Agnostic on the Engine, Opinionated on the Steering Wheel: A Different Approach to AI
Created on 2026-02-06 08:49
Published on 2026-02-11 09:15
Why we refuse to sell you technology and what we focus on instead
Every week, someone asks me which AI platform they should buy.
ChatGPT or Claude? Microsoft Copilot or Google Gemini? Should they build their own models or license from vendors? What about the open-source options?
I understand the question. With hundreds of AI platforms competing for attention and billions of dollars in marketing spend telling you that each one is the answer, the technology choice feels like the most important decision.
It is not.
And our refusal to answer that question is the core of what makes AIR APAC different.
The Engine Versus the Steering Wheel
Here is how I explain our approach.
AI is the engine. It provides the power. The capability. The raw potential to transform how work gets done.
But an engine without a steering wheel is not transportation. It is a hazard.
The steering wheel is what I call the Human Layer. Leadership. Data. Capability. Process. Governance. Culture. The system of human judgment and intervention points that directs the engine’s power toward useful destinations.
Most organizations obsess over the engine. Which model has the best benchmarks? Which platform has the newest features? Which vendor is winning the capability race?
Meanwhile, they ignore the steering wheel entirely.
This is why 95% of AI initiatives fail.
Not because they chose the wrong engine. Because they never built the steering wheel.
Why We Are Agnostic on the Engine
Let me be direct about our position.
We do not sell AI platforms. We do not take referral fees from vendors. We do not have preferred technology partners who pay us to recommend their solutions.
This is deliberate.
The engine is commoditizing.
Three years ago, there were meaningful differences between AI platforms. Today, the major providers offer remarkably similar capabilities. GPT-4, Claude, Gemini, and the leading open-source models can all perform the core tasks that most organizations need.
The differences exist, but they are at the margins. For most business applications, the choice between major platforms matters less than vendors want you to believe.
Six months from now, the rankings will shift again. The platform that leads today will be challenged tomorrow. The capability gaps will narrow further.
Optimizing for engine selection is optimizing for a moving target.
The engine is not where you fail.
MIT’s research studied over 300 AI implementations. They found that 95% produce zero return. When they investigated why, they did not find technology failures.
They found approach failures.
“This divide does not seem to be driven by model quality or regulation, but seems to be determined by approach.”
Organizations with inferior technology but superior approach outperform organizations with superior technology but inferior approach.
The engine is almost never the constraint.
The engine decision distracts from harder work.
I have watched organizations spend months evaluating platforms. They run proofs of concept. They benchmark performance. They negotiate contracts.
Meanwhile, their leadership remains misaligned. Their data stays siloed. Their people lack capability. Their processes remain broken.
The platform evaluation feels productive. It generates meetings and documents and decisions. It looks like progress.
But it is often procrastination dressed as strategy.
The harder work is building the Human Layer. The platform decision is easier. Organizations gravitate toward easy.
Why We Are Opinionated on the Steering Wheel
If we are agnostic on engines, we are decidedly not agnostic on steering wheels.
We have strong opinions about what makes organizations ready for AI. We have frameworks developed across twenty-five years of transformation work. We have seen what succeeds and what fails.
Leadership must lead, not delegate.
AI transformation is not a technology project that IT handles while executives approve budgets. It is a strategic reorientation that requires executive understanding and engagement.
Leaders do not need to understand how neural networks function. They need to understand what AI will change about their competitive position. They need to articulate a vision in concrete terms. They need to personally use AI tools so their visible adoption signals organizational priority.
We are opinionated about this. Leaders who delegate AI without understanding it will fail.
Data governance must be operational, not theoretical.
Every organization has data policies. Few organizations have data that flows where it needs to go when it needs to get there.
The question is not whether you have governance documents. The question is whether you can get data from System A to System B in less than a week. Whether the same question produces the same answer regardless of which system you query. Whether someone is actually accountable for data quality.
We are opinionated about this. Theoretical governance does not enable AI. Operational governance does.
Capability must include judgment, not just usage.
Training people to use AI tools is easy. Training people to judge AI outputs is hard.
The shift from creator to auditor is the defining professional transition of our era. People must develop the ability to evaluate whether AI outputs are correct, appropriate, and aligned with organizational intent.
We are opinionated about this. Usage training without judgment development creates dangerous dependency.
Processes must be redesigned, not just automated.
If you automate a broken process, you get a faster broken process.
Most organizations layer AI on existing workflows without examining whether those workflows make sense. They pave the cow paths, encoding inefficiency into their AI systems.
We are opinionated about this. Automation without redesign scales dysfunction.
Governance must enable, not just restrict.
Governance is not the Department of No. Governance is the Department of How.
Good governance gives people clear boundaries within which they can move quickly. It creates confidence through clarity. It enables experimentation by defining what experimentation looks like.
We are opinionated about this. Governance that only restricts will be circumvented. Governance that enables will be adopted.
Culture must support learning from failure.
AI transformation involves uncertainty. Experiments will fail. Approaches will prove wrong. Organizations that punish failure will not transform.
We are opinionated about this. Psychological safety is not a nice-to-have. It is a prerequisite.
The Architect Versus the Builder
There is a distinction that helps explain our positioning.
Some firms are Builders. They implement technology. They write code. They configure platforms. They are valuable when you know what you want to build and need skilled hands to build it.
We are Architects.
Architects do not lay bricks. They design structures. They ensure the foundation can support what will be built on top. They create blueprints that Builders then execute.
