Blumberg Capital On What Investors Really Want From AI Founders Now

[00:00:00] - [Speaker 0]
I'm only able to record over 60 interviews a month and cover events around the world. Thanks to partners like NordLayer, who not only support the show, but the entire Tech Talks network. But today, I wanna talk about a challenge I hear all the time from founders. You know security matters, but you don't have the budget or the luxury of a big team to run enterprise grade tools. And yet, you're still a target.

[00:00:25] - [Speaker 0]
Now the NordLayer new browser feels like it was built with this exact scenario in mind because it gives you enterprise level control over how you access your SaaS apps, move data, and browse the web. But most importantly, without the cost or complexity that often comes with it. Best of all, it's easy to deploy, simply to manage, and will feel very familiar for your team from day one, which means you'll actually get adoption instead of resistance. And that is the part that often gets overlooked. But the best security tool is of course the one that people actually use.

[00:01:02] - [Speaker 0]
So if you're scaling a business and want to get ahead of risk without slowing growth, take a look at nordlayer.com/browser, and you can find out more information there. But now, on with today's show. Picture, not just about technology, but about how ideas, ambition, and timing all come together to help reshape industries and even entire economies. And my guest today is David Bloomberg. He is a venture capitalist, founder, and managing partner at Bloomberg Capital.

[00:01:44] - [Speaker 0]
And he's someone that spent decades backing companies at the earliest stages, including names like Nutanix, Braze, and Double Verify. But in our conversation today, he will share the mindset that guides those early investment decisions and talk about why the most successful founders are usually the ones solving difficult, meaningful problems and doing so not just with technology, but deep understanding of the industries they serve. So we'll explore today how AI is shifting from simply generating answers to actually running workflows, automating processes, and changing how businesses operate day to day. So from health care to mining to finance, David will offer real world examples today of where AI is already transforming productivity and why he believes we're now entering a period of unprecedented innovation. So if you've ever wondered how investors spot the next generation of industry defining companies or if you're a founder, what you should be thinking about right now as AI accelerates and how you get to appear on the radar of a VC, this conversation is packed with useful insights.

[00:04:23] - [Speaker 0]
But enough for me. Let me introduce you to my guest now. So thank you for joining me on the podcast today, David. Can you tell everyone listening a little about who you are and what you do?

[00:04:36] - [Speaker 1]
Sure. Well, I'm, David Blumberg, founder and managing partner of Blumberg Capital. We are an early and growth stage venture capital investor, plus more. I'll explain the more. I grew up in a town called Fresno, California.

[00:04:52] - [Speaker 1]
I it's a middle class agribusiness town. So, you know, I knew a lot of farmers and ranchers and spent my days either playing tennis or hiking in the mountains and good good, you know, life, I would say. Great parents, great family. And I was able to lucky enough to go to Harvard College where I studied international relations and economics. Thought I wanted to work in Washington.

[00:05:14] - [Speaker 1]
I was quickly disabused of that notion because I took three internships there, and I realized that instead of solving the problems, what I found that most of government does is causes problems. At the same same time, I was working while my way through school running a business. It still goes to this day. One of the Harvard distribution it's called Harvard Distribution Services. It's part of Harvard student agencies.

[00:05:38] - [Speaker 1]
And what I learned there is the joy of being an entrepreneur. The feedback loop is so quick between when you get an order and you solve a customer's problem and them saying thank you and then they're paying you for it, and then you're able to hire more people. Or if you screw up, they say, that was bad. And then you say, uh-oh. How could I fix it?

[00:05:58] - [Speaker 1]
And you get, again, this feedback loop. So it's very immediate, very powerful, and and rewarding. So, I was hiring a lot of my friends at school. I paid my way through through through college. I think it helped me get into the business school, which I got into, which was Stanford, also a great place to to be educated at the time.

