Every day headlines scream that AI is coming for your job. Another panic-inducing article about mass layoffs. Another breathless prediction about the end of human work as we know it. But here is one simple truth. Most of these stories are built on noise, not facts. Noise about the models. Noise about AI layoffs. Noise about how AI can replace us all. Hype, not reality.

And I'm starting a new series of videos dedicated to AI hype versus reality. This series isn't about telling you that everything is fine. It's about showing you what's actually happening. In each episode, we'll follow the same format. First, we'll dive into the hype, the headlines that dominate your social feeds and water cooler conversations. Then we will expose the reality, what the actual data is showing, stripped of emotion and agenda. Then we will talk about the human factor, the real unspoken forces shaping the changes that nobody talks about. And finally, I will give you actionable strategies on how to stay afloat.

Here's what I'm going to prove to you in the next 20 minutes. If you work in customer service, support, customer success, or any other client-facing role in tech, you're not about to be replaced by AI. Not even close. But you should be worried, just not from the direction everybody's screaming about. I'm going to show you the real data behind the AI layoff headlines, exposing what's actually happening at companies claiming to go AI first, and give you the strategy that will make you more valuable, not less, as AI adoption accelerates. If you're ready to stop the panic and start preparing, let's dive in.

The hype. AI replacement support and other customer service roles has been one of the most discussed and controversial topics in tech throughout 2025. The narrative is driven by major layoffs at tech giants Amazon, Microsoft, Meta, Intel, and others attributed directly to AIdriven efficiencies and automation with tens of thousands of jobs cut in the first half of 2025 alone. This way of job cuts emerges against a backdrop of companies aiming to trim costs and leverage AIdriven efficiencies and restructuring efforts. These layoffs were not mere reactions to financial constraints, but were indicative of a broader trend in which white collar positions long considered secure were now vulnerable.

So this is a headline and now we're going to move past the big screaming title and read without doing any further research. Just read what's written in the article. And what we're going to see is that all of a sudden the 2025 layoffs are not being attributed purely to AI. Let's talk about the reality. We're going to open Techrunch's comprehensive list of 2025 layoffs in tech and go company by company and I will prove to you why AI replacing customer service is a lot more of a panic mode rather than the reality.

Let's start with the big ones. Bumble announced in an SEC filing that it will cut approximately 240 jobs or 30% of its workforce to enhance operational efficiency and allocate the resulting savings to the development of new products and technologies. According to a CNBC report, the layoffs will help the online dating app save $40 million annually per the report. Now, from here, let's look at the CNBC report. Shares of the dating company have plunged since their debut on the public markets in 2021. Its market value has plummeted from $7.7 billion to about $673 million as of Wednesday's close.

Now, do you see anything here mentioned about automation or anything saying 100 support specialists were replaced by AI? No. But the noise is telling you you Bumble is slashing 30% of staff. AI is coming for your job. But when you open the source, all of a sudden the big and hairy AI monster is not so big and hairy anymore. And if you put yourself in the shoes of the Bumble founder and you realize that the company is public and your shares have plummeted, you're going to get your together and restructure to keep your company alive. So once again, you hear layoffs, but if you look into why the layoffs are happening, it actually makes sense and they have nothing to do with AI.

As someone not interested in the online dating and not keeping pulse on what's happening in the online dating world, even I know that online dating trend is on the decline. I come across physical ads all over the city about the offline events for singles, cooking classes, dance classes, drinks, outdoor film festivals. So, no, I am not surprised that the profits of Bumble are dropping. And AI has absolutely nothing to do with these layoffs.

Moving on, let's look at a bigger guy, Google. Looking at this title, Google reportedly cuts smart TV budget and prioritizes AI and YouTube. On June 23rd, Google reportedly reduced its investment in its smart TV division, specifically cutting the budget for Google TV and Android TV by 10%. Did anybody hear AI or automation layoffs? Let me bring you back to where we started for a second. We started from Fang is laying off people because of AI. But when you open the links, when you open the sources, turns out that the layoffs have nothing to do with automation. And if anything, they're creating jobs in the different division.

