The AI Growth Stack: What a Fractional CGO Actually Uses and Why
People ask me how big my team is. When I tell them my answer, they pause.
“So it is just you?”
Not quite. I have a content strategist and writer, a branding and design partner, a business analyst, and a developer. Each one is a trained cognitive function I built inside Claude Code. Not a chatbot I prompt on the fly. A structured skill with defined judgment, rules, and output standards.
That distinction matters. Most AI commentary tells you which tools to add. I am going to argue something different: the right question is not which tools to buy. It is which specialist judgment calls you can now run inside a trained system, and how you wire those systems to each layer of your growth stack.
Here is how I do it, and why the architecture matters more than the tool list.
1. The stack is not a tool list. It is a set of cognitive functions.
Most founders approach AI the way they approach hiring: find a tool, give it a task, and see if it performs. That produces a fragmented collection of subscriptions that do not talk to each other and do not compound.
The better frame is functional. Every growth system requires four types of judgment:
- Acquisition judgment: which segments to target, which channels to back, how to allocate budget
- Activation judgment: how to move a new user from sign-up to first value
- Measurement judgment: what the data actually means and what to do next
- Retention judgment: who is at risk, who has upside, and what intervention fits each
For most of my career, each function required a specialist or an agency. Now each function can be architected as a trained cognitive layer. The tools are just the execution surface.
💡 Key Takeaway: Before you ask which AI tools to use, map which growth functions you are currently underserving. The stack follows the diagnosis, not the other way around.
2. Acquisition is where AI earns its cost fastest.
The acquisition layer is the easiest place to see AI impact. The feedback loop is short, and the metrics are unambiguous.
At this layer, I use AI for three things: channel strategy research, creative brief generation, and performance analysis. These are not automations. They are judgment calls that previously required a senior media strategist or a 48-hour agency turnaround.
The evidence from the region supports the case. 81% of Southeast Asian companies have moved beyond AI experimentation into pilot and scaling phases, compared with 63% globally. The acquisition layer is already running on AI at scale in this market. The question is not whether to use it. It is whether your system is designed or improvised.
What I built: a business analyst skill trained on growth frameworks, commercial logic, and my own proof points. It does not produce a generic analysis. It produces analysis calibrated to the specific constraints of a founder-led business in SEA.
💡 Key Takeaway: If your acquisition decisions still depend on a weekly agency call or a manual reporting cycle, you are not competing with the companies that have wired AI into this layer.
3. Activation and retention are where the economics actually compound.
Acquisition gets the attention. Activation and retention are where the unit economics actually work.
In Singapore, AI adoption among large enterprises rose from 44% in 2023 to 62.5% in 2024, while SME adoption rose from 4.2% to 14.5% over the same period. That uptick is not happening at the top of the funnel. It is happening in lifecycle systems: churn prediction, next-best-action logic, personalised re-engagement.
Lazada illustrates what this looks like at scale. Their 2024 whitepaper found that 88% of Southeast Asian shoppers say AI-generated content directly influences their purchase decisions, and they built infrastructure to match: AI Lazzie, a GenAI shopping assistant that handles discovery, decision support, and personalised offers across six markets. That is not a personalisation feature. That is the activation and retention layer rebuilt around behavioural signals.
Most founders I work with do not need that scale of infrastructure. They need a system that can do three things:
- Identify which new users are not activating and why
- Flag which retained users are showing early churn signals
- Recommend a specific next action rather than a generic re-engagement email
I run this through a combination of trained analytical prompts and structured data outputs. No custom model. No data science team. The architecture is the differentiator, not the budget.
💡 Key Takeaway: Activation and retention AI does not require a data team. It requires a clear decision framework and a system trained to apply it consistently.
4. Measurement is the layer most teams skip, and it breaks everything else.
You can have strong acquisition and good activation and still make bad decisions if your measurement layer is broken.
The most common failure: teams use platform metrics to make business decisions. Cost per click, impressions, and open rates. These are activity metrics. They tell you what happened inside the tool. They do not tell you what moved the business.
However, working faster on the wrong metrics is not a measurement improvement. It is accelerated noise.
What I built at this layer: a reporting and analysis skill that translates raw platform data into commercial language. Revenue impact. CAC trend. LTV trajectory. Payback period. It does not replace a data analyst. It replaces the three hours a week I used to spend reformatting dashboards before I could think clearly about what the numbers meant.
💡 Key Takeaway: AI in measurement is not about faster reports. It is about replacing activity metrics with decision-grade commercial outputs.
5. The failure mode is not bad tools. It is no architecture.
The strongest counterargument I hear from founders: “Our problem is not tools. It is that our foundations are weak. Adding AI on top of bad systems just automates the mess.”
That is correct. And it is exactly the point.
The founders who are getting nothing from AI spent money on subscriptions before they had a decision framework. They added a personalisation layer before they had clean behavioural data. They automated outreach before they had a retention model.
The AI growth stack only works when it maps to a growth system that already has a defensible structure. Acquisition feeding activation. Activation data informing retention. Retention economics shaping acquisition investment. AI sits at each node, but the architecture between nodes is what produces compounding returns.
This is not a tool recommendation. It is a design problem.
Final Thoughts: The question is not which AI tools to use. It is which functions to rebuild.
The founders winning with AI are not the ones with the longest tool list. They are the ones who asked a harder question: which specialist judgment calls in my growth system can I now run as trained cognitive infrastructure?
Acquisition strategy. Creative briefing. Activation logic. Churn prediction. Measurement interpretation. Each one is now architecturable. None of them requires a six-figure hire or an agency retainer to execute at a high standard.
The gap between those founders and everyone else is not the budget. It is design clarity.
If you want to map your current growth system against this framework and identify where AI creates the most commercial return for your specific business, book a discovery call or connect with me on LinkedIn.
A note before you close this tab. The fact that you read this far tells me something. You already sense that the way you’ve been thinking about growth might be incomplete. That instinct is worth following.
Mervyn Chua is a growth-transformation consultant helping founders and CEOs build the strategic clarity and systems to grow in an AI-first world. If this raises questions worth exploring for your brand, let’s talk.
