Strategy · 4 min read
From AI pilots to enterprise-wide deployment: four lessons from my AIMX Singapore panel
I joined AIMX Singapore as a panellist to discuss why so many enterprises stall between AI pilots and real, organisation-wide deployment. Here's what I argued on stage, and why.
Originally published on LinkedIn
Key takeaways
- Treat AI ecosystem partners like dating, not marriage. Long lock-in deals can cost more in lost revenue lift than they save.
- If you can't define the ROI of an AI use case, you probably never defined the ROI of that function without AI.
- Don't wait for perfect data. Start with proxies you already have, then refine.
- The AI playbook is an evolution of digital transformation. If you skipped that step, build it now, with AI in mind.
In August 2026 I joined a panel at AIMX Singapore, hosted by AIMX Network, alongside Olivier Gergele, Walter Lee, Jim Lim and Miguel Ho. Our brief was a question almost every enterprise leader is asking right now: how do you get from AI pilots to enterprise-wide deployment?
A few points from the discussion stayed with me. These are the ones I made on stage.
Date, don't marry.How should enterprises choose AI ecosystem partners?
The analogy I gave was simple: "Date, don't marry."
Conversations about AI partnerships tend to follow a familiar script. Do we commit to this? Do we build this or buy that? CIOs and CDOs need to break out of the traditional model of "let's secure a 10-year deal that's more efficient."
With AI advancing as fast as it is, today's most advanced model could be tomorrow's redundancy. The 20% you think you're saving with a 10-year deal could cost you the 50% uplift in revenue effectiveness you'd have gained by staying flexible in what you adopt.
The value definition was never there.Why is AI ROI so hard to define?
ROI matters when you use AI to enable business processes. Yet enterprises often get stuck because they can't define the ROI of making something better, faster or more effective.
Why? This isn't an AI problem. It points to something more fundamental: the business never defined the ROI of that function in the first place.
Take AI-powered business intelligence. When the CFO asks what the ROI is, more often than not the company has no way to calculate the ROI of business intelligence at all. They know they need it. But because AI is an additional spend, the ROI question gets put in the spotlight.
So the real question to ask is this: what is the ROI of what you're doing without AI, and what is it with AI?
No. Start with proxies.Do you need perfect data before scaling AI?
Everyone will tell you data is the foundation of any scaled AI deployment. True. But what exactly about data? Is it interoperability? Access? Quality? Or all of the above?
And what happens if you don't have it, or it isn't clean enough, or it takes too much effort to get right?
Start with proxies, meaning what you already have, and then refine. If you wait for 100%, you'll never get there, or you'll spend too much trying.
It's the next chapter.Is AI transformation different from digital transformation?
This was my personal favourite point of the session:
"The playbook is an evolution of the digital transformation journey. If you didn't have that, build that first, with AI in mind."
If you'd like to talk through what this means for you or your organisation, WhatsApp me or email me. I'm happy to chat.
The newsletter
Weekly insights in digital, data and AI.
Work with Konverta.ai
Stuck between AI pilots and real deployment?
Talk to Konverta.ai about your AI partners, the ROI case and the data you already have, and how to scale AI across the organisation.
Prefer a form? Contact us.
