Hello reader, welcome to a special Saturday edition of HX Weekly!

This week we are sharing a story from one of my great mentors and close colleagues, Jacob Doft.

Jacob and I attended Wharton together at the same time and was one of the first people to teach me how to pick stocks. Over the next few decades, we collaborated many times and almost worked together a couple as well.

He is super smart, super knowledgeable and also one of the best human beings I have ever met. He recently published this story on Tablet and we thought it was worth sharing.

It is a tale of modern investing, the wild world of AI and - ultimately - human nature. We hope you enjoy the story as much as we did!

Jacob is just getting started with his writing career and you can follow him at @jwdoft on X/Twitter and @jakedoft on Substack.

A Tie in Tel Aviv

How a fake fighter pilot, a real AI professor, and a chatbot nearly pulled off the perfect Israeli startup con, and what it cost the man who believed them

The first thing I noticed about Erez Ben-Ishay was the tie. It was silk, dark blue, in a perfect Windsor knot, and he was wearing it in Herzliya, Israel, in April, at a hotel where every other man was in sandals.

I have spent a good deal of my adult life in Israel. I have met generals, tech founders, billionaires, and at least one person who I’m fairly certain had a live feed of the ayatollah on his iPhone. None of them wore a tie. The country doesn’t really do ties. A tie in Israel is like an umbrella in the desert. It’s not that it’s wrong, exactly, it’s that it raises a question about what the person thinks is going to happen.

I didn’t say any of this. I said, “Nice tie.”

Erez said, “This is how I do business.”

I said, “Uh-huh.”

And that, more or less, is how I entered a two-year relationship with a fraudulent AI company, a fake fighter pilot, a brilliant professor who was being conned by her own CEO, and a chatbot that was doing most of the work. It is, on balance, the strangest thing that has happened to me. And I spent 30 years on Wall Street, where strange things happen before lunch.

I want to explain why I’m telling you this story, because the reason matters. I am neither a journalist nor an investigator. I am a retired hedge fund manager who got pulled into something, and who, for reasons I’m still examining, could not pull himself out. Some of those reasons are noble. I wanted to protect people I loved. Some of them are less noble. I was retired, and I had become useless, and this was the most alive I had felt in years.

My favorite television series is The X Files, the show whose tagline is “The truth is out there.” What I always loved about the show is that sometimes the truth turned out to be outrageous, a genuine conspiracy, a real monster, something that justified every ounce of paranoia. And sometimes the truth was mundane. A hoax. A misunderstanding. A man in a mask. Either way, the truth was usually irrelevant, because the search itself had become the point, and no answer could ever be as compelling as the question.

I think about that now, looking back at PromAI. For two years I searched for the truth about this company, and when I found it, it was simultaneously outrageous and mundane. Outrageous because the entire corporate reality had been generated by a chatbot. Mundane because at the center of it all was just a man in a suit who lied, for reasons that even he may not fully understand. The truth was out there. I found it. And it was, as my son Joey would later articulate better than I could, smaller than I expected.

I should tell you a little about myself. A Google search will inform you that I am a retired hedge fund manager, a venture capitalist, and a pro-Israel philanthropist. What it won’t tell you is that I am fundamentally a listener. I don’t talk much in meetings. I’m told my face doesn’t move much. This isn’t intentional. It’s just the face I was given, and it happens to be useful in a profession where everyone else is performing confidence at full volume. Underneath that face, I want you to know, it is loud. I am questioning everything all the time. Every instinct I have is followed immediately by a second instinct that says: But what if you’re wrong? 

One more thing, for context: In my world, a few million dollars is the threshold where an investment becomes real. Anything below that is a token, a gesture, a way of keeping your seat at the
table without truly committing. Anything
above it means you’ve done your work and you believe. This is a financial and emotional red line.

The story begins at Passover in 2023, at a hotel in Herzliya, on a morning so beautiful it felt like Israel was showing off. My father-in-law, Bill Weiss, told me over breakfast that his best friend Howard Miller was bringing over a company for him to see. An Israeli AI startup. Howard had invested significantly. He was beyond excited.

Bill is a piece of work, and I say that with complete love. He’s a Brooklyn-born software entrepreneur who sounds exactly like Bernie Sanders arguing about a parking ticket but whose politics land somewhere to the right of Ted Cruz. If you closed your eyes and just listened to him talk, you’d assume he was organizing a union. Or a stickball game in Flatbush. Then he’d tell you his thoughts on immigration and you’d need to sit down. He eats Israeli breakfast the way it’s meant to be eaten: 17 small plates, everything simultaneously, talking with his mouth full about three different subjects. Fork and knife? Out of the question. We have a close relationship. He trusts my judgment. I trust his heart.