All right, let's look for different lengths. Let's take another one. Microsoft. Microsoft is contemplating additional layoffs that could happen by May. Business Insider reported the company said to be discussing reducing the number of middle managers and non-coders in a bit to increase the ratio of programmers to product managers. First of all, as a product manager, I am happy that the PM market is finally starting to balance itself out. The amount of people who got a PM title during pandemic hiring surge was ridiculous. As someone who fought for a PM job for years, I feel like my job got really devalued more than an Argentinian peso during pandemic when people decided that PM job is writing Jira tickets. Point is, this is not automation. This is big tech correcting their own overhiring decisions from 5 years ago.

Moving on, Canva. Canva is the only real example of people being laid off because their job is now being done by AI. Now in Canvas case, it only covers technical writers and today is about customerf facing rules. So we're not discussing technical writing in this episode. I will make a separate episode about UX. But in a nutshell, yes, technical writers do have a legit reason to worry. But again, this is not customer service and not support.

Okay. Now, let's pick a company that is highly dependent on customer service. HelloFresh, a meal kit company, is closing one of its sites in North Texas and cutting over 207 workers. HelloFresh is shuttering its distribution center and consolidating to another site in Irving to manage the volume in the region. Again, no mention of automation.

Okay, maybe this is all a lie and I'm not looking well enough. Fine, I look better. I'm searching specifically for AI related and automation related layoffs. Looking at AI job displacement of 2025, once again focusing specifically on customer service rules. AI chatbots reduce telemarketing costs by 80% making human customer service rapidly obsolete. When did you last speak to a human when calling customer support? Exactly. Well, I do remember the last time I spoke to a human and pretty much every single time I need to get in touch with support, I do speak to a human.

Also, read this again. AI chatbots reduce telemarketing costs by 80%. Yes, in theory they do. And we're talking about very well-trained and tuned chat bots. Not a word about support, delivery managers or implementations. The only example of customer service being replaced by AI that I found in my research and the one that I can attest to is Amazon. I think the way Amazon handles their customer support queries is very well automated. But even then, there are plenty of use cases where the chatbot will not do the job. I have personally called Amazon support several times, especially when it came to billing problems.

I remember once ordering a gift for my friend in Germany and I had to make that purchase on the German Amazon. So to order a product off of Amazon in a different country, I had to create a country specific Amazon account. So the whole international user flow isn't really solved. And when I had issues with that order, the chatbot wasn't able to help me. On top of it, I was not able to resolve this issue with Amazon Canada because I had to call the German Amazon to resolve the issue.

And now I would like to take a closer look at a very interesting case that happened with a Swedish company called CLA, which is a major European fintech unicorn, and study their case a little bit closer. What happened back in 2022? CLA made headlines when it laid off around 700 workers, which is a significant portion of its workforce at the time. The CEO was vocal about embracing AI tools that could take over tasks such as customer support, translation, content creation, and even executive level decision-making. Fast forward to 2024, and things did not go as planned. Clara got a surge in customer complaints, a major dip in user satisfaction, loss of users, and growing frustration on social media. And all of this was because of the dominance of AI and customer support.

Customers complained that automated responses were too generic, repetitive, and simply unhelpful when dealing with real life problems. Things that were specifically cited, impersonal, and repetitive AI interactions, inaccurate or confusing answers, difficulty resolving complex or sensitive issues, loss of empathy and human touch. Now, did Clara rehire the roles that let go two years prior? Yes, Clara is now actively rebuilding its human support teams.

So key lessons learned from Clar's experience. AI should assist, not replace human interactions. It is great when used for pre-screening for support agents. Before a human agent becomes involved, Clard's AI system performs a comprehensive pre-screening test to collect and synthesize critical information. Details like the name, account history, transactions, previous interactions, preferred language, sentiment, and emotional tone of the customer's message.

I'd like to add a few points to this. From my personal work experience, I have worked on an AI agent for customer service teams that handles basic customer complaints or questions. But for anything that's worth escalating, it acts as an assistant to the support agent almost like a real-time Chad GBT. The agent analyzes customer question and either provides a template answer to the support agent that they can send in the meantime while it's looking for the answers across the knowledge base or CRM and offers a support agent several variations of the response that it was able to find. It is then on the support analyst to determine accuracy of the information and add nuance or details they deem necessary.

As a result, nobody lost jobs. Customer service folks improved the speed of response by almost 20%. This is especially important for companies whose customer service teams are offshored and where the language of the company is not the native language of the customer service agent. Are there more examples on top of CLA? Yes, I got to mention that the examples I'm going to be providing are not specific to customer service roles. I was looking for examples of pure CX rule replacement, but please understand that I work with secondary data and I got to operate with what I have. The results are often published for groups of roles, customer success and support, project management and account management. So, just keep that in mind.

