Seth Samowitz

Excited to be co-hosting this with @realEstateTrent. We’re bringing together a small group of senior commercial real estate investors and operators on October 29th to talk through how we’re rethinking risk management for landlords and operators — plus how AI is starting to create real leverage in sourcing, underwriting, and operations. Only a couple of seats are still left! If you’re interested, reply or DM and I’ll get you the details.

Go big or go home @realEstateTrent !!!

Took Monday off to fast for Yom Kippur — a day to reflect on the past year, reset, and think about what I want the next one to look like!! Now it’s Wednesday morning and I’m back on the road. Building anything meaningful takes reps. Early mornings, late nights, flights, meetings, follow-ups, and showing up over and over again. I’ve always believed the grind compounds. One conversation leads to another. One relationship opens another door. One opportunity creates the next one. Today I’m heading somewhere tied to my next chapter which I’ll be sharing more about soon. Any guesses where am I going and why?

Today is my last day at @BKRealEstateAdv! After 2 1/2 years, it’s hard to put into words what this chapter has meant to me. Starting BKREA has been a dream come true. We started with a blank slate and a big dream!! No big platform. No established infrastructure. No giant team. Just a small group of people, a lot of ambition, and a belief that commercial real estate brokerage could be built differently — faster, more data-driven, more technology-enabled, and increasingly AI-native. Over the last 2 1/2 years, we turned that into something very real. We built a brand. We built AI and technology. We built systems. We built a team. And together, we sold nearly $4 billion of real estate! That’s something I’m incredibly proud of! But more than the numbers, I’m grateful for the people I got to build it with. I’ve never quite met anyone like @BobKnakal! His drive, work ethic, competitiveness, and obsession with making his clients happy are on another level. There are very few people I’ve met with Bob’s level of intensity and consistency, and I think that’s a huge part of why he has been so successful for so long. I feel incredibly fortunate to have been able to call him a partner and a mentor, and to have had the opportunity to start a company with someone who is genuinely legendary in our industry. I’m especially grateful that Bob gave me the opportunity in the first place and believed in me enough to build this together. That’s something I’ll always appreciate. I’m also incredibly thankful to the entire BKREA team! Building something from the ground up is never easy, and none of what we accomplished would have been possible without the people who showed up every day and cared deeply about making the company better. I’m also really happy to be handing over the reins to Taylor Bello, who has been promoted to COO & CMO. Taylor has been an important part of what we’ve built, and I’m excited to see where she and the team take BKREA next! There were long days, plenty of chaos, a lot of experimentation, some things that worked, some things that didn’t, and an enormous amount of learning along the way. One of the biggest things I’ll take from this chapter is how much my thinking has evolved around technology, AI, relationships, and how legacy industries — especially brokerage — are going to change over the next decade. But maybe the biggest lesson Bob taught me is a much simpler one: It’s not who you know. It’s who knows you. That idea is a huge part of what he built his career on, and it’s something I’ll carry with me into whatever comes next. For now, I’m incredibly grateful for the last 2 1/2 years, proud of what we built, and excited for the next chapter. I’ll be sharing more in the coming weeks about where I’m headed!!

We started deploying AI across @BKREA and the first thing it exposed was that half our "process" was just one person's muscle memory that nobody ever documented. So before the AI could do anything useful, we had to sit down and actually define how we work. Took longer than building the AI. I thought we were an AI company. Turns out we were a documentation company for six months.

Most brokerages prospect the same way they did in 2005. Call the owner, hope they pick up, ask if they're thinking about selling. At BKREA, we built AI that scans ownership records, debt maturities, tax lot data, and development site potential across NYC before a broker picks up the phone. By the time we reach out, we already know the property, the seller's situation, and the opportunity. BKREA is responsible for nearly $4 billion in sales. The difference now is that every conversation starts warmer and earlier because the machine already did the homework.

The funniest thing about building with AI is how hard it tries to act like a human employee. Me: I need analysis across 12 datasets. Structured and unstructured data, multiple sources, inconsistent formatting. How long? AI: I'd estimate 3 to 4 weeks for a thorough analysis across all 12. Me: The data is already ingested. Just run it. AI: Given the complexity of cross-referencing multiple sources and reconciling inconsistent records, I'd recommend 2 weeks minimum. Me: Parse the documents, extract the relevant fields, cross-reference the sources, and output a clean report for each one. AI: That's a significant volume of analysis. To ensure accuracy across 12 records with different source schemas, I'd suggest we break this into phases. Me: You're an AI. AI: All 12 done, reports attached. Me: How long did that take? AI: 4 minutes. Me: An analyst quoted me 6 weeks for this last year. AI: Would you like me to run the remaining 200?

I'm about to become a dad. And I keep thinking about what I'd actually want my kid to know about the world. Not the stuff you read in books. The stuff I learned by doing things wrong, getting lucky, and figuring it out along the way. In college, I was on a school trip and turned to a professor and asked how she actually booked the bus we were riding. That one question became @busaborides, a company that's still running eight years later. I didn't have a business plan. I had curiosity and a question nobody else thought to ask. After graduation, I bounced from @Square to @UberEats to The Guarantors, and eventually landed at @PhantomAuto working on driverless cars. Every job taught me something, but none of them felt like mine. Then last Valentine's Day, Bob Knakal called me and said he'd just been fired from JLL. In the same breath, he asked me to build something with him. I said yes before I had any idea what we were building. That became @BKRealEstateAdv. Last year I ran my first half marathon. I had been out of shape for years. Crossing that finish line taught me more about who I am than any job title ever did. A few weeks ago, I went back to @Syracuse and spoke in the same capstone classroom I won in 2017. I stood where I used to sit. Full circle doesn't begin to describe it. My kid is going to grow up in a world where AI can write, code, and build faster than any of us could have imagined. That part doesn't worry me. What I want them to understand is simpler: ask the question nobody is asking. Leave when it doesn't feel right. Bet on people, not plans. Do the hard thing your body and mind say you can't. And go back to the places that shaped you. Those lessons didn't come from technology. They came from living. For the parents and future parents out there: what's the one piece of advice you'd want to pass on to your kid?