Howard, I should say, had told Bill something that certainly got our attention: The founder, a woman named Dina Goren-Bar, had been Howard’s girlfriend of 10 years. They’d broken up, but he was still a major investor. Despite the personal situation, Howard said the technology had commercial potential worth billions. And it involved artificial intelligence.

I need to tell you about Howard, because he is the most important person in this story and the one who deserved the least of what happened to him. Howard Miller was not a reckless man. He was not a fool. He was a successful businessman, sharp enough to have made real money over a career, generous enough to share it freely, and trusting enough that his generosity sometimes got ahead of his diligence. He had invested in PromAI originally out of loyalty to Dina, the woman he’d loved, as well as deep reverence for her intellect. When you love and respect someone for a decade and they build something they believe in, you write the check. That’s not stupidity. That’s devotion. The tragedy is that devotion and due diligence don’t always live in the same room, and Howard’s heart had more pumping capacity than his skepticism.

He was also one of the most likable human beings I’ve ever met. Everyone who knew Howard has a story about something kind he did that he never mentioned to anyone else. He was the guy who showed up at the hospital when you didn’t ask him to. The guy who remembered your kid’s name. The guy who, when you were having a bad day, would call you for no reason and talk about nothing until you felt better. And he was my father-in-law’s closest friend, which meant he was family.

I brought my son Joey to the meeting. Joey was 18, freshly admitted to Wharton, about to spend his gap year in Tel Aviv, and looking for something to do with his free time. I thought he should start sitting in on meetings like this. Not because I expected him to have opinions, though he always does, but because learning to read a room is a skill that takes decades, and the clock doesn’t start until you put the kid in the chair. Observing is everything.

Joey is also a kinetic thinker. He can’t sit still, reads while pacing, and processes information best when some part of his body is in motion. Even in that first meeting, he was tapping his pen, shifting in his seat, bobbing slightly to some internal rhythm. He looked like a young man who’d accidentally wandered into a conference room on his way to a rave. People found this charming.

Erez entered the room with the tie and the suit and a handshake that was about 15% too firm. Behind him was Dina Goren-Bar, who was elegant and serious and clearly the intellectual engine of whatever this company was.

They explained the product. PromAI had developed an AI-powered software platform that optimized manufacturing lines, identifying errors, reducing waste, and improving quality. The business model was that PromAI would capture half of the cost savings it generated for its clients. If they saved a company $10 million a year, PromAI kept $5 million.

This got my attention for the wrong reason.

“That’s not how enterprise software works,” I said. “Success-based pricing is almost impossible to quantify. How do you attribute the savings to your software versus a hundred other variables?”

Erez didn’t flinch. He said they already had agreements with Medtronic and Elbit, two significant companies that had embraced the model.

Now, maybe they had. I wasn’t there. I hadn’t seen the contracts. It’s entirely possible that Medtronic’s procurement department had lost its mind and agreed to give an Israeli startup half its cost savings in perpetuity. Stranger things have happened in enterprise software. Not many stranger things, but some. I filed it away and moved on. There’s no point in arguing with a pitch.

Then Erez began talking about himself.

He had been an engineer his whole career, he said, while concurrently serving as lead test pilot for the Israeli Air Force. He had personally tested every aircraft in Israel’s fleet. He had invented the thumb drive. When I asked what kind of planes he’d flown, he said, “All of them.”

I looked at him. A fighter pilot. A test pilot. An inventor. A CEO. A man who had done everything, flown everything, built everything. This was either the most extraordinary person I’d ever met, or it was a performance. I couldn’t decide.

I don’t always decide things in meetings, but when I do, I usually screw up. I decide things later, alone, when nobody is watching. This is, I think, the single most important habit I developed in 30 years of evaluating companies: Never react in the room. React in the car. React in the shower. React at 2 in the morning when you wake up and the thing that didn’t sit right suddenly makes sense. But never in the room. The room is for listening.

Then Erez noticed my son.

It began as normal interest. Questions about Joey’s background, his gap year, Wharton. But it escalated with a velocity that activated something in me. Erez leaned forward. His focus narrowed. And then, without any discussion of qualifications or roles, he offered Joey a job on the spot. A paying job.

After the meeting, I pulled Bill aside and listed my concerns. The suit. The revenue model. The rush to hire Joey.

Bill looked at me. I drank my coffee.

“So we’re not investing,” he said.

“We’re not investing yet. Let me dig. If it’s real, it’s real. I could be wrong about all of this. I’ve been wrong before.”

