The Last Caller
After fifteen years handling customer complaints, Maria takes her final call before the AI takes over—and discovers what it means to truly be heard.
The clock on Maria's cubicle wall read 4:47 PM. Thirteen minutes left. Fifteen years, three months, and seventeen days, down to thirteen minutes.
She adjusted her headset for what might be the last time and glanced at her monitor. The queue showed 47 waiting calls. Tomorrow morning, the AI would handle all of them. No wait times. No hold music. No Maria.
Her screen blinked with an incoming call. She clicked accept without thinking, muscle memory from thousands of identical motions.
"Thank you for calling Velocity Wireless, this is Maria, how may I help you today?"
"Finally! I've been on hold for forty-five minutes. Forty-five! Do you people even—"
"I apologize for the extended wait time, sir. I understand how frustrating that must be." The words came automatically, smooth as water over stones worn down by years of repetition. "Let me see how I can help you today."
She'd said variations of that sentence approximately 127,000 times. She'd done the math once, during a particularly slow Tuesday afternoon in 2019. The number was probably higher now.
"My bill is wrong. It's been wrong for three months, and every time I call, someone promises to fix it, and then next month it's still wrong. I'm paying for premium data I never ordered, and I want my money back."
Maria pulled up his account. James Weatherby, 2847 Oak Street, Cincinnati. Customer since 2014. She could see the notes from his previous calls—four of them, same issue, different agents. Each note ended with "escalated to billing" or "will be resolved next cycle."
"I see the issue, Mr. Weatherby. I'm very sorry you've had to call multiple times. Let me fix this for you right now."
She navigated through three different systems, her fingers moving across the keyboard with practiced precision. The override codes, the credit adjustments, the service tier modifications—she knew exactly where everything was, how everything connected. She'd learned the systems when they were new, watched them evolve, memorized every shortcut and workaround.
The AI would know them too, of course. Instantly. Perfectly. But it would never need shortcuts, would never have to remember when the billing system used to process refunds differently, would never carry the institutional knowledge of thousands of similar calls.
"Okay, Mr. Weatherby, I've removed the premium data package from your account, backdated the adjustment to when it was originally added in error, and issued a credit of $267.43 to your account. You'll see that on your next statement, and your monthly bill will be $89.99 going forward, same as it was before."
Silence on the line. Then: "Really? Just like that?"
"Just like that. Is there anything else I can help you with today?"
"No, that's... thank you. I mean, really, thank you. I was ready to cancel, honestly. But you actually fixed it."
"You're very welcome, Mr. Weatherby. Enjoy the rest of your evening."
She ended the call and sat back. Twenty-three seconds over her average handle time, but her supervisor wasn't watching anymore. The Performance Dashboard that had governed her days for fifteen years had been turned off last week. No more metrics. No more targets. Just thirteen minutes left to be useful.
Another call blinked in. She clicked accept.
"Thank you for calling Velocity Wireless, this is Maria, how may I help you today?"
"Oh thank God, a real person. I thought I was going to get that robot thing. My phone just died, completely dead, and I have pictures of my granddaughter on there that I never backed up, and the guy at the store said they're gone, but they can't be gone, she's only three months old and I—"
The caller's voice broke. Maria could hear the tears.
"Ma'am, I'm so sorry you're going through this. Let's see what we can do. Can you tell me what happened?"
The story tumbled out. Phone dropped in the sink while doing dishes. Wouldn't turn on. Store technician said the logic board was fried. Pictures gone. Videos gone. Her daughter lived six states away, and these were the only photos she had from the last visit.
Maria knew what the policy said. Water damage wasn't covered. Customer should have enabled cloud backup. Nothing could be done. The AI would explain this efficiently, sympathetically, and move to the next call.
"What's your daughter's name?" Maria asked.
"What? Oh. Sarah. Why?"
"And your granddaughter?"
"Emma. Emma Rose. But I don't see how—"
"Hold on just a moment." Maria put the call on hold and opened her personal email. She found what she was looking for—a technical support forum she'd bookmarked three years ago, discussing recovery methods for water-damaged devices.
She pulled the call back. "Ma'am, I can't promise anything, but I want you to try something. Do you have rice at home? The regular kind you cook with?"
"Rice? Yes, I have rice."
"Here's what I want you to do." Maria walked her through it—power off the device completely, remove the SIM card, submerge it in rice, wait 48 hours minimum. Then there was a technician, a third-party repair shop in the caller's city, who specialized in data recovery from water-damaged phones. Maria had found a review thread about them last year when her own niece had a similar problem.
"I'm texting you the address right now. His name is Marcus. Tell him Maria from Velocity Wireless sent you. He might be able to recover the photos even if the phone doesn't work."
"You can do that? Send me information about another company?"
No, she couldn't. It was explicitly against policy to recommend non-authorized repair services.