IBM IBM replaced over 8,000 HR and support staff with its in-houseuilt AI platform, Ask HR, aiming to automate 94% of internal queries and documentation. By automating these processes, IBM saved an impressive $3.5 billion across 70 different roles, providing significant efficiency gains. But IBM soon realized that complex HR issues such as policy judgment, exceptions, and grievance resolutions were not being solved by AI either efficiently or accurately. Reports surfaced that IBM began rehiring hundreds of staff not for basic support, but for roles demanding empathy, discretion, and nuance decision-making. As a result, IBM shifted to a hybrid model, combining AI for repetitive tasks with human specialists for complex cases.

Let's look at Dolingo. I would say that Dolingo is more of a bad example in terms of framing and positioning of AI. Dualingo announced a shift to an AI first strategy in April 2025, laying off or not renewing contracts for a significant portion of its support and content moderation teams, replacing them with AI tools for content creation and user support. The move caused an intense backlash from users and employees with concerns over declining content quality and loss of the human touch. Shortly after, they faced bad publicity and push back from the general public on social media after announcing the move, particularly on Tik Tok. The top comments on virtually every recent post have nothing to do with the video or the company and everything to do with the company's embrace of AI. As a result, Dolingo wugged back on its AI only approach, emphasizing a hybrid model where AI augments but does not replace human workers.

And now after we have realized the headlines versus what the companies are actually doing and saying, I would like to talk about a very real risk for customer service and tech and what AI is actually doing is that it's exacerbating that risk and the risk is offshoring and nearshoring. We spoke about nearshoring trends at length in multiple videos in the where to run series. But what I would like to highlight for all customer service teams is that if there is anything you should be worried about in the near future, it's the offshoring. Because one of the things that was an obstacle before was the language barrier.

For North America, the top destinations for offshoring customer success is Latin America and India. In India, English is commonly spoken, but the time zone is an issue. Latin America overlaps for timing, but English is their second language and customer service teams are very much language dependent when it comes to escalations, complaints, price negotiations, etc. With AI, this risk goes down significantly because even when used for assistance, LLM can translate and interpret the nuances and the tone to a non-native speaker. So the likelihood of offshoring goes up not because AI is going to replace you but because AI is making your job easier.

Now let's talk about the real advantages of AI and customer service teams to keep things fair. AI customer service can provide numerous substantial advantages when implemented properly. It boosts operational efficiency. It can handle thousands of repetitive tasks without fatigue. It can also scale to adjust quickly during peak demand, eliminating long wait times and improving the speed of customer service. AI systems are operational around the clock. They provide 247 service around the globe. Advanced data analytics can make it highly personalized and tailored to customer behavior, purchase history, and usage patterns. AI customer service can do rapid troubleshooting and access vast knowledge bases instantly to resolve an issue quickly.

And now let's talk about the downsides. Implementation complexity is a significant challenge. It requires substantial adjustments in organizational workflows and culture. Misalignment can lead to frustration which really undermines brand loyalty. Now, if you read up on the biggest implementation challenge, you will see that the most critical issue is being attributed to the lack of human interaction and how it impacts customer satisfaction. I will address this in the next part, the human factor. But I personally disagree with this. I don't think that the lack of the human touch is the biggest problem. I think it really depends on the customer demographic and the nature of the app.

The products where the demographics is younger, so for example, between 18 and 30, the lack of human connection is not the biggest disadvantage really. To me personally, it isn't. I would love for the chatbot to fix all my issues and not having to deal with people and not having to call anyone for simple admin tasks like checking the status of my support ticket or closing a debit card. At the same time, as someone who is currently working on a SAS product where the demographics is 40 years and older, I can see a very clear behavioral trend to prefer a human in the support department over a chatbot. So, all I'm saying is that in my view, the trend around a lack of human touch is debatable.

Technical errors and malfunctions are an inherent risk in AI systems. And the cold turkey switch to AI can very well damage customer trust and the company's reputation, especially if it's a company that's been on the market for a long time and offers a product suite similar to competitors. Privacy and ethical concerns from customers are additional critical issues given AI's extensive reliance on personal data.