I genuinely wanted it to be real. Not for the money, although I love making money. But because Howard was deep in this, and Bill loved Howard, and the best possible outcome is always the one where nobody gets hurt. Every instinct I had said something was off. But instincts are not evidence, and I have learned, sometimes painfully, that the distance between a hunch and a fact is the distance you have to leap before you’re allowed to act on it.

And I will tell you something else, something I didn’t fully recognize at the time but that I’ve come to understand in the months since: I didn’t want to walk away because I didn’t want to go back to being retired.

I had been off Wall Street for two years. I had money. I had time. I had freedom. And I had, if I’m honest, a growing sense that I was becoming an observer of my own life. I read the papers. I took meetings. I gave advice. But I was no longer a participant in anything. I wasn’t making decisions that had consequences. I wasn’t in the arena. I was watching the arena from a very comfortable seat, and the comfort was slowly killing something in me that I didn’t have a name for.

Then Erez walked in with his tie, and a company that might be real or might be a fraud, and a son of mine sitting in the room, and a father-in-law’s best friend with millions at stake, and suddenly I was back. I was reading a room. I was evaluating a claim. I was protecting people. I was doing the thing I had spent 30 years training to do, and it felt like a double espresso after a long nap.

Erez followed up with a formal email asking me to invest $3 million at a $15 million valuation. The money, he explained, would be used to build out the team, launch U.S. sales, and finish the product. He cited the Elbit and Medtronic deals as evidence of traction.

The email was polished, professional, and, as I would learn much later, written entirely by ChatGPT.

I asked my analyst, Francis Lee, to start pulling the company apart. Francis is the kind of person who finds a discrepancy in a document and experiences something close to physical pleasure. He is extremely good at his job. He began running backgrounds, pulling competitors’ corporate filings, and mapping the claims against publicly available evidence.

Around this time, Erez and Dina came to my office in New York, accompanied by Howard. Despite it being summer, Erez arrived wearing a bomber jacket. In July. In Manhattan. I wanted to ask if he was cold or just committed to the bit, but I didn’t. Listening mode does not include commentary on outerwear.

I sat the three of them on the sofa in my office. Francis and I took the chairs across from them. I asked about the air force, because I was genuinely curious.

Erez said he still flew missions. He still tested planes. He was still actively involved in the air force. The bomber jacket, he explained, was a gift, because he had “won Top Gun” in 1996.

I asked: the Israeli Top Gun?

No, he said. The American one. The real one. He was the only Israeli pilot to ever win Top Gun.

He later sent us a photograph of a crystal trophy.

We requested two things for our diligence: a product demonstration and conversations with their paying customers. They agreed. Then we heard nothing for months.

In the silence, I made a decision about Joey that I want to explain carefully, because it’s the decision in this story I’ve thought about the most, and the one I’m least certain I got right.

PromAI had offered him the paid internship, and Joey wanted to take it. He was 18, curious, and bored in the way that smart 18-year-olds get bored when they don’t have enough to do. Under normal circumstances, I might have told him to find something else. These were not normal circumstances. I had a company I couldn’t evaluate from the outside, a CEO I didn’t trust, and an investor I loved who was in too deep.

Besides, this was the spring of 2023, and artificial intelligence was beginning its boom. The technology was suddenly everywhere, in the news, in the markets, in every conversation I had with anyone under 40. Whatever PromAI’s problems were, the company was building AI software, and the chance for Joey to spend a year inside an AI company, learning the tech’s potential, appreciating how it worked, getting fluent in the language and the culture of it, was genuinely attractive. The world was moving in that direction. Having my son ride with it, even inside a company I had doubts about, seemed like an education worth the risk.

I also recognized that having Joey inside PromAI gave me something I couldn’t get any other way: proximity to the truth. Joey wasn’t a spy. I would never have framed it that way to him. But he was observant, he was smart, and he was there, every day, in the same room as the people I was trying to understand.

I told Joey to take the job.

I told him to pay attention and tell me what he saw. He said, “So you want me to spy.” I said, “I want you to observe.” He said, “Pops, that’s the same thing.” He was probably right.

Joey talks about it now with a kind of incredulous laughter. “I still can’t believe I was part of that,” he says, and shakes his head, the way you shake your head at something that’s too absurd to be upsetting and too real to be funny. It’s become one of those stories he tells at school, and every time he tells it, some part of me flinches, because every detail he recounts is a detail I let happen.

His gap year program leaders, Yehuda and Rav Shlomo, two serious men with extensive networks in the Israeli military, visited him at the office and met Erez. They called me afterward with grave concern. Not a single person in their contacts had heard of Erez Ben-Ishay. This was notable. Israel’s military community is not large. A lead test pilot for the air force would be known.

