Klarna went all in. Salesforce laid off 4,000 people. Jack Dorsey cut 40% of his company. What's really behind those headlines — and why it could be an opportunity for you, not a threat.
It's not just a buzzword. This is really happening.
In 2024 and 2025, the world of work began to change in ways that were previously only speculated about. Major global companies started openly saying what would have been unspeakable just recently: we're replacing people with AI agents.
Let's look at the concrete numbers.
Klarna
700
full-time employee equivalent replaced by AI
Saving $40M/year • resolution time: 11 min → 2 min
Salesforce
4,000
laid off in customer service
AI handles 50% of interactions • costs down 17%
IBM
7,800
positions earmarked for AI replacement
$4.5B in savings by end of 2025
Amazon
14,000
corporate positions eliminated in October 2025
CEO Andy Jassy confirmed it in an internal memo
Microsoft
6,000
laid off, many of them engineers
30% of Microsoft's code is now written by AI
Duolingo
AI‑first
new company strategy since 2025
Only hiring where automation isn't enough
Jack Dorsey and Block: 40% of the company gone because of AI
In February 2026 came perhaps the most radical move of all. Jack Dorsey, founder of Twitter and CEO of Block Inc. (parent company of Square and Cash App), announced the layoff of approximately 4,000 employees. That's nearly 40% of the entire company.
Dorsey didn't sugarcoat it. In his letter to shareholders, he wrote plainly:
"
Intelligence tools have changed what it means to build and run a company. A significantly smaller team can now achieve more and perform better.
Jack Dorsey, CEO Block Inc. — shareholder letter, February 2026
Dorsey went even further. He predicts that most companies will reach the same conclusion within the next year. His goal, in his own words, is to build a "smaller, faster, intelligence-native company."
Wall Street's reaction to the layoff announcement? Block's stock went up. Investors see this as a signal of efficiency, not weakness.
"
AI agents now handle 50% of all customer interactions. Our support costs dropped by 17%.
Marc Benioff, CEO Salesforce
What do the global numbers say?
These aren't isolated examples. It's a trend that's happening globally and accelerating.
AI impact on the job market in numbers (2025)
Companies planning to reduce headcount due to AI41%
Companies already automating human work with AI37%
It's clear that this is not a question of "if." It's a question of "when" — and more importantly, "how well."
But wait. Not every story doesn't end with success.
Here's where it gets really interesting. Not every company did it right.
Klarna made a mistake and had to backtrack.
After massively deploying AI in customer service, customer satisfaction started to decline. Interactions were fast, but customers felt short-changed. They missed empathy, contextual understanding of complex issues, and simply a human touch. In 2025, Klarna had to start rehiring human agents. They explicitly admitted that cutting costs on customer service was a mistake.
By Q3 2025, their AI agent was still doing the work equivalent of 853 employees and had saved a total of $60 million. But the company returned to a hybrid model where AI handles simple queries and humans take care of more complex cases.
Timeline: Klarna and AI
Early 2024
AI assistant replaces the work of 700 employees. $40M saved. The company celebrates.
Mid 2025
Customer satisfaction drops. Complaints about impersonal and frustrating responses.
Late 2025
Klarna resumes hiring humans. CEO admits the mistake. Transition to a hybrid model.
Q3 2025
AI still does the work of 853 employees, total savings $60M. But with humans alongside it.
The lesson from Klarna's story is clear: deploying AI on everything at any cost is the wrong approach.
So what does this mean for your business?
Companies that use AI most effectively don't do one thing: they don't mass-fire people and replace them with a chatbot. Instead, they identify specific processes where AI delivers the most value and deploy it precisely there.
Here are the types of processes where AI agents excel:
Customer support (first line). Repetitive queries, order tracking, FAQ, complaint forms.
Lead generation. Qualifying potential customers, first contact, data collection.
Phone line. Voicebots that can handle dozens of calls at once, 24 hours a day.
And where do humans still win?
Complex problems requiring context and empathy
Business negotiations and long-term relationships
Creative and strategic work
Communication in crisis situations
AI agents don't replace people. They replace routine.
This is exactly the nuance that separates companies afraid of AI transformation from those that use it as a competitive advantage.
A properly deployed AI agent frees your people from monotonous, repetitive work. Instead of having your customer service team answer "where's my order?" for the thirtieth time that day, they can focus on customers with real problems and build relationships that lead to loyalty.
That's exactly what we do at Chatbot.Expert. We don't implement AI to replace people. We implement it so people can do work that has real value.
What does a proper implementation look like step by step?
The most successful companies we've worked with approached AI systematically. Here's the approach that works.
1
Map your processes
Take one week and track what your team repeats. How many calls, emails, and chats are about the same topic? That's where your first opportunity lies.
2
Start with a small pilot
Don't jump straight into your entire customer service. Pick one channel, one type of query, and test AI there. Klarna made this mistake and paid the price.
3
Set satisfaction metrics
Track not just response time and costs, but also NPS and customer satisfaction. Numbers will alert you before you feel the problem in reviews.
4
Define when to hand off to a human
Every good AI agent must know when it's not enough. Set clear rules for escalation to a live operator.
5
Scale what works
Once the pilot shows good results, expand it. Add channels, languages, query types. That's how you build a sustainable AI transformation.
Is AI right for your process? Find out in 30 seconds.
Check what applies to your team. The more items you check, the greater the AI potential for your business.
0 / 6 points
Frequently Asked Questions
An AI chatbot is a conversational window on your website where customers type questions and get answers. It's a visible assistant for customer communication. An AI agent is an invisible background process that works with your data, connects via API to your business systems, and autonomously executes tasks and automates business processes.
It depends on complexity. A simple chatbot can be ready in a few days. More extensive solutions with CRM, ERP, or e-commerce integration take longer. We aim for the fastest and smoothest deployment possible.
No. The most successful companies use AI so their people can do more valuable work, not as a direct replacement. Klarna tried otherwise and had to backtrack. The right approach is a hybrid model where AI handles routine and people handle relationships.
Yes, the chatbot automatically adapts to the language the customer starts communicating in. We support Czech, Slovak, English, and dozens of other languages.
Yes. An AI agent or chatbot works around the clock and can handle customers and queries day and night, without breaks and without sick days. That's one of the biggest differences compared to a human team.
We take care of the chatbot for you. We regularly update and optimize it within the agreed scope of services. You focus on your business.
An API is a standardized interface that allows systems to exchange data and execute precisely defined operations. This enables an AI agent to read and write information in your e-shop, ERP, or CRM and automate business processes.
Where to start?
If you're considering deploying an AI agent in your company, start by answering three questions:
Which processes are slowing your team down the most because they're repetitive and predictable?
Where do you spend the most time on communication that's structured and doesn't require creativity?
What are your monthly costs for these processes?
If the answers point to dozens of hours and thousands of dollars per month, it's time to talk.