Before you hire Builders, you need an Architect. Before you configure platforms, you need to know what you are configuring them for. Before you implement technology, you need to ensure your organization can absorb it.
This is what we do.
We assess readiness across the six dimensions. We identify gaps. We design approaches that address those gaps. We create the conditions under which technology deployment can succeed.
We do not implement the technology. We ensure the organization is ready for the technology.
This means we often tell clients things they do not want to hear.
You are not ready to deploy AI at scale. Your leadership is not aligned. Your data is not accessible. Your people lack the capability to judge AI outputs. Your processes need redesign before automation.
These are uncomfortable conversations. But they prevent expensive failures.
What Independence Means
Our technology agnosticism has a practical implication: we have no incentive to sell you anything except clarity.
Vendors make money when you buy their platforms. They are incentivized to emphasize capabilities and minimize limitations. They are incentivized to encourage deployment even when you are not ready.
Consultants who take implementation fees are incentivized to recommend implementation. The more technology they deploy, the more they earn.
We make money when you get ready. Our success is measured by whether you can deploy AI effectively, not by which AI you deploy.
This changes the conversation.
We can tell you that you are not ready without losing revenue. We can tell you to wait without undermining our business model. We can tell you that the platform choice matters less than the organizational readiness without contradicting our financial interests.
Independence is not just a positioning statement. It is a structural feature of how we work.
A Story About Engine Obsession
A major Malaysian bank licensed a customer service AI from a US vendor. State-of-the-art technology. Excellent performance in American markets. Strong benchmarks.
It failed spectacularly in Malaysia.
The AI could not handle Singlish. The natural code-switching between English, Malay, Chinese dialects, and Tamil that Malaysians use in everyday conversation.
A customer might say: “Eh, my account got problem lah, the money never come in one.”
That is a legitimate customer service query in Malaysian English. The US-trained AI interpreted it as gibberish.
The AI also could not navigate cultural context. The indirect communication patterns where “maybe” often means “no.” The relationship signals embedded in how people phrase requests. The face-preserving language that avoids direct confrontation.
It failed in ways no American test would catch.
Every failure damaged trust. Adoption stalled at 15%. The investment was written off.
This is what happens when you focus on engine and ignore context.
The technology was excellent. The engine was powerful. But without the steering wheel of local understanding, that power was useless.
No platform evaluation would have prevented this failure. Only Human Layer work would have. Only understanding the context into which the AI would be deployed.
What We Actually Do
Let me be concrete about what this approach looks like in practice.
We assess.
Using our six-dimension framework, we evaluate where your organization actually stands. Not where you think you stand. Not where your vendor assessments say you stand. Where you actually stand.
This involves uncomfortable questions. Can your CEO articulate what AI will change about your competitive position? Can you get data from one system to another in less than a week? Can your people distinguish good AI output from bad? Are your processes designed or accidental?
Most organizations score lower than they expect. That gap between expectation and reality is precisely what needs to be closed before technology deployment.
We prioritize.
You cannot fix everything at once. The assessment reveals many gaps. Our job is to help you identify which gaps matter most.
Leadership alignment problems block everything else. Data accessibility problems prevent most AI applications from working. Certain gaps are prerequisites; others can be addressed in parallel with deployment.
We help you sequence the work so effort flows to where it creates most value.
We design.
The Human Layer does not emerge accidentally. It must be designed.
Who decides what AI can do? Who reviews outputs? Who intervenes when something fails? How does capability develop over time? What governance structures enable rather than restrict?
We work with you to answer these questions and document the answers in operating principles your organization can follow.
We prepare.
Before technology deployment, there is preparation work. Leadership alignment sessions. Data accessibility projects. Capability development programs. Process redesign initiatives. Governance framework development.
This is the invisible work that determines visible success. We guide you through it.
We do not implement.
When you are ready to deploy technology, you work with Builders. Implementation partners. Technology vendors. System integrators.
We help you select them. We help you manage them. We help you evaluate whether their work aligns with your readiness.
But we do not compete with them. Our work is done when your organization is ready. Their work begins when the implementation starts.
The Question Behind the Question
When executives ask me which platform to buy, they are usually asking a different question underneath.
They are asking: “How do I avoid failing at AI?”
The platform question feels concrete. It feels answerable. It feels like the kind of decision executives are trained to make.
But the real answer to “how do I avoid failing at AI” is not a platform recommendation. It is readiness assessment. It is Human Layer development. It is the invisible work that makes technology deployment possible.
The platform matters less than you think. The readiness matters more than you know.
The Invitation
If you are looking for someone to tell you which AI platform to buy, we are not the right fit.
Plenty of firms will give you that recommendation. Many will be influenced by referral fees and implementation revenue. Some will give you genuinely independent advice.
We are not in that business.
If you are looking for someone to help you understand why AI initiatives fail and how to build an organization where they can succeed, that is precisely what we do.
We are agnostic on engines because engines are commoditizing and engine choice is rarely the constraint.
We are opinionated on steering wheels because the Human Layer is where organizations actually fail, and where intervention actually helps.
AI is the engine. You are the steering wheel.
Our job is to help you become a better steering wheel.
Are you focused on the engine or the steering wheel? What would shift if you spent as much time on readiness as you spend on platform evaluation?
If you want to assess your Human Layer across all six dimensions, comment “SCORECARD” below. I will send you the assessment I built for mid-market APAC leaders.
The technology will keep getting better. The question is whether your organization will be ready to use it.