I considered the possibility that Erez was simply classified, one of those top-secret types that Israel produces in unusual numbers. It was possible. The thing about intelligence work is that it provides perfect cover in both directions: The guy who is actually secret can’t tell you he’s secret, and the guy who is lying about being secret can tell you he can’t tell you. I filed this alongside everything else and kept my mind open, even as the file was getting heavy.

In the meantime, we checked the Top Gun story. We contacted the U.S. Navy Office of Information at the Pentagon and sent them the images Erez had provided.

The trophy was fake. The patch was from the movie. The Navy had never heard of him. Even better, Top Gun is not a competition. There is no winner.

It’s an odd thing, confirming that someone has been lying to your face. You’d think it would feel satisfying. The instinct was right, the data confirmed it. But mostly I felt dread. Because the lie wasn’t directed at me, not really. I hadn’t invested anything. The lie was directed at Howard, who had invested in Dina’s company because he had faith in her, and who had conceded the Top Gun legend because it made the story even better. Howard wasn’t a dupe. He was a man whose love for Dina had extended, naturally and understandably, to the company she’d built and the people she’d chosen to run it.

“The bomber jacket, he explained, was a gift, because he had ‘won Top Gun’ in 1996. He later sent us a photograph of a crystal trophy.”

I decided not to tell Howard about the Top Gun fraud. I told him only that we couldn’t get confidence in the commercial opportunities and that we were going to pass. It was the polite exit. The kind thing. And I suppose it was also, in retrospect, a mistake, though I’m not sure what the alternative would have produced. Howard believed with the force of a man whose investment was an extension of a relationship, not a spreadsheet.

Nor does a lying CEO necessarily mean a bad company. Some of the most valuable technology firms in history were led by people with a complicated relationship with the truth. I couldn’t shake the possibility that underneath Erez’s theater, Dina had developed something important. I kept challenging my own conclusions. Maybe I was being too cynical. Maybe my instinct about the tie was just a bias against formality. Maybe the Top Gun thing was an exaggeration, not a fabrication. I didn’t believe these maybes, exactly, but I made myself consider them, because the day you stop considering them is the day you start being wrong in a different direction.

And underneath all of this: I didn’t want to let go. Not of the investment opportunity. I hadn’t invested, and at this point I probably wasn’t going to. What I didn’t want to let go of was the investigation itself. For decades, I had woken up every morning with a problem to solve. Then I retired, and the problems stopped, and I discovered that, for me, the absence of problems is its own kind of problem. PromAI gave me something to do. Something with stakes. Every phone call, every document, every late-night text from Joey, every new lie from Erez, was another piece of a puzzle, and I was the one assembling it, and it made me feel necessary in a way that retirement had not.

In early 2024, Erez sent me another formal email. Invest now, at a $20 million valuation. I politely declined. Still no customer conversations. This was going to be my permanent excuse: Show me the customers, or leave me alone. And if, by some miracle, they actually had customers, I would happily change my mind.

They promised to connect me. They never did.

Later, Erez and Dina came to my office yet again, Howard between them on the sofa. Francis and I were direct: no investment without customer verification. Out of respect for Howard, we were blunt.

I left the room to get a coffee. When I returned, the energy had shifted. I looked at Francis. “What happened?”

Erez had announced that he and Dina were now “together.”

I looked at Howard, sitting between his ex-girlfriend and the man she was now dating. “You OK with this?” I asked.

Howard said absolutely. And I believed him. He wanted her to be happy. He wanted the company to succeed. If Dina being with Erez was part of the package, Howard would accept it, because Howard accepted things. That was his grace and his vulnerability.

They did eventually arrange a Zoom for me and Francis to speak with a customer, a man named Nir Livshin, who was supposedly the general manager of a military drone program at Elbit. The call began. I started gently: How long have you worked at Elbit? Where’s your office? Did your team lose anyone on Oct. 7?

I was about three minutes in when my phone started lighting up like a slot machine. Francis was texting me. Jake. Jake. Jake. Jake.

I hadn’t received texts this frantic since the time Joey ran out of toilet paper.

I looked at my phone under the desk. Francis had pulled up Nir Livshin on LinkedIn. He was head of IT at PromAI. He didn’t work for Elbit. He was sitting in the PromAI office, pretending to be a customer, on a Zoom call with me.

I texted Joey.

Do you know someone named Nir Livshin?

Yeah Pops. He sits next to me.

Can you look over and see if he’s on a Zoom, with me?

Yup Pops.