The human factor. There are several things that play in your favor. Customer experience and trust often suffer when companies go AI only and companies start suffering reputational risk even before they start replacing jobs. I think that people who work in tech, people who build AI apps are prone to think that everyone around them is well-versed in what's happening in the AI world. So when a companies make announcements like that, to someone who understands how AI works, it looks like an interesting experiment that they're going to watch closely. to someone outside of the tech bubble. It sounds like this company is depriving people of jobs.

Dolingo is actually a very good example. Dolingo says that much of the feedback is coming from people who don't understand what AI first means. A lot of feedback we've seen comes from a place of passion for Dolingo, which we really appreciate. To clarify, AI isn't replacing our learning experts. It's a tool they use to make Dolingo better. Everything we create with AI is guided by our team and learning design experts. But to people who are already skeptical about AI, this statement is just noise.

Companies in general remain excited about the potential cost savings of AI, sometimes for good reason. Dualingo stock, for example, is at an all-time high. But as bullish as the executives might be, however, that excitement has not made its way into the consumer space. Now, on top of this, add corporate politics. And this mostly applies to large corpse, especially legacy corpse where corporate politics plays probably a larger role than the products those companies produce. Now put yourself in the shoes of a middle manager or they decide to cut all ties with the existing support staff and then a few months later they get a backlash like CLA, IBM or Dolingo. Can you imagine the level of embarrassment and the threat of job loss they're going to face?

I had a chat with somebody a couple of months ago, a person that I don't know very well and they don't know what I do for a living. They were interviewing for a job and I suggested that perhaps it would be helpful for them to use an LLM of choice to act as a resume coach. And the person I was chatting with was probably, I don't know, 27, 28 years of age. And when I suggested that, their response was, "Oh, so you believe in AI?" Now, they do not work in tech and I'm not expecting anyone outside of tech to really understand how artificial intelligence works. But that conversation made me realize that I live in a bubble where everyone around me knows how machine learning works for example and that when you meet a person outside of that bubble they treat AI as a religion a religion that they can believe in or not.

Now on top of this add another layer which is enterprise SAS versus consumer SAS. If you're in customer success at a consumer SAS company I can honestly see why you can be worried. Although like I said before, what you should be worried about is your role being outsourced, not replaced by I. Even if it happens, the world of enterprise SAS awaits you. Now, we are a long way from support and CX replacement at enterprise SAS companies, especially legacy tech. Why? A couple of reasons.

The first one has to do with the price of enterprise software contracts. You know how often when you go on a website of an app for devs, designers, PMs, anyone really, you often see individual plan free. Let's say there is, you know, a plan for small teams for five people, which is like $50 a month. And then for enterprise package, call sales. Now, as someone who works in product, there are a lot of discussions on the pros and cons of surfacing enterprise pricing and that it's time we show the pricing and stop losing leads. But in practice, the vast majority of enterprise tech still wants to go through sales. Why? Why is an enterprise pricing shown on the websites? Because the costs vary and they vary greatly and you can be looking for the price ranging from $10,000 to hundreds of thousands of dollars.

And it is also not shown because no company is going to put a $20,000 transaction on a credit card, at least for now, which is why sales is looped in for all enterprise contracts. And the future relationship between the company and the customer gets dictated by the price. This is why there is a whole cohort of CX folks whose titles start with enterprise customer success or enterprise support team or enterprise account management. These customers want that service. They want a dedicated CX person even if a chatbot is faster because this is how those contracts have been done for decades. Yes, it is likely that it will get less dependent on human CX staff in the future, but this is going to take time. And I don't have to tell you how deeply legacy culture is rooted in enterprise tech in North America.

Take for example any massive tech enterprise. Levels over levels over levels of sales CX and support. Now, the technology has been good enough for a long time now to be able to get rid of half of those people. Yet, here we are. Do not underestimate the weight of corporate politics.

And now the last layer of this pyramid, company policy. Again, this is especially relevant to enterprise tech. Price negotiation and price formation in tech is so convoluted that if you were to map it out, that map would be the size of Brazil. The price that enterprises pay for the service depend on a million factors. how long the customer has been with them, one-off discounts, promotions, relationship with CX, etc. The closest analogy I can give you in this case is the price of a phone plan that you have if you are with one of the major telecom providers.

I'll use an example from my life. I live in Canada. In Canada, we love monopolies. Telecom is a great example. We have three or four or five big monopolies that control the entire telecom market. One of them is called Rogers. So if you put a hundred people in a row and every single one of them is with Rogers and every single one of them has the exact same plan, the same data, the same number of calls and messages, same everything, I am ready to bet that all of those people would be paying different price for the exact same plan.

