Driving AI Adoption at Your Philippine Hub with Forward-Deployed Engineers (FDEs)

Forward-Deployed Engineers (FDEs) — How to Make Use of This New Role for AI Success at Your Philippine Hub

We explain the new role of "FDE," which solves the problem of AI adoption stalling at the pilot stage, from the perspective of Japanese companies in the Philippines. Learn it in practical terms — the concrete five steps at a Manila hub, a sense of costs, and how to comply with the data-protection law.

---

Part 1: Why This Matters

Step 1: The Philippine Business Context (3 min)

Many Japanese companies in the Philippines use Metro Manila and Cebu as BPO (business process outsourcing) hubs, shared-service centers (consolidated hubs for accounting and HR), and manufacturing bases. At these sites, AI adoption is emerging as a management issue, but you constantly hear that "we ran a PoC (pilot), but it never reached production."

The cause lies not in the AI model's performance but in the "adoption design" — how to build AI into the work. The reason major U.S. cloud providers are rushing to hire "forward-deployed engineers" (hereafter FDEs — engineers who embed at the customer to make AI adoption succeed) is precisely this challenge. For Japanese companies with Philippine hubs as well, whether to hold an FDE-type role in-house or source it externally becomes a deciding factor in the success of future AI use.

"Imagine a scene after the morning meeting at an office in Makati, where you consult the local IT head, Mr. Juan: 'Tokyo head office wants us to scale up AI adoption from next term. But last year's PoC ended in form only. This time I want to proceed with the FDE concept.' Mr. Juan tilts his head and asks, 'What's an FDE?' This material builds a foundation that lets you explain it to someone like Mr. Juan — and to Tokyo head office's management, too."

Step 2: Key Points from the Original Article (5 min)

We have organized the facts of the original article for business professionals at Philippine hubs.

Source: Computerworld — "Here's one career emerging from the AI shift: 'forward-deployed engineers'" (May 15, 2026)

This table was created for learning purposes based on facts from publicly available information. For details, please refer to the original article linked above.

Related: see How AI Helps Philippine SMEs Build a Practical Adoption Roadmap for a detailed explanation.

Step 3: Comprehension Check (5 min)

Q1. Is the main role of an FDE (forward-deployed engineer) to write code, or to make the customer's AI adoption succeed?

Hint: The original article explains that they "focus on the customer's successful outcomes rather than on writing code."

Q2. According to LinkedIn's data, by how many times did FDE-related job openings grow between 2023 and 2025?

Hint: It is a figure that greatly exceeds the 13-fold growth in AI engineer roles.

Q3. What does the original article cite as causes of AI adoption failing at non-IT companies? Name three or more.

Hint: Four factors are given — vision, talent, budget, and the complexity of adoption.

Q4. Which company did Microsoft form an FDE-related partnership with in March 2026?

Hint: It is a major global consulting and business-outsourcing firm.

Q5. What is one future role of FDEs suggested by the Gartner analyst in the original article?

Hint: It is work related to AI usage fees (the cost of processing units called tokens).

---

Related: see How AI Partner Selection Helps Philippine SMEs Cut Project Risk for a detailed explanation.

Part 2: Putting It Into Practice

Step 4: Adoption Steps in the Philippines (10 min)

We have organized the steps for driving AI adoption at a Philippine hub using an FDE-type role into five stages.

Step 5: Common Mistakes and How to Avoid Them (5 min)

Mistake 1: "Passing Tokyo head office's instructions straight to the local site"

Bad example: You hand the AI adoption plan decided at the Japanese head office to the Manila team merely translated into English. Local work flows and language differences aren't reflected, the front line takes it as "this has nothing to do with us," and it becomes a dead letter.

Good example: Together with the local IT head, you create a version adapted to the Philippine work flow. In team meetings, you explain while showing concrete examples, and always make time for questions at the end.

Mistake 2: "Stuck at PoC (pilot), unable to move to production"

Bad example: You spend six months on an elaborate pilot, the demo succeeds, but because you hadn't decided the governance mechanisms and division of responsibility needed for production operation, it ends up shelved.

Good example: From the start, on the premise of production operation, you build something working in three months. Demand of the person in the FDE role, as the result, not a demo but "a state where it's built into actual work and can keep being used."

Mistake 3: "Putting data-protection considerations off until later"

Bad example: You input data containing customer information into an external AI service without checking. Later you become a target of an NPC investigation, and the local subsidiary's trust is damaged.

Good example: At the early adoption stage, you create criteria for judging, by type of data, "whether it may be input into AI." You set things so data isn't used for training, and ensure audit logs can be kept.

---

Part 3: Going Deeper

Step 6: Related Technical Terms (5 min)

Forward-deployed engineer (FDE) is an engineer who embeds at the customer's site and sees AI adoption through to the end. When a Japanese insurance company adopts AI document review at a Manila BPO hub, the FDE plays the role of accompanying the project from designing the work flow to the start of operations.

Agent / agentic AI (AI that carries out work autonomously) is AI that, without being instructed one by one, carries out multiple steps on its own when simply given a goal. At a Cebu accounting shared-service center, it is beginning to be used as a mechanism that handles invoice reading, data entry, and sending approval requests in one continuous flow.

Token (the smallest unit by which AI processes text) is the unit into which AI divides text, and usage fees are also calculated by this token count. Because Japanese text tends to require more tokens than English, you need to be careful with cost estimates when using a mix of Japanese and English at a Manila hub.

