Chainet LogoChainet
Back to Home
Why Not Leaving McKinsey Was Scarier Than Starting an AI Startup

Why Not Leaving McKinsey Was Scarier Than Starting an AI Startup

Yuliya Fomina spent 3+ years at McKinsey across five countries before walking away to build Tailo, an AI agent that converts website traffic into warm leads. She shares why staying comfortable felt riskier than starting over, how B2B sales won't be replaced by AI, and why her two-person team isn't hiring anytime soon.

November 21, 2025
14 min read
By Rachit Magon

Listen on

Most people think the risky move is walking away from a prestigious consulting job. Yuliya Fomina realized the real risk was staying put while years slipped by, advising but never building, strategizing but never shipping.

After spending 3+ years at McKinsey working across Russia, Kazakhstan, Indonesia, Malaysia, and Singapore, plus earlier stints at Google, P&G, and Unilever, she had the dream resume. The kind that makes parents proud and LinkedIn recruiters drool. But when she stopped feeling energy from her work, when her growth stalled despite changing projects and teams, she knew something had to give.

Today, Yuliya runs Tailo with just one other person. No plans to hire. Their AI agent platform turns anonymous website visitors into qualified leads by actually understanding what people want, not just answering generic questions. They're deliberately not scaling their model yet because they're learning what actually works versus what sounds good in a pitch deck.

This is a conversation about making the leap, building in public, and why B2B salespeople might be the last humans standing when the AI dust settles.

Key Takeaways: The Consulting Escape to AI Reality

On Making the Leap:

AI Sales Reality Check:

Building With AI:

Q: You wanted to be a vet as a kid. When did entrepreneurship become the dream?

Yuliya Fomina: Oh no, I thought I'm going to be a vet in my childhood. I was like a very idealistic child with very dreamy thoughts about life and just helping people, you know, rainbows and sunshine. But somewhere closer to my high school years, that's where I really felt that yeah, building something, really being an entrepreneur is probably my passion. But I haven't jumped into that until I had enough confidence in myself. If I could turn back time, I would have started earlier. But at the time I was in an environment which didn't encourage being an entrepreneur that much. I'm coming from a culture where you have a conventional successful path that you need to follow and any digression from that path felt very disruptive.

🔥 ChaiNet's Hot Take: Cultural expectations are invisible cages. Breaking out isn't rebellion, it's survival.

Q: How did your family react when you left McKinsey?

Yuliya Fomina: My mother sat me down to talk. Yeah, for real, that was like that. The initial reaction I would say was not extremely positive and I saw how many doubts there were. Honestly, on the surface for my family I was exceptionally successful, someone who does a great job with their life and then boom, that disappears. It took them a while to accept I would say, a lot of talks on my side also to bring confidence that what I'm doing actually brings me much more happiness and it's working, so why not?

🔥 ChaiNet's Hot Take: Success on paper means nothing when you're dying inside. Your family's comfort with your career isn't worth your misery.

Q: What made you actually pull the trigger and leave?

Yuliya Fomina: I just realized that life is too short to be doing something that I just do not feel energy anymore from. When I originally started at McKinsey I felt like I could give really a lot of my energy into work and when I realized that it's not happening anymore, something looked wrong. If you do not come to work just wanting to do things, that's a problem. Nothing actually helped. I tried many pathways, I changed projects, I changed teams, I changed different angles. It just stopped working for me. I felt like my growth has stalled a little bit. I didn't see myself happy anymore if I continue doing this. I realized I needed to come back to the passion I always had and try my own thing because I felt like I would regret much more not trying than even if it fails, it would be still better than I didn't try.

🔥 ChaiNet's Hot Take: Comfort is a slow death. The pain of staying often exceeds the pain of leaving, we're just bad at math.

Q: Tailo competes with companies like Intercom. How are you different?