Because how they got that plan has a thousand different variations. They could have bought it on Black Friday sale plus they got an employee discount because the employee liked them very much. plus they got a new customer discount and then they somehow paid $30 a month. And then you take customer B who bought the exact same plan but it was on the back to school sale, not Black Friday and the pricing that day was different. And then you take customer C who paid the full price but they called and renegotiate. This is what I mean by convoluted.

You can see how this multiplies. Now with B2C contracts, it's easier and more straightforward because you're dealing with one customer. With B2B contracts, it's a maze. And in my experience, when I saw customer success negotiating price with customers, half the time they play it by the ear. Now, imagine a scenario where a company that handles all of these negotiations with a B2B customer through a chatbot. Can you imagine how they're going to be training that model, how they're going to be fine-tuning, and what kind of work they'd have to do to be able to trust a chatbot to do those negotiations effectively.

The biggest problem here would be the trust, not the ability of AI to do the price negotiation. The company is not going to trust a chatbot to do that kind of work because for large tech companies, the number one priority is revenue. And it's a very delicate balance. But enterprise tech is not at a point where they're going to trust AI to do that kind of negotiation. And there is going to be a massive resistance from within the company to let AI handle pricing. So, no, AI as of now cannot eliminate customer success fully, and we're still pretty far from the moment when and if that happens. Not because AI cannot do it, but because there is a massive human factor at play.

What's actually changing and how to stay afloat? Throughout the entire video, I've been telling you why you shouldn't fall for AI hysteria. Now, does this mean you're safe? No. AI is a risk to your job, not because it can do it instead of you, but because it enables people who couldn't do it before enter your job pool. And offshore teams cost less for businesses because they simply live in places where life is cheaper. And we live in the world of capitalism, ladies and gentlemen. So, the choice is obvious.

Now, what this means is that you have some time, and that time is not indefinite, but there is a little bit of a buffer. Here's what I would do if I were in a customer success or support role right now and did not want to change careers but wanted to stay afloat. Learn Agentic AI and agent configuration. Agentic AI represents the most advanced form of artificial intelligence to date. Unlike traditional chatbots that follow predefined rules, Agentic AI operates autonomously. It comprehends natural language. It organizes workflows and it makes instant decisions grounded in data that it gathers and analyzes.

If we talk about your future, not tomorrow's future, but let's say the next 10 years, yes, a good percentage of your jobs may get replaced by AI agents. So the core change in your everyday job will be the ability to manage agents, not clients. To have the competencies to operate agents, you have to know how they work. You have to know the different types of agent function. Simple reflex agents, model based agents, goal-based agents, utility based agents, learning agents, etc. You need to be able to configure agent parameters and boundaries. You need to be able to handle multi- aent orchestration. And for that, you need the skills on how to design modular agentic workflows, how to manage shared context and memory, and how to operate agentic SAS. You have to master prompt engineering and be very well-versed in techniques.

If you want to stay afloat, I recommend that you research trend around customer service rules turning into agent managers. Again, this is not going to happen tomorrow, but if you want to stay afloat long term within the realm of customer success in tech, this is what I would recommend looking at.

Remember how in the where to run series, I told you that we are not in the AI layoff era yet, and that it won't happen as drastically as the hype makes you believe. Now, I'm building this series to show you why it won't happen the way it seems. But will the jobs change? Yes. Will some of the roles be completely eliminated? Yes. But the elimination does not mean that all of you are going to be axed. Your titles may change. Your responsibilities may change. Your skill set will need to be a lot stronger. And yes, some of you may need to re-qualify entirely.

Will there be more AI layoffs? Yes, and some of them will be hysteria. But it'll be a contained hysteria because introducing AI into everyday workflows and making it secure, ethical, and safe is new to the vast majority of tech businesses. Many will learn from the lessons of CLA, Dolingo, and IBM. Even if you get laid off, the world of enterprise tech is waiting for you. And the world of legacy is definitely waiting for you for the next 10 years. But don't waste your time. Stop the panic. Stop the hysteria. Get control over your anxiety and start upskilling and networking so you can stay afloat and thrive. As always, I hope this helps. I would love to hear your feedback on this first episode in the new series and we'll see you next time. Thank you very much. Bye.