[00:06:15] - [Speaker 1]
So my green light, I think, that went off over my head about the difference between working in government, which is, like, what I thought was supposed to solve big problems, and what I do now is that what I realized through long story, I won't bore you with all the details, but it was partly writing my thesis. It was partly through this running this business at college and then partly through my academic studies. I realized it's really science plus capitalism that solve most problems. It's this combination of human drive, ambition, initiative, and the reward of gain, and gain is good. I'll explain why.

[00:06:52] - [Speaker 1]
And then the innovation, which sometimes come through determined research efforts and sometimes comes through luck and through accidents and through spontaneous discoveries of things that weren't intended, but, oh, that's useful for such and such. I was trying for x, and I came out with y, and that's gonna be valuable valuable. So it's that combination of new physical reality and then the human ability to drive it to a useful purpose for humanity. That motivates me greatly because, again, I think that's the way the world progresses materially to help people live better lives, to alleviate poverty, to alleviate difficulties, and so on. And then the reward is, you know, what we receive for helping our fellow human beings do better in their lives with improvement of productivity.

[00:07:46] - [Speaker 1]
That's mainly what it's about, and I'll bring that back to what we do today. Venture capitalists and our firm in particular is very focused on investing in entrepreneurs that are addressing that Pareto optimal boundary of productivity. We're trying to push the edge of the envelope outward toward optimal outcomes between resource use, time of work, and the outcome and capital. Actually, those are three inputs. So we're trying to help mostly businesses do things more productively so they could serve their customers more efficiently, and we drive the whole virtuous cycle upwards and onwards.

[00:08:26] - [Speaker 0]
What a great story, a journey that you've been on, and I would echo what you said there. And after, what, 4,000 interviews on here, the role of serendipity in everything comes up time and time again. And and looking back at your career now, mean, you've backed companies like Nutanix, Braze, and Double Verify at very early stages. And I'm curious, if you now look back at those early investments, what what were the signals that helped you recognize founders or ideas that have the potential to become industry defining companies? I suppose it's not just one thing, but I'm curious what you see and and what you feel were the sparks that you saw then.

[00:09:06] - [Speaker 1]
Well, since you're British, I think I'll quote my mother who was quoting, I believe, an English gentleman from the, I don't know, eighteenth century, a nobleman who said that before I had children, I had three theories about raising them. Then I had three children, and I no longer have any theories. That's true. That same thing. So I do have some theories, and yet each of these companies, Nutanix, Braze, Double Verify, teach us somewhat different lessons.

[00:09:32] - [Speaker 1]
I think all the lessons are valuable, so I'll just throw them out one by one. Let's start with Nutanix. That was the one of the first ones. That one, one of our best early indicators was we had a prepared mind collectively as our firm. Why?

[00:09:47] - [Speaker 1]
Well, Bloomberg Capital has a number of about 25 people. We're in four different offices, Miami, New York, San Francisco, and Tel Aviv. So we're we think where a lot of the great entrepreneurs of the world are located. We try and work with them. Our investors come from all over, and most of our companies are trying to hit The US market as early adopter for first customers, for first hires, mostly on the market facing side.

[00:10:11] - [Speaker 1]
Wherever the companies start, they usually end up with very good engineers, Whether they're in Silicon Valley or in Miami or New York or Tel Aviv or London or somewhere, the engineers, they usually find them. But if they're going into The US market from overseas, say, or Israel or England, then they're gonna wanna find local people here to help lead the sales and marketing efforts and so on. So we we try and help a lot with that. The thing that we do is have this prepared mind by leveraging what we call our innovation council. Our innovation council consists of 100 plus between one hundred and two hundred CIOs, chief information officers, chief security officers, chief technical officers, chief marketing officers, so on.

[00:10:55] - [Speaker 1]
People who buy software, who manage big divisions of companies with spending allocations and budgets, and they're looking for better solutions. They're looking for productivity improvements. They're looking for new ways of serving their customers and doing things better. So we try and sell to them. And so what we do is we gather these folks eight times a year virtually and in person in New York and in San Francisco, and we ask them what are the problems they're seeing, what are the challenges they're confronting, what are the ambitions that they have for new product entry, and so on.