I want to pause on this moment, because it is the most absurd thing that has ever happened to me as a professional, and I include in that category the time I watched a hedge fund manager weep during a presentation about soybeans. My 18-year-old son was sitting 10 feet from a man who was, at that very moment, on a video call with me, pretending to be someone he was not, at a company where my son was an intern. If Joey had rolled his chair 3 feet to the left, he could have waved at me through Nir’s camera.

I was not angry. I was offended. I have spent my entire career evaluating people who stretch the truth, shade the numbers, or present aspirations as accomplishments. That’s the startup game. But this was not stretching. This was a man in an office pretending to be someone at a different company, on camera, in real time, while the investor’s son sat close enough to hear what he was saying. It was lazy. It was sloppy. And it was, in a strange way, disrespectful. Not only to me, but to the craft of deception itself. I’ve been lied to by professionals. This was amateur hour with a ring light.

I called Howard. This was the call I’d been avoiding for months, and it was one of the hardest things I’ve ever done.

I told him everything. The Top Gun trophy was fake. Nir Livshin wasn’t an Elbit executive; he was the IT guy at PromAI. Every customer document they’d sent us was PDFed on a Mac, Erez’s Mac, and the spelling mistakes were the kind an Israeli would make, not an American. The contracts were doctored. The logos were fabricated. None of it was real.

Howard was devastated.

He asked me what he should do. I told him to get out. To separate himself from the situation. To never write another check.

But even as I said it, I still held a sliver of doubt in reserve. Not about Erez. Erez was clearly a liar. But about the technology. Dina was a legit AI professor with scientific publications and a real product. Forty engineers were building something. Maybe they were building something good. Maybe the tragedy of PromAI was that a brilliant product was being led by a delusional man, and if you could remove the man, the product might be worth something. Sometimes you’re so focused on the lie that you miss the truth sitting behind it. I’ve seen worse people build better companies.

The months that followed were an exercise in watching people I cared about drift further into something I couldn’t stop.

Howard and Bill called regularly. PromAI was getting interest from Boeing, Tesla, Airbus. The FAA wanted PromAI to help fix Boeing’s manufacturing problems. They sent me the FAA meeting minutes. Boeing had supposedly offered to invest $30 million at a $1 billion valuation. Howard started talking about how he was going to be a billionaire. I wanted to tell him that billionaires don’t usually become billionaires by investing in a man in a costume. But I didn’t say that, because Howard’s enthusiasm wasn’t naive. It was hopeful. And hope, in a man as kind as Howard, is hard to interrupt.

“Some of the greatest technology companies in history were run by lunatics. Some of the most successful founders were liars who happened to be building something real at the same time.”

Erez, for his part, seemed unbothered by my distrust. He remained obsessed with convincing me to invest. On one of my trips to Israel, he staged an elaborate meeting, waiting for me in a conference room with his outside counsel, who formally handed me the proposed Boeing contracts as if the lawyer’s physical presence constituted verification. I suppose the theory was that documents become more real when a person in a blazer hands them to you.

I read the contracts and asked one thing: to speak to the person at Boeing who’d negotiated them. In fact, I said, I would fly to Seattle or Chicago or wherever and meet the executive in person. I wanted to sit across from a Boeing employee, in a Boeing building, and hear them say this was real. Part of me genuinely hoped they would. Most of me knew they wouldn’t.

Erez agreed. Then, once again, he disappeared.

When he resurfaced, weeks later, he explained that he’d been managing the operation to bomb Iran. This was a common theme. Erez’s disappearances always had geopolitical explanations. He was flying combat missions over Gaza at night. He was being sent to the White House to help prepare for one of Netanyahu’s visits. He was an emissary of the IDF. And when he traveled between the U.S. and Israel, it was because “a special operation is brewing.”

I doubted he could fit in a cockpit. But I kept this observation to myself.

He also claimed that as a VIP, he was invited to sleep in the Pentagon’s special guest wing when he was in Washington. This was almost certainly another layer of insulation: If anyone ever questioned why Erez had gone silent for days while supposedly in D.C., he had a ready-made explanation that was both unverifiable and too absurd to challenge.

Around this time, Howard was sending me videos of Joey at work. Howard had taken to secretly filming my son at his desk. Joey with noise-canceling Beats, dancing in his seat, bouncing to music, doing some kind of full-body rhythm while simultaneously getting his work done. The videos had the wobbly quality of a 70-year-old man trying to hold his phone steady while laughing. Howard was charmed by Joey in the way that genuinely kind people are charmed by youth and energy. He saw in my son something uncomplicated and alive, and it made him happy.

