Mira worked at a business process outsourcing center in Manila for four years. She started as a night-shift customer service agent answering basic billing questions for a telecom client, worked her way up to senior agent, then team lead. She had health insurance, a steady paycheck, and was saving to send her daughter to a better elementary school.
Then the AI voice system arrived.
At first, management framed it as a helper. Something that would take the easy calls so agents could focus on the harder ones. Mira and the other top performers on her floor — the ones with the highest CSAT scores — were asked to record hours of their voices. Different tones. Apologetic. Reassuring. Firm but polite. "It's for training data," her project manager said. "You'll get a small bonus."
She sat in a soundproof booth for three days, essentially teaching a machine how to sound like her — how to comfort an angry customer the way she did, how to say "I completely understand your frustration" in the exact tone that once won her Employee of the Month.
The rollout happened faster than anyone expected. Within weeks, the AI was handling 74% of all incoming calls. First response time dropped from minutes to seconds. Resolution time fell by half.
And then the layoffs came.
Of the 40 agents on Mira's floor, 36 were let go. Aiza didn't make the cut. Neither did Carlo, a soft-spoken agent who used to sing old Eraserheads songs under his breath during slow hours. On his last day, he stood by the elevator and said, "Sana all makakuha ng bagong trabaho agad" — hoping everyone finds new work quickly — and then he was gone.
Mira was one of four kept for escalations — the cases the AI can't handle, the customers who demand a real person. She now sits in a booth that used to hold four agents and holds just one. Her work-life balance is, ironically, better than ever, because the AI absorbs the volume that used to burn her out so badly she cried in bathroom stalls.
Sometimes, late at night, a customer calls in furious and then says, "Finally, a real person," relief flooding their voice, like she's rescuing them from something. She doesn't tell them she's one of four humans left on a floor that used to hold forty. She doesn't tell them that somewhere in the system is a version of her own voice, trained on her own patience, still answering calls she will never hear.
She just says, "Yes po, I'm here now, let's sort this out together." And for a moment, that's enough.