Yuliya Fomina: Our primary differentiation is we are specifically built to drive results. If you look at Intercom for example, their core is just to answer questions. That's primarily what they do and they're good at that. We're looking at a different angle. We enter clients with a sales agent that listens to what people actually are here for. It's not a "help me with anything" approach. It's walk them through the product, understand what they're here for. This is invaluable information that sales teams are getting. Even if this lead is not converting, they have a huge dataset now that says okay, what are people coming to our website for? Are they even quality people? Are we attracting the right audience? If they're not converting, what's going on? Which video snippets that Tailo shows are converting? It's a big analytics platform built specifically to drive conversion rates, not just answer questions.

🔥 ChaiNet's Hot Take: Most chatbots are glorified FAQ pages. Tailo's betting that understanding intent beats answering questions.

Q: Walk me through a real example. How does Tailo work differently?

Yuliya Fomina: When we start building an agent we always talk to teams a lot to understand what they're looking for. A company could have a problem with conversion rates when they have visits but nobody books a demo, they just leave. That's one problem. Another problem could be we have so many leads we cannot understand which one is good and which one is bad, how do we prioritize? Completely different problems. For conversion rates, we build a fancy agent that looks good, not your regular basic bot that just irritates you, but something that indicates okay, that's useful, that would help me understand the product. Our agent does discovery, shows exactly the right parts of the product. What quite often happens is there's a big 30-minute video about the product and I don't need this 30-minute video. I just need to understand whether you have a feature I care about. Help me realize this product is worth my time. That ends with a CTA with opportunity to book a call with the team.

🔥 ChaiNet's Hot Take: Nobody wants your 30-minute demo. They want to know if you solve their problem in 30 seconds.

Q: Is Tailo a self-service platform or do you build custom solutions?

Yuliya Fomina: We have a lot of predetermined workflows. We do not yet have self-serve and we're deliberately avoiding this right now because we need to understand more of what is the full portfolio of cases. For our clients right now it will not be easy for them to build themselves. What we do is we can spin off an agent for them relatively quick. They cannot yet but we're getting there. Frankly our model is not yet scalable but it's also deliberate because by doing this manually first we realize we get much more information on how it should look like. It's all done deliberately for now.

🔉 ChaiNet's Hot Take: Doing things that don't scale isn't a bug, it's the feature. You can't automate what you don't understand.

Q: How does building AI at a startup differ from advising on AI at McKinsey?

Yuliya Fomina: The biggest difference is you own much more when you are building your own AI. When it comes to working at McKinsey and advising clients on how to implement AI, you are not walking in their shoes. It's like oh, it's nice to have a workflow agent like this, I think I see a gap, you could improve it if you have this agent, here is our recommendation, good luck building it, and yeah we can build it for you if you pay us a little bit more. It's a little bit outside-in perspective. You do not live in this, you do not own it. When you're building your own AI, I'm literally building Tailo's AI for our clients together with them, it's a completely different level of ownership and stakes. I'm literally reading through all the conversations that our agent has with our clients manually. Something I would never do at McKinsey. If my agent doesn't work with some clients I know it will not work with other clients. I need to make sure it all clicks.

🔥 ChaiNet's Hot Take: Consultants recommend. Founders bleed. One gets paid regardless, the other only eats if it works.

Q: Will AI replace human sales agents completely?

Yuliya Fomina: I don't think so to be honest. There will be companies who would make a conscious decision eventually to just replace all the sales agents with AI. It would be in theory not a good decision I think but that's something some companies would do just from a cost-cutting perspective. But in B2B especially you will continue buying from people. AI would help you maybe in our case just trigger interest, make sure that there are less no-shows than there are now. But completely replacing human touch and trust, I don't believe in that. There are B2C industries out there which currently have too many clients to manage and it is not possible to connect to everyone. What AI would enable is potentially possibility to have one-on-one conversations with each B2C client because of the scale. But in B2B, no.

🔥 ChaiNet's Hot Take: B2C sales will democratize through AI. B2B sales will stay human because trust doesn't scale through algorithms.

Q: Where do you use AI in your daily life?