I watched those videos alone in my office, usually more than once. My son, dancing in his chair, inside a company that I was increasingly certain was a fraud, 10,000 miles from me, oblivious and joyful. I felt something large and complicated. Love and fear and pride and the particular anguish of a parent who can see a danger that his child cannot, and who put the child there deliberately, and who would probably do it again, and who wonders what kind of father that makes him.

Erez sent another formal email. Invest at $23 million. Then $28 million to $30 million. I ignored each one. But Bill, my father-in-law, told me that missing out on the Boeing deal, which they claimed would yield $249 million per year in revenue, would be too much for him to handle. He put in a nominal amount, the kind of check I wouldn’t consider a real investment. Not because he believed, exactly, but because he wanted to support Howard, and because on the off chance that the Boeing contract was real, he didn’t want to be the one who sat it out. It was not a conviction bet. It was an insurance policy against regret. Schmuck insurance, it is specifically called. Yet in his heart he saw what I did.

He just couldn’t say it, because saying it meant telling his best friend that the thing Howard had built his hopes around was a sham.

Yet I continued to wonder if I was wrong. If I was the dupe. This is the part of the story that’s hardest to explain to people who haven’t worked in venture capital: the persistent, nagging fear that your skepticism is costing you the opportunity of a lifetime. Some of the greatest, most valuable technology companies in history were run by lunatics. Some of the most successful founders were liars who happened to be building something real at the same time. The line between “visionary who exaggerates” and “fraud” is, in many cases, visible only in hindsight. Maybe I was too suspicious. Maybe I was the problem. I didn’t think so. But I’ve met a lot of people who didn’t think they were the problem, and some of them indeed were.

Eventually, Bill called me and said that while he understood all my concerns about Erez, he had accepted an invitation to go to London with Erez for an aerospace conference. Maybe he needed to see for himself. Sometimes the only way to know is to go.

Then came the Boeing contract.

Howard emailed it to me, triumphant. The finished agreement. Signed, sealed, and delivered. I opened it the way I open everything: slowly and carefully, looking for the details that don’t want to be found.

The name of Boeing’s COO, Stephanie Pope, was misspelled. There was a corporate stamp on the signature page, which is something that essentially nobody uses in modern business. The last time I saw a corporate stamp on a legitimate document was in a museum, and it was behind glass. The PDF metadata showed the document had been created on a Mac. Boeing, I am certain, does not issue contracts from MacBooks. And the Boeing entity that signed the contract was different from the Boeing entity listed in the body of the agreement.

I mentioned these observations to Bill and Howard. I begged them, again, to get more evidence. To call the individuals named in the documents. To verify independently. It is possible that Howard confronted Erez at this point. We aren’t sure.

In October of 2024, Howard had a grandchild. He invited Erez to the bris. It was, by all accounts, a warm and joyful occasion. The kind of day where everyone is their best self and the world feels briefly reparable. Erez drove Howard home afterward.

Two days later, Howard did not show up to the office he shared with Bill. His girlfriend couldn’t reach him. Bill went to Howard’s home. Nobody answered. He fetched the superintendent to get the keys.

They found him dead.

He was seated in a chair, fully dressed in a suit. He had a scotch in one hand and a smoke in the other. He had died very suddenly. According to Jewish law, no autopsy was performed.

What actually killed Howard Miller will never be known.

I’m going to dwell on that for a moment, because I’ve been sitting with it for months, and I haven’t gotten to the bottom of it yet. Howard was the nicest man I knew. He believed in the company because he believed in Dina, and he believed in Erez because she did, and the chain of trust was built on love, every link of it, and that made it both the strongest chain I’ve ever seen and the most precarious.

Whether the stress of discovering the fraud contributed to his death, perhaps. Whether a confrontation with Erez in those final days played a role, I don’t know. What I know is that Howard is gone, and that the man who drove him home from his grandchild’s bris was lying to him about everything, and that the truth, whatever it was, however much of it Howard glimpsed, may have been the last thing he processed.

Howard deserved better. Better than Erez, better than PromAI, and better than the slow, stupid cruelty of being lied to by someone he trusted. Could I have saved him? Was I too careful, too measured, too committed to the polite approach? Maybe if I’d been louder earlier, if I’d told him about the Top Gun trophy when I first learned it was fake, things would have been different. I question this constantly. I’ll think about it for the rest of my life.

Despite everything, Bill went to London with Erez. He reported that potential customers marveled at the software. Erez emailed me again: $33 million valuation, last chance before the Boeing contract kicked in. He’d hired a seasoned COO and a general counsel from a major Israeli law firm.

Hmm. A serious lawyer would presumably have done due diligence before joining a company. Maybe this was evidence that something real existed beneath the lies. Perhaps the product indeed had substance.