"I just did," Maria said. "Give it a try. And ma'am? Even if the rice trick doesn't work, Marcus is really good. He recovered photos from my niece's phone after it went through a washing machine."
"I don't... I can't thank you enough. I was so upset, and you actually listened, and you're helping me, and..." The tears were back, but different now. "Thank you."
After the call ended, Maria sat very still. Seven minutes left.
She'd been written up twice in her career. Once for excessive empathy—spending too long on calls trying to help customers with problems outside her scope of support. Once for going off-script during a particularly difficult interaction with someone who'd just lost their job and couldn't afford their phone bill. Both times, her supervisor had explained that empathy was good, but efficiency was better. Time was money. Metrics mattered.
The AI would never get written up. It would be perfectly empathetic within the exact parameters defined by the efficiency algorithms. It would sound caring without actually caring. It would solve problems without ever feeling the weight of them.
Another call came in. Maria clicked accept.
"Thank you for calling Velocity Wireless, this is Maria, how may I help you today?"
"Hi, I think I need to upgrade my plan? I'm traveling next month and I need international coverage, but I don't really understand the options."
Maria smiled, even though the caller couldn't see it. She'd learned early in her career that smiling changed your voice, made it warmer. The AI would probably simulate this too, modulating its tone based on acoustic analysis of human emotional states.
She walked the caller through the options, explaining not just what each plan cost but when it made sense, when it didn't, which countries had better local SIM options, which airports had good WiFi. She told him about the time she'd traveled to Portugal and made the mistake of using data without a plan and came back to a $800 bill. She made him laugh.
At the end of the call, he said, "You've been really helpful. I'm glad I got a person instead of a robot. Those AI things never really understand what you're asking, you know?"
After he disconnected, Maria sat looking at her screen. Three minutes left.
She thought about the trainers who'd shown her the systems fifteen years ago, most of them laid off in the first wave of optimization five years back. She thought about the cubicles around her, empty now except for three other veterans finishing out their final shifts. She thought about the new AI—"Aria" they were calling it—and the demos they'd shown the team last month. It was impressive. Fast. Accurate. Never tired. Never frustrated. Never sad.
One minute.
The queue still showed 42 waiting calls. By 9 AM tomorrow, Aria would have cleared them all. Average handle time: 187 seconds. Customer satisfaction: projected 94.2 percent, up from the current team average of 89.7 percent. Cost per call: $0.31 versus $11.50 for human agents.
The math was simple. The decision was rational. The future was inevitable.
Thirty seconds.
One more call blinked in. Maria's hand hovered over the mouse.
She clicked accept.
"Thank you for calling Velocity Wireless, this is Maria, how may I help you today?"
"Oh, hi, I just wanted to say thank you. You helped me last week with a billing issue, and I just got my statement and it's all fixed, and I wanted to call and tell you that you were great. That's all. Thank you."
Maria's throat tightened. In fifteen years, she could count on one hand the number of people who'd called back just to say thank you.
"That's... that's very kind of you. I'm so glad I could help."
"You did more than help. I was having a really bad week, and you were patient with me, and you listened, and you made it right. I know you probably have a thousand other calls to handle, but I wanted you to know it mattered. You mattered."
The clock on her screen turned to 5:00 PM.
"Thank you," Maria said quietly. "You have no idea how much I needed to hear that today."
After the call ended, Maria sat in her cubicle, headset still on, looking at the blank screen. Around her, the other remaining agents were standing up, removing their headsets for the last time, gathering their personal items from drawers and walls.
She thought about the last caller—someone who'd taken time out of their day to call back, to say thank you, to remind her that behind every ticket number and account ID was a person. Someone who needed help. Someone who needed to be heard.
The AI would be faster. More efficient. More cost-effective. It would solve problems with ruthless precision and perfect memory. It would never get tired, never get frustrated, never need a break.
But it would never call its niece to recommend a good repair shop. It would never share a travel story to help a nervous customer feel less alone. It would never know what it felt like to hear someone say, "You mattered."
Maria removed her headset and placed it carefully on her desk. She picked up the framed photo of her family—the only personal item she'd kept in her cubicle—and walked toward the exit.
At the door, she paused and looked back at the room full of empty cubicles. Somewhere in a data center, Aria was probably already running, processing the queue, optimizing for efficiency.
Maria hoped the AI would be good at its job. She hoped customers would get their problems solved. She hoped the metrics would improve.
But she also hoped that somewhere, somehow, there would still be room for the kind of help that couldn't be measured in handle time or satisfaction scores. The kind that came from one person choosing to see another person's pain and doing something about it, even when the policy manual said otherwise.
She walked out into the parking lot. The January sun was setting, cold and clear. Tomorrow, she'd start looking for whatever came next. But tonight, she'd remember that last call. That last thank you.
That last reminder that for fifteen years, three months, and seventeen days, she'd mattered.
This story explores themes discussed in my article How AI Will Replace Customer Service Representatives: Timeline 2026-2028.