Guardrails (mechanisms to prevent AI from running off the rails) are restrictions placed on both input and output so that AI doesn't give inappropriate answers. When using AI for customer service at a Manila call center, you'd set restrictions such as not including personal information in answers and not quoting figures without basis.

Outcome-based pricing is a mechanism where you pay for an AI service according to the results actually achieved, rather than usage time. In the Philippine BPO industry, performance-linked contracts have long been used in part, and this thinking is spreading in AI adoption support as well.

Step 7: Applying It to Your Own Company (10 min)

Put into words why your AI adoption has stalled

Something to think about: If you have AI projects that ended at the PoC stage, try sorting their failure factors into the four categories of "vision," "talent," "budget," and "complexity." In many cases, multiple factors overlap.

Next action: List your AI-related initiatives over the past 24 months and create a one-page internal document that rates failure factors and restartability on a three-point scale.

Decide "who" plays the FDE role at the Manila hub

Something to think about: Who is best suited — a Japanese expatriate, the local IT head, or an external AI support firm? Comparing them along four axes — language, business understanding, technical skill, and cost — makes the decision easier.

Next action: For the three candidate patterns (expatriate-led / locally led / outsourced), summarize first-year costs and expected outcomes in a table, and propose it to the local subsidiary's management meeting.

Start making AI usage fees visible monthly

Something to think about: AI usage fees quietly balloon once you start using them. For a Philippine hub, dollar-denominated billing affects both pesos and yen, so you need a setup where head office and the local site can see the same figures.

Next action: Pull monthly usage fees (dollar-denominated and peso-converted) from the AI service's admin screen, and decide the format of a report shared between the local subsidiary and head office.

---

Part 4: FAQ

Q1. Which is better — growing an FDE in-house or sourcing one externally?

We recommend a two-stage approach: partner with an external support firm for the first one to two years and grow local staff in the meantime. In the Philippines, IT talent is highly mobile, and relying on a single in-house specialist tends to create a gap when they leave. Pairing an experienced external FDE with core staff who work locally for the long term makes knowledge more likely to stay in-house.

Q2. Will the time difference and language gap between the Manila and Japan teams be an obstacle to using FDEs?

The time difference is only one hour, and English is usable as a common language, so coordination is easier than with European or American hubs. However, if the Japanese head office keeps the habit of making decisions in Japanese documents, the translation burden falls disproportionately on the Manila side. Deciding a policy of creating the main FDE-related documents in English from the start speeds up local decision-making.

Q3. In relation to the data-protection law, how should we judge what data may be put into AI?

The Data Privacy Act of 2012, administered by the National Privacy Commission (NPC), sets procedures for obtaining consent when handing personal information to a third party and for cross-border data transfer. Because many AI services have servers overseas, be sure to create a procedure to consult your in-house data protection officer (DPO) before inputting customer information.

Q4. How much do the initial and operating costs of AI adoption come to in pesos?

It depends on the scope, but for a small-scale adoption targeting one department, many companies anticipate a first-year total on the order of PHP 3 million–8 million. External support costs make up more than half, followed by AI service usage fees (often dollar-denominated), in-house labor costs, and training costs. Factoring in exchange-rate swings, we recommend leaving about 10% of slack in the peso budget.

Q5. If local staff become anxious that AI will eliminate their jobs, how should we explain it?

In the Philippine BPO industry as it stands, the mainstream division of labor is that "AI takes over simple work and people concentrate on judgment and customer service." In an FDE-type approach, carefully explain in pre-adoption meetings that local staff's business knowledge is indispensable to AI design. In fact, the FDE's work itself is built on collaboration with the front line's business experts.

---

Tips for Making the Most of It (3 Tips)

Start with one task, three months, and one person in charge Starting broad and shallow means nothing reaches production anywhere. Choose just one task to tackle first at the Manila hub, and set the constraint up front of starting operations in three months. Narrow the person in charge to one, and building the result of AI adoption into that person's performance evaluation raises its priority in-house.

Always pair a local business expert with the FDE Neither the FDE alone nor local staff alone will succeed. The FDE brings knowledge of AI and adoption; local staff bring the realities of the work and the characteristics of the customers. Hold a weekly regular meeting without fail for at least the first three months, and tie the two sides' knowledge together.

Make AI usage fees visible every month, converted to pesos AI service fees are often dollar-denominated and subject to exchange-rate effects. Once you start adoption at the Manila hub, build a mechanism that shares both the dollar bill and the peso-converted amount with accounting monthly. When the cost picture is visible, head office and the local site can make decisions on the same premise.

---

Bonus: How to Use PH AI Works

PH AI Works supports the use of AI and technology for Japanese companies expanding into the Philippines and for Japanese business professionals in the Philippines. Centered on FDE-style hands-on support, we help with AI adoption rooted in local work.

Examples of what you can consult us on as a next step:

• Re-examining AI-related initiatives that stalled in the past at the Manila hub and identifying work that can be restarted
• Working together to sort out which work to start AI adoption with at BPO and shared-service hubs for the most readily visible results
• Designing in-house study sessions to convey the concepts of AI and forward-deployed engineering to local staff

Please feel free to get in touch. Consultations are free.

---

References and Sources

• Computerworld — "Here's one career emerging from the AI shift: 'forward-deployed engineers'" (May 15, 2026)
• National Privacy Commission (NPC) — the Philippine National Privacy Commission