At Digital Transformation World in Copenhagen this year, there were many cheerleaders for AI - and chief among them was Anton Bonifacio (pictured), who holds the unique dual role of Chief AI Officer and Chief Information Security Officer at Globe Telecom.
Bonifacio took the reins of Globe’s newly established AI Development and Enablement Group (AIDE) last year, and the operator has already celebrated several key milestones in its AI journey.
In this interview, Developing Telecoms explores how the Philippines-based operator is balancing rapid innovation through AI with the critical need to protect digital security in a country where millions are gaining connectivity for the first time. Bonifacio also shares how Globe is building strong foundations to scale AI efficiently and cost-effectively, ensuring it avoids the costly missteps often associated with deploying legacy technologies.
You’ve spoken a lot about Globe’s strategy on stage. Let’s start with the basics - why avoid a use case–driven approach to AI?
Anton Bonifacio (AB): Our concern, informed by past experiences with tech adoption, is that if we approach AI purely from a use-case perspective, we risk waking up in a few years with a pile of overlapping, expensive technologies. It’s already happening - vendors are rushing in, everyone’s selling something. A chatbot for HR? That’s $500,000. A contact centre AI solution? Fifteen different vendors will pitch you that.
So, you end up with 15 GenAI solutions from 15 vendors, all overlapping. TM Forum is trying to address this problem with its Open Digital Architecture (ODA), because many of these AI components are actually reusable. But if we don’t think ahead, we’ll just recreate legacy problems in a new AI form.
So how are you avoiding that?
AB: We take a “foundation first” approach - what I call our “kitchen strategy.” First, we divide use cases into two buckets: internal productivity and complex business requirements. The former can be tackled with democratised tools - we empower our 6,000 employees with Gemini and low-code/no-code platforms. For more complex cases, we do co-building between our engineering team and vendors.
The key is to build reusable components - common landing zones in AWS or Google Cloud Platform for example, or enterprise-wide data stores like Databricks. That way, we aren’t spinning up separate environments for every new use case. The infrastructure is central, not the use case.
What kind of “dishes” are you cooking up for customers?
AB: The menu is still evolving, but three focus areas stand out. First, enhancing customer interactions - reaching customers at the right moments and servicing them better. Second is data democratisation. Telcos have so much data, but very little of it is truly usable. GenAI can help make that data consumable and actionable.
And third, internal productivity - not just for developers but across the employee base. We’re investing in tools that make daily work easier and more efficient.
How has Globe evolved since you took on the role?
AB: AI isn’t new to Globe - we already had strong data and MLOps foundations through our Enterprise Data Office and Finance teams. What’s changed is the accessibility and mindset. Innovation has always been part of our DNA - GCash (Globe’s fintech arm) is a good example. That makes it easier to communicate a foundation-first AI strategy rather than just chasing short-term ROI.
So what does a “complete” AI-powered Globe look like in five years?
AB: It’s not just about what you serve - it’s whether the people consuming it are happy. In Globe, we talk about a “circle of happiness”: happy employees lead to happy customers, who in turn create happy shareholders. If we can maintain that cycle while using AI, then we’re winning.
Let’s talk about employee empowerment. You’ve given tools to 6,000 employees. Why Gemini, and how do you measure success?
AB: We’re a Google Workspace company, so Gemini was a natural choice. But it’s not the only one. We built our own RAG toolkit using Flowise, we use enterprise APIs from ChatGPT, and give staff options suited to their needs — whether that’s a simple chatbot or something more custom.
Impact is clear. Within the first 4–6 months, we had 400 bots created, most by non-technical teams. One standout was a B2B team that built a QA chatbot to analyse call recordings - no tech background required.
How are you spreading this innovation across the business?
AB: We’ve built a movement - AI advocates, internal hackathons like “Playoffs,” and constant evangelisation. The difference is our hackathons aren’t just showcases - employees can immediately build production-grade AI applications.
Are there any use cases that could transform the whole industry?
AB: Honestly, many of them are already known - call centre co-pilots, automated networks, enhanced customer service. The bigger question is: how do you execute? Piecemeal or with conviction? My view: make big bets and do it right the first time.
Let’s talk security. How do you balance innovation and risk?
AB: I wear both hats - CAIO and CISO - so I have a unique view of both sides. That gives me clarity on how high innovation can “jump” while staying within our risk tolerance. Understanding how development teams work helps me ensure security is embedded - we push DevSecOps and shift-left practices aggressively.
What kind of security challenges are unique to GenAI?
AB: About 80% are familiar - API security, application security. But there are new challenges too: prompt injection, securing LLMs, ensuring outputs aren’t manipulated. It’s encouraging to see frameworks like OWASP (Open Worldwide Application Security Project) releasing GenAI guidelines, but many CISOs are still focused on the wrong things - like whether their data is being trained on - when they should be looking at operational security of AI outputs.
Given that millions of Filipinos are coming online for the first time, how do you handle your responsibility in this space?
AB: We see digital enablement as a national economic issue. GCash has already brought many unbanked Filipinos into the digital economy. But with that comes a need for trust. That’s why we set such high standards — like completely blocking SMS messages with links. Not filtering — blocking.
What similar “high-bar” practices are you bringing into AI?
AB: Not many yet, as our AI platforms are still maturing, but the same principle applies - we won’t hesitate to make tough decisions if it protects users. Even at the cost of revenue.
You’re co-building with Google. What makes a successful co-build?
AB: Most operators say they co-build, but really, it’s co-design. Few have strong internal engineering teams to truly build with partners. That’s what we’re pushing at Globe - actual co-development. GenAI is still new - no one’s lightyears ahead. This is the time to build that engineering capability, experiment, and avoid getting locked into opaque, black-box solutions.
Is the talent pool strong enough for this?
AB: We’ve built a solid team, but we’re still learning. The important thing is to start — let teams make mistakes, let them build. The learning curve isn’t too steep yet, so the timing is good.
With billions being invested into AI, are we headed for a bubble?
AB: Maybe. But ideally we can avoid it by doing things right the first time. Foundations matter more than flashy products. We need to temper the race for short-term ROI with the long-term belief that AI will shape the next 10–30 years.
Regulation is still evolving. How are you preparing?
AB: The Philippines tends to follow the EU’s lead - our data laws mirror GDPR. If the EU AI Act comes into force, we expect similar rules. But yes, there’s a tension - you want to regulate, but not stifle innovation. It’s a balance we’re figuring out daily.
Finally, what’s the next five years of AI in telecoms going to look like?
AB: It comes down to this: can you scale your AI properly? We have the benefit of hindsight from cloud and legacy infrastructure. Now is our chance to get it right the first time - and that’s what we’re focused on at Globe.
