the multipliers: precision, training, coordination. I learned this concept from Peter Treselyan, a former British Special Forces instructor who had trained Navy SEALs. We were running crisis training for CEOs in Australia, and it changed how I think about competitive advantage.

Your mid-market company is the Mongol force. You can get your entire leadership team in one room. You can make a decision this quarter and implement it next quarter. You don’t need fourteen layers of approval to redesign a workflow.

The enterprise is Napoleon’s army: massive, slow, burdened by logistics. They’re filling out requisition forms for AI approval. You can deploy next Tuesday.

But here’s the critical caveat. Force multiplication only works if the force is coordinated. A Mongol unit with bad intelligence, no communication network, and riders who don’t trust their horses isn’t a cavalry. It’s a stampede. The leverage works in both directions.

The Yam: History’s First Context Graph

There’s one more lesson from the Mongols that almost nobody talks about, and it might be the most important one.

The Mongols built the Yam.

The Yam was a relay communication network that stretched across the entire empire. Relay stations every 25 to 30 miles. Fresh horses and riders at each one. Intelligence could travel from Hungary to Mongolia in days, not months.

Without the Yam, the Mongol cavalry was just fast horses going in random directions. With it, every decision could be informed by intelligence from the other side of the empire. Speed had context. Action had judgment. The system was greater than any individual unit.

This is what I call the Context Graph: the accumulated record of not just what your organisation does, but why it does it. The reasoning behind decisions. The exceptions, overrides, precedents, and cross-system context that explain the judgment behind the data.

Most enterprise software records transactions. What happened. The Context Graph records decisions. Why it happened.

If an AI agent looked at your last fifty key decisions, would it find the reasoning in the database? Or would it have to interview a human to understand why you did what you did?

The organisations building their Context Graph now are building their Yam. When their competitors are still sending messages by horseback, they’ll be operating with real-time intelligence across the entire organisation.

The APAC Terrain

One more thing about the Mongol horses. They survived on the steppe because they were bred for the steppe. They could handle extreme cold, navigate rough terrain, and forage where nothing seemed edible. European horses, bred for green pastures and temperate weather, died in the same conditions.

This matters for APAC.

What works in Silicon Valley doesn’t automatically land in Seoul. Or Singapore. Or Jakarta. Or Kuala Lumpur. The regulatory environments are different. The cultural dynamics (face, hierarchy, relationship-first business) are different. The linguistic complexity is different.

I watched a large Southeast Asian bank license a US-built customer service AI. It couldn’t handle Singlish code-switching. It failed in ways that no American test environment would ever catch, because the context of how people actually communicate in this region was never in the system.

The cost of deploying AI that doesn’t understand local context, what I call the Context Tax, is months of wasted momentum and trust that should never have been lost.

APAC needs horses bred for APAC terrain. Not European warhorses imported from Silicon Valley.

The Question

I spent twenty years learning the lesson that Electrolux taught me in one brutal conversation with Chairman Gunnar Broberg.

Results without relationships is destruction. Technology without readiness is the same thing.

The Mongols understood this instinctively. The rider and the horse were one unit. The speed was grounded in intelligence. The small force was amplified by preparation, not just ambition.

The question for every mid-market leader in Asia-Pacific is not “Do you have AI?”

The question is: “Are you and your AI one fighting unit? Or are you just sitting on a horse you don’t know how to ride?”

The answer depends on six dimensions: Leadership. Data. Skills. Process. Governance. Culture. Together, they form the Human Layer. They are almost certainly the thing you’re not measuring. And they are almost certainly the thing that will determine whether your AI investment succeeds or fails.

The scorecard takes 15 minutes. Six dimensions. A score that tells you what you already suspect but haven’t been able to quantify. airapac.org/scorecard

The window is closing. MIT’s interviews with seventeen procurement and IT leaders established consensus: a strategic positioning window is closing between mid-2026 and early 2027.

The Mongols didn’t build the largest empire in history by waiting for better horses.

They built better riders.