I joined a Zoom with Erez, Dina, and the new executives. The COO turned
out to be Maya, Dina’s very capable daughter. I listened politely, then asked for a private follow-up with just Maya and the general counsel.

On that call, I asked if they’d diligenced the company. Maya said she’d spoken to the head of Unit 8200, the Israeli military’s cyber intelligence unit, and they’d said Erez was “OK.”

“OK?” I said.

“Yes. OK.”

I let the word linger for a moment, the way you let a foul odor sit before opening that taxicab window. “OK” is no term of endorsement. “OK” is what you say about a restaurant where nobody got food poisoning. “OK” is the minimum viable adjective. It is the word they use when they have nothing to say and don’t want to say it. “OK” is an arrow in the quiver of a bullshitter.

“I’m not going to invest,” I said.

Passover again. April 2025. I flew to Tel Aviv a week before my family, to attend a cousin’s wedding and see some friends.

My friend Arik picked me up at the hotel in his Tesla Model X, the one with the airplane yoke steering wheel, which he gripped with the casual authority of a man who once redesigned aviation security for an entire airline and now drives to dinner the same way he probably drove to classified briefings: fast, confident, and with both hands on the yoke. Two other men were in the backseat. One of them was Yaki, the former head of Yahbal, an elite unit within the Israeli equivalent of the FBI. Yaki is a weathered man in his 70s who has spent a lifetime extracting truth from people who didn’t want to give it. He has the quiet conviction of someone who has seen everything and is no longer surprised by any of it. He’s the kind of man who orders coffee and somehow the coffee is intimidated.

We had over an hour to drive. I said, “Yaki, tell me what you can learn about someone named Erez Ben-Ishay. He claims to be head test pilot of the IAF.”

Yaki began typing on his phone. Arik and I discussed some other business. The two men in the backseat said nothing. I watched the Israeli highway roll by in the dark and waited. I had been waiting for information about this company for two years. Another 30 minutes was nothing.

Thirty minutes later, Yaki looked up.

“Erez has 30 Facebook friends in common with me,” he said. “They are all desperate ladies and hookers. He is married and was never a pilot. Nobody has heard of this guy. Nothing checks out.”

I said, “He’s not married. He’s dating Dina Goren-Bar.”

Yaki looked at me. He didn’t say anything. He didn’t need to. The look said: He is married. I don’t care what he told you. It also said: I found this out in 30 minutes on my phone while sitting in a car. What have you people been doing for two years, Habibi? He had a point.

I sat with this in the dark car and felt something that I would describe as the absence of surprise. When you’ve been watching a building lean for two years, you don’t gasp when it falls. You just try not to choke on the dust.

Two days later, Maya and the general counsel WhatsApped me and asked to meet. I expected another pitch for money.

They sat down in the Kempinski lobby and thanked me. They said I had saved their careers.

After my conversation with them, they’d realized that they too had never directly spoken to a customer. Not one. They had taken Erez’s word for everything, the way everyone had taken Erez’s word for everything, because Erez’s narrative appealed to each and every person who ever missed out on a deal, didn’t get that invite to a party, or badly needed a break.

So they, together with Dina, went to Erez’s home to confront him. His home was disheveled. His daughter said he’d gone to the emergency room the night before with what might have been a heart attack. They went to the hospital.

And there, in a hospital room in Israel, Dina Goren-Bar walked in and found Erez sitting up in bed next to his wife.

His wife.

Dina, who had believed she was Erez’s girlfriend for years, who had restructured her personal and professional life around this man, who had introduced him to investors as her partner, who had sat on a sofa in my office next to him while he announced they were “together,” saw a woman she had never met, sitting at Erez’s bedside, and understood in a single frame that everything was fake. The customers. The contracts. The air force career. The relationship. Her entire life for the past several years had been built on a foundation that did not exist.

I wasn’t there. I can’t tell you what her face looked like. Maya told me that Dina has yet to recover.

They took Erez’s phone and computer. When they opened the laptop, everything was there. Preserved in his ChatGPT history, like a diary he never meant anyone to read. Every fake customer communication. Every fabricated Boeing contract. Every doctored logo. Every formal investment email he’d ever sent me, the fundraising solicitations that arrived in my inbox at regular intervals for two years, all of it generated by a chatbot. Even the follow-up emails mentioning Joey by name. Even the messages are calibrated to sound warm but professional. All of it: typed into a prompt, generated by an algorithm, copied into Gmail, and sent to me as if a human being had written it.

Here is where I am still stuck.