Yuliya Fomina: Well, honestly, my favorite use case is brainstorming on birthday presents. It always has been my big problem and I almost built a startup that will help people find birthday presents. I think AI can do a really good job at brainstorming ideas for that. But frankly speaking, a lot of cases, starting from I'm in a foreign country and I don't understand how to turn on this washing machine, I just ChatGPT video call and it's like magic. I'm an active user of AI in a lot of cases.

🔥 ChaiNet's Hot Take: The best AI use cases aren't revolutionary, they're annoyingly practical. Birthday gifts and washing machines beat most enterprise solutions.

Q: How is AI optimizing your actual work at Tailo?

Yuliya Fomina: We currently are just a two-people team, two of us, and we are not thinking about hiring yet in the recent future because how much you can do with AI in a day is just incredible. Every month it's even more because we also learn how to use it better. We can build a feature in a day with AI because it just really helps automate a lot of work that used to take much more time. We use it heavily and we believe ChatGPT Pro subscription is like the best investment we ever had. It's a bit scary at times for junior workforce because AI is almost like my assistant these days. I don't need a person, an intern who would analyze the spreadsheet because it can be done by ChatGPT. We use it all the time.

🔥 ChaiNet's Hot Take: Two people doing what used to take ten isn't productivity, it's the extinction event for entry-level work.

Q: What's been your biggest failure and what did you learn?

Yuliya Fomina: On the first ever job, like a real job I ever had, I got fired from it. It's not many people know about it but yes, I'm just openly saying that. It happened very early in my career. I was like a teenager I would say. But it shaped a lot of my adult life because when you get that shock content early on that really helps. What I learned immediately is find roles where you believe is something where you can succeed. It's very important to learn what are your weaknesses, what are your strengths. If you realize what is that where you can be your best ever employee ever, you need to find that and succeed there because you get both success and satisfaction out of it. If you cannot be someone who can run, it doesn't make sense for you to get into that space because you would be a failure there but you can be a huge success somewhere else, you just need to find it. At times what happens in societies, especially very traditional ones, you have a traditional job you need to take, you need to be an accountant because your father said so but you are not built to be an accountant at all. A lot of people I see are pushing themselves into spaces that are not built at all for them and that brings them failure at the end of the day and sadness about their life because work is a big part of their life.

🔥 ChaiNet's Hot Take: Getting fired young is a gift. It forces you to find your fit before wasting decades in the wrong game.

Q: What's been the hardest part of building Tailo as an entrepreneur?

Yuliya Fomina: Every day you need to choose one path out of 100 that you in theory can take. You need to pick one because you have only one day. Having conviction that this is the right path, being mentally ready that it should be the right path but you understand it could be eventually not the right path, it's hard. There is no one there that would tell you you need to do this. There is even no one you can rely on someone's judgment to make decision for you. At McKinsey it was very easy. You can come to your partner, say okay I think I need to do X, what do you think? They say okay yes you should do X and you kind of have this internal conviction that this is the right thing that you need to do. But in entrepreneurship it's really like you just believe in it and go through with it. That's the only way.

🔥 ChaiNet's Hot Take: Corporate life is paint-by-numbers. Entrepreneurship is a blank canvas where every brushstroke could ruin the painting.

Q: How did your first version of Tailo look and what did you learn?

Yuliya Fomina: We originally started with the hypothesis of fully replacing product demos. We felt that AI could do it and I think it still can actually, but what we realized is people are not ready to talk to AI for more than 5 minutes. It's just, there is a certain level of, it's hard to convey but data just showed us that full demo of a product, no person is ready for that to go through with AI. For many it probably felt a little bit wrong, a little bit strange. We realized that right now what really worked is what AI could do really well is show not a full demo but a short part of the interface that person is interested in. Our conversations became shorter but then they started to bring much more results into actual trust into the product, into talking to eventually the team. Short doses of AI works much better than long 20-minute discussions that no one is ready for right now.

🔥 ChaiNet's Hot Take: Users will tolerate AI in sips, not gulps. Five minutes is the attention span ceiling, not the floor.