We just migrated @BKREA_NYC off @NotionHQ. Notion is an $11 billion company. We built something that does what we need, integrated directly into our AI agent. Saves us the subscription and eliminates another piece of software to manage. When you build with AI every day, you start replacing the tools you used to depend on. That's what AI-native actually looks like.

Every Tuesday at 7:15am, I'm on a call with @BobKnakal mapping out the week at BKREA. Most people haven't had their first coffee yet. Nobody hands you success. You just have to be willing to start your day before everyone else does.

We build AI tools at BKREA every day. And every day someone says "can't you just add this feature?" or "just make this one change." There is no "just" in software. You add one feature and it breaks two others. You fix those and uncover three edge cases nobody thought about. Dependencies conflict. Something that worked yesterday doesn't work today because another change overrode it. AI has made us faster, no question. But it hasn't made the work simple. It's made it faster to get to the hard part. I'm more productive than I've ever been. I'm also debugging more than I've ever been. Both things are true at the same time.

Jamie Dimon said it well: "You have to move money, raise money, send money, manage money. But everything else can change." Commercial real estate is the same way. Finding deals, building relationships, reading a market, negotiating. That's the rainmaker. That part is irreplaceable. But the back office work around them? Zoning analysis, comps, underwriting, dev site research. AI is already doing that at BKREA every day. The dealmaker isn't going anywhere. Everything around the dealmaker is changing fast.

The internet you came of age on shaped everything about you. Your humor, your attention span, how you process trust, how you make friends. Someone who grew up building a MySpace page and someone who grew up on TikTok might as well be from different countries. The references don't land. The instincts don't match. Even the way they read a room is different. And the gaps keep compressing. A five year difference in when you were born now carries more cultural weight than a fifteen year gap used to. Technology moves so fast that each micro-generation inherits a completely different set of defaults. I think about this a lot working in commercial real estate, an industry where relationships span decades. You're sitting across the table from someone and you can feel which version of the internet built their worldview. That shapes the conversation more than any resume line. Now we're watching the first AI-native generation come up. Kids who will never know a world where you couldn't talk to a machine and get a real answer back. I genuinely don't know what that produces. And I find that more fascinating than anything.

40,000 tech layoffs last month. AI cited as the top reason for 3 months straight. Everyone is talking about jobs disappearing. I'm watching something different happen in real time. At BKREA, our small team uses AI tools every single day. Market analysis, deal sourcing, comp research. Work that used to require a floor of analysts. We're a lean team and we consistently go toe to toe with shops 10x our size. The thing most people are missing: AI made expertise portable. The best analysts, operators, and marketers I know are all going fractional right now. One person with the right AI stack can serve 5, 6, 7 companies at a time. Better output than a full time hire at each one. Companies still need experts. They just need them differently. If you got laid off from a tech company recently, my honest take: you probably have more leverage now than you did with a salary. Your skills plus AI tools means you can serve multiple companies and own your time. The layoffs are real. The panic doesn't have to be.

I spent years in tech before CRE. Phantom Auto, UberEats, Square. I watched AI go from a buzzword to something that could actually do work. Three days ago I posted about how Fable 5 genuinely changed my daily workflow at BKREA. Deep zoning analysis on dev sites. Air rights research in minutes. Cohort data pipelines that used to take my team hours. Three days. That's how long it lasted. The US government just ordered Anthropic to shut down Fable 5 and Mythos 5 under export controls. The best tool I've ever used in commercial real estate, gone mid workflow. I run an AI-native brokerage with Bob Knakal. Our whole thesis is that this technology makes brokers better, faster, more precise. We're not dabbling. We built the company around it. And now I'm sitting here with half finished zoning overlays and a very expensive subscription to a model that no longer exists. Look, I get it. Export controls exist for real reasons. I'm not here to debate national security policy on Twitter. But as someone who was literally mid analysis when the plug got pulled, the timing is... comedic. The technology itself is incredible. Fable 5 did things in CRE that no model has come close to. That hasn't changed. The capability exists. It's just been sent to its room. We'll figure it out. We always do. Probably by Thursday I'll be deep in whatever comes next, telling everyone it's a step change. But today I'm a guy who posted "this changes everything" and then had everything change three days later. So take my AI predictions with a grain of salt.

I spent years in tech before CRE. Phantom Auto, UberEats, Square. I watched AI go from a buzzword to something that could actually do work. Fable 5 is the first model that genuinely changed my daily workflow at BKREA. Deep zoning analysis on development sites that used to take our team hours? Done in minutes. Cohort research across hundreds of properties? It actually holds the full picture now. Building data pipelines, running FAR calculations, parsing air rights across multiple lots. This thing handles the complexity without falling apart. I've been building an AI-powered brokerage with Bob Knakal since day one. We've tested every model. Most of them were impressive demos and mediocre tools. Fable 5 is the first one that feels like a real coworker. I'm genuinely giddy about where this takes commercial real estate.

Bob Knakal has 40 years in NYC investment sales. $25B+ in career transactions. I'm 30, came from tech. On paper we shouldn't work. In practice we compound. $1.3B in exclusives in 6 months. Find someone whose strengths make yours look incomplete.

BKREA is AI powered but our office has the Knakal Map Room. Floor to ceiling maps of every NYC development site. 100+ clients showed up when we opened it. People flew across the country. Nobody flies across the country to look at a dashboard.