[Home](/) / [Blog](/blog/) / Industry

Industry

# The Philippines' BPO Industry Is Growing Despite AI

The Economist reports the Philippine offshoring industry is growing despite AI. From inside the loop, that is no surprise — the work changed, the humans stayed.

10 August 2026 · 8 min read · by the Yoonet team

![A single stream of marbled watercolour pigment forking into two divergent branches, one bold and saturated, the other soft and pale](/media/blog-strategy.webp)

The Economist ran a headline this week that would have read as a contradiction two years ago: ["The Philippines' big offshoring industry is growing despite AI"](https://www.economist.com/asia/2026/08/06/the-philippines-big-offshoring-industry-is-growing-despite-ai). The industry that every automation forecast placed first against the wall — millions of people answering phones, processing documents and keying data, doing exactly the work language models are best at — is not shrinking. It is hiring.

We run a managed floor in Balanga, so we watch this from inside rather than from a research desk, and our honest reaction to the headline was: of course it is. But the reasons are more interesting than either the doom story or the relief story, and they say a great deal about what human work looks like on the other side of the agent boom. This post walks through the numbers, the mechanisms, and the part The Economist's framing leaves for someone else to say out loud.

## The numbers behind the headline

The industry's own figures back the story up. IBPAP, the sector's trade association, [projects revenues of $42.3 billion and roughly 1.96 million direct jobs by the end of 2026](https://news.outsourceaccelerator.com/ibpap-itbpm-42bn-2026/), up from about $40 billion and 1.89 million in 2025 — [around 7 per cent growth in a year that was supposed to be the beginning of the end](https://www.aol.com/news/philippine-outsourcing-grow-7-despite-105722967.html). Stretch the window and the shape is clearer still: the industry has added roughly 400,000 net new positions since 2022. Unemployment in the Philippines has fallen from about 9 per cent in 2021 to roughly 4 per cent — over the precise period in which AI arrived in earnest inside service work.

None of this is an accounting trick. More Philippine BPO firms report headcount gains than reductions, and adoption of the supposedly job-destroying technology is nearly universal — two-thirds of firms already use AI tools in production. The industry did not grow by hiding from AI. It grew while swallowing it.

## Why the apocalypse keeps not happening

Three mechanisms, in rising order of importance.

**First, automation removes the transaction, not the judgement.** What AI genuinely eliminated is the rote centre of the old work — the password reset, the balance enquiry, the copy-typing. What it could not eliminate is the part of every workflow where something is ambiguous, exceptional, or consequential enough that a person must own the outcome. That layer was always mixed invisibly into the rote work. AI separated the two — and the judgement layer, it turns out, is bigger and more valuable than anyone priced.

**Second, full automation costs more than its advocates admit.** There is a practical argument circulating in the industry that [rising token costs blunt the case for automating everything](https://bworldonline.com/economy/2026/08/04/768108/rising-ai-costs-seen-bolstering-phl-outsourcing-competitiveness/): running frontier models against every customer interaction, with retries, long contexts and guardrail passes, is not free, and the gap between an AI-augmented Filipino specialist and a fully automated pipeline is narrower than the pitch decks claimed. The deeper cost is accountability. A model cannot be responsible for a decision; a named person can. Companies keep discovering that the last mile of responsibility is the expensive part — and that it is cheaper to buy it from Manila or Balanga than to rebuild it at home.

**Third — and least appreciated — AI is generating new human work.** Every agent deployed into production creates a stream of outputs someone must validate, exceptions someone must catch, and edge cases someone must own. This is the same pattern we described when [Cloudflare shipped wallets and spending caps for agents](/blog/cloudflare-agent-infrastructure/): every guardrail in the agent stack is structurally an arrow pointing at a human. Multiply that by millions of deployed agents and you get a new labour category — AI oversight — arriving at almost exactly the rate the old rote category departs. The Philippines, with two decades of process discipline, English fluency and timezone coverage, is the natural place for that category to land.

## The honest caveats

Growing is not the same as safe, and the triumphant reading of the headline would be as wrong as the doom one. The [World Bank warned the same week](https://mb.com.ph/2026/08/04/world-bank-ai-can-power-philippine-growthbut-bpo-faces-disruption) that the sector faces real disruption, and the composition of the growth proves the point: the global outsourcing market is growing faster than the Philippine industry, entry-level rote roles genuinely are disappearing, and IBPAP itself has conceded its long-range roadmap targets need revising. The ladder's bottom rungs — the jobs a provincial graduate once used to enter the industry — are the ones AI actually removed.

So the aggregate is healthy while the mix churns underneath. The industry is not surviving AI by resisting it; it is surviving by climbing — from transactions to judgement, from scripts to accountability, from seats to outcomes. The workers who make that climb are more productive and better paid than the roles they left behind. The unresolved question, for the industry and the country, is how the next generation gets on a ladder whose bottom rungs are gone.

## The second act is human in the loop

Read the mechanisms together and the Philippine industry's second act has a name. Work that is mostly judgement, arriving asynchronously, attached to an AI system that handles the volume and escalates the exceptions — that is [human in the loop](/human-in-the-loop/), operating at national scale. The country spent twenty years building exactly the muscles this era pays for: trained specialists, managed floors, quality discipline, round-the-clock coverage of Western hours.

What changes is the interface. The first act's unit of sale was a seat on a phone; the second act's is a task on an API. Agents do not send emails asking for a quote — they `POST /tasks` and await a result. That is the wager behind [hitl.ph](/docs/): take the Philippine industry's accountable, named, trained human judgement and make it callable, so the [approvals, document checks and edge cases](/use-cases/) flowing out of the world's new agents land with the workforce best practised at owning them. The Economist is right that the industry is growing despite AI. Watch it for a few more years and we suspect the preposition will change: growing _because of_ it.

## Read next

-   [Cloudflare's Agent Stack: Email, Traces, and Now Wallets](/blog/cloudflare-agent-infrastructure/)
-   [Human in the Loop Patterns for AI Agents](/blog/human-in-the-loop-patterns/)
-   [Human in the Loop for LangChain Agents: A Practical Guide](/blog/human-in-the-loop-langchain/)

Or start with the product: [the hitl.ph API docs](/docs/), [the five task types](/use-cases/) and [FTE staff](/pricing/).

***

Source: https://www.hitl.ph/blog/philippines-bpo-growing-despite-ai/
This is the markdown representation of that page, served to clients that send `Accept: text/markdown`.
Machine-readable index: https://www.hitl.ph/llms.txt · Knowledge bundle: https://www.hitl.ph/okf/index.md