Q: What's been the most surprising thing about having customers use your product?

Yuliya Fomina: What was surprising is that once a client outreached to us and told us okay look guys, you do really cool things, can you give me a space where I can fully edit how exactly it talks, how exactly it responds? This was one of the wakeup calls for us that people want very custom flows in the agent and the editing of it should be really simple in plain English. Because originally we were like okay look, we have our own script prompt which we protect, it's our proprietary knowledge of how it should really walk a person through the demo and we were super protective originally of it. We thought oh my god if you just give it away, our product, what exactly are we here for? And then we realized it's not the point. Actually the point is giving space, giving flexibility, opportunity to influence this agent and make sure it behaves the way you want it to behave and given opportunity to do that easy. This client wanted it completely different and the next client comes, they wanted completely different from the client before. We were like okay, this doesn't work like that, you cannot just come with your template and use it for every company.

🔥 ChaiNet's Hot Take: Proprietary prompts are fool's gold. Flexibility that's easy to use is the real moat.

Q: What's your honest advice to people stuck in comfortable corporate jobs thinking about starting something?

Yuliya Fomina: First, it's important to make this jump only when you know what it means. Only know the scenario in your head when this will go wrong. It's very important to not romanticize this jump because some people I know do and it's a problem. Things will go wrong for sure at some point and if you do it you just need to picture worst-case scenario, what's the worst case that would happen if I made this jump. If you are comfortable with this scenario, if you are, then yeah, then maybe that's the right answer, to make the jump. I just see many people kind of having a lot of talk about oh I will do it, I will do it, but do not do it eventually and that could be a right call for them because very few people are comfortable with being a founder, very few people are comfortable to leave the safety net and that's actually a good thing. Stability and safety net, it's not the time for shaming it, it's very important for life. But if you're sure, if you're okay even with the worst scenario, then yeah, then go for it and consume it fully, learn from this experience which is very interesting.

🔥 ChaiNet's Hot Take: Picture your worst-case scenario. If you can live with it, jump. If you can't, staying put isn't failure, it's self-awareness.

Final Thoughts: The Risk of Playing It Safe

Yuliya's closing wisdom: "Don't do it for money. I have friends who've been in corporate positions for the longest and they're at least three to four times more, they have bigger networks than me. That's a very interesting point which is often misunderstood by a lot of young entrepreneurs."

The bottom line: Yuliya's journey from McKinsey to a two-person startup isn't a story about trading stability for chaos. It's about recognizing that the biggest risk isn't failure, it's spending years in a role that stops challenging you while telling yourself you're being responsible.

Her insights on AI sales are refreshingly honest. B2B sales isn't going anywhere because trust still matters. But B2C? That's getting democratized hard. The future isn't AI replacing humans everywhere, it's AI enabling humans to do what only humans could do at impossible scale.

For anyone sitting in a comfortable corporate job wondering if they should make the leap, Yuliya's advice is brutal and necessary: picture your worst-case scenario. Really picture it. Not the romanticized "I'll sleep on a friend's couch for six months and then hit unicorn status" fantasy. The actual worst case. Can you live with it? If yes, jump. If no, staying isn't weakness, it's wisdom.

Q: How can people connect with you and learn more about Tailo?

Yuliya Fomina: You can find me on LinkedIn, that's where I'm most active. For Tailo, check out our website where you can see how our AI agents work in action. We're always happy to talk to teams who are struggling with converting website traffic into actual conversations with potential customers.

Final words: The corporate ladder promises safety but delivers stagnation. Startups promise chaos but deliver growth. Neither path is inherently better, but only one forces you to confront who you actually are versus who you thought you'd become. Yuliya chose discomfort over complacency, and whether Tailo becomes a unicorn or a learning experience, she's already won by refusing to let comfort decide her future. The machines are coming, but they can't replace the human need to build something that matters. Choose your hard: the hard of staying safe or the hard of betting on yourself. Both hurt. Only one helps you grow.

Related Shorts

Explore short-form videos that expand on the ideas covered in this blog.