Erez never stole money from the company. He never tried to sell his stake. He raised the valuation from $15 million to $33 million over two years, but he never cashed out. He fabricated an entire corporate reality, customers, contracts, financials, and he did it with a persistence that was Gatsby-esque. But he never tried to extract wealth from the operation.

So what was the point?

The best I can figure is that the corporate fraud and the personal deception were two pieces of the same mosaic. Erez built the fighter pilot legend to explain his nocturnal absences to his wife. He built the Pentagon-sleepover story to explain to Dina why he went silent in Washington, if he was even there. Every lie in the business served a parallel function in his personal life, and vice versa. Together, they supported the weight of a life that was entirely constructed and, apparently, entirely unsustainable.

He was, in a sense, running two simultaneous cons, one on his investors and one on his wife, using the same material. The suit was both a business costume and a disguise. The bomber jacket was both a prop and an alibi. And ChatGPT was the tool that made it all possible: an anonymous machine accomplice that could generate professional documents, credible emails, and convincing corporate language more rapidly than any human liar could produce on his own. Erez didn’t need to be a good writer. He didn’t need to be a good forger. He just needed to type a prompt and hit enter.

This is, I think, the part of the story that’s most relevant to anyone who isn’t personally invested in the fate of an Israeli AI startup. We have entered an era in which the tools of institutional trust, contracts, letterheads, professional correspondence, corporate identity, can be generated in seconds by anyone with 20 bucks per month and Wi-Fi. The barrier to creating a convincing lie has plummeted. Don’t get me wrong, I have nothing against language models. I even used AI to edit and check the grammar of my initial draft and would have used it for my final version, had my editors not stepped in and informed me Tablet has a strict policy of publishing only human beings. In today’s divisive world, I am thankful they still consider me, an ex-hedge fund guy, human.

Erez Ben-Ishay was no sophisticated criminal. He was a mediocre man with a ChatGPT subscription and a taste for adventure. And he nearly got away with it, not because his lies were good, but because the people around him wanted to believe, and the documents looked real enough to sustain the illusion.

I think about Howard, who invested out of love and believed out of loyalty. I think about Dina, who built something genuinely brilliant and handed it to a man who turned it into a stage set. I think about Bill, who put in a token check to support his best friend, because that’s what friends do, and because, hey, you never know.

I think about Joey, dancing at his desk inside a company that didn’t exist in the way it claimed to exist, headphones on, full of energy and life, oblivious and beautiful and 18 years old. Howard’s shaky phone videos of my son are the things I can’t stop watching. They’re the thing that makes this story something other than a case study in fraud. They’re a record of innocence inside a machine of lies, filmed by a kind man who would be dead within months, sent to a father who could see the danger but couldn’t yet name it, and who put his son there anyway, and who watches the videos over and over, alone, looking for something he can’t quite identify. Forgiveness, maybe. Or proof that it was worth it. Or just the sight of his kid, happy and unaware, in a place that didn’t deserve him.

And I think about the question I still can’t answer: Was the software actually good?

Because the strange, maddening, irreducible fact is that 40 engineers built something for two years. The product may have been genuinely valuable. And if it was, if underneath all the fabricated contracts and fake customers and ChatGPT-generated correspondence, there was a real technology that could really optimize manufacturing lines, then the tragedy of PromAI isn’t just that a con man fooled a group of investors. It’s that a brilliant woman’s life work was consumed by a fraud so total that nobody will ever know what it was actually worth.

That’s the thing about lies of this scale. They don’t just destroy what’s fake. They make it impossible to trust what’s real.

Joey is at Wharton now. He’s doing well. He still can’t sit still. He still shakes his head when the subject comes up. “I was inside a movie set, Pops. I was literally inside a fake company, vibing at my desk, and I had no idea.” He says it with wonder now, not anger.

Joey told me recently that his time at PromAI taught him more than any class he’s taken. I asked him what, specifically, he learned.

He thought about it, bouncing his leg the way he always does, his brain working at the speed of his body.

“It’s like Russian nesting dolls,” he said. “You keep opening them because you think there’s something inside. But the last one is solid and empty. That’s the truth. It’s smaller than you expected. And I think that’s what happens when AI is doing the lying for you. The machine never runs out of layers. It will generate dolls forever. It doesn’t stop because it doesn’t know what it’s seeking. It’s just pattern after pattern after pattern. Erez found the perfect partner. A liar that never gets tired and never asks why.”

I looked at my son. I didn’t say anything. I didn’t need to.

He’s going to be fine.

Jacob is just getting started with his writing career and you can follow him at @jwdoft on X/Twitter and @jakedoft on Substack.

Do you have any thoughts, questions, or feedback? Tell us more in the comment section or at [email protected].