[00:11:28] - [Speaker 1]
And so we have that as sort of a list of what we're looking for. And then we're out at conferences. We're out reading the papers. We're out, you know, cruising around on social media and and cocktail parties and so on, looking for entrepreneurs solving some of these problems. So to Nutanix, we had heard for months from the CIO council that digital storage growth was accelerating at an exponential rate, and it couldn't stay on the same cost curve.

[00:11:56] - [Speaker 1]
The CFOs of these companies, the chief financial officers, were demanding that they find a lower cost curve. You have to find a cheaper solution. You can't just keep buying more and more storage systems from EMC and the other big providers. They were too expensive. They were difficult to maintain, and you had to buy a big chunk at a time that wasn't modular and easily added additive.

[00:12:20] - [Speaker 1]
So when the founders of Nutanix came to us, they had a solution that was modular. It was based on commodity hardware. Didn't matter what kind of hardware. It was interchangeable, and it was concentrating the value and the flexibility in the software layer. And that's what they did.

[00:12:39] - [Speaker 1]
So it's called hyperconverged infrastructure. Now it has a number of names, and it's expanded their offering. But, essentially, they were the first to virtualize the digital storage domain, and now they're working in hybrid both in on premise and in cloud, which is what everybody wants. Anyway,

[00:12:57] - [Speaker 0]
the you know,

[00:12:57] - [Speaker 1]
they went from we were the first investor, I think, at a $8,000,000 valuation. Now they're, you know, public on the stock exchange for something like $11,000,000,000 today, doing really well, employ, I think, over a thousand people all over the world and really solving people's problems. So the thing we had there was prepared mind was number one. Also, there were not a lot of competitors, and most of their competitors were doing it in a completely different way, the traditional way. So these guys were the rebels.

[00:13:24] - [Speaker 1]
So that meant they had a sort of an open field when they would come to a customer. They could sort of put all their competitors into one barrel and say, they do it that way, and here's why our solution has some superior aspects.

[00:13:37] - [Speaker 0]
And and, also, before you join me on the podcast today, I was doing a little research on you, and you said that this year, 2026, is gonna be the year that intelligence becomes operational. So I'm curious here. From your unique vantage point as an early stage investor, what does that transition look like as AI moves from just generating insights to actually taking the reins and and taking action inside real business workflows? What do you see here?

[00:14:06] - [Speaker 1]
Well, it's a wonderful question. I'll try and be brief, but it deserves a very long answer, maybe a whole lot in and of itself, but I'll try and be quick. But let's go you're British. So I'm gonna give homage to our, you know, forefathers from England because Bletchley Park, Turing is where a lot of the earliest earliest vintages of AI started with some of that code breaking and the pioneering work that they did. Then move forward maybe thirty years to MIT in the nineteen seventies, and I was privileged to be actually on the board of an AI company in the nineteen eighties, mid eighties.

[00:14:40] - [Speaker 1]
So that was very, very early. Now we didn't succeed in this company. It was a Canadian company based in Edmonton. It was doing parallel processing, but they didn't have NVIDIA. They didn't have these GPUs, the the graphic processor units, which, by the way, were never invented for AI.

[00:14:57] - [Speaker 1]
They were invented for video gaming. Okay? And then they were then they found a new use again, serendipity. Again, they found a new use case in crypto mining. And then now, lo and behold, lo and behold, they are great for this multiple processing parallel processing that is required for neural network analysis that is the basis of AI.

[00:15:17] - [Speaker 1]
So the hardware had to be ready before the software could work. And then the software had to be designed newly because the old software was sort of single threaded, and we needed the multi. Alright. So kudos to all the great big shoulders we stand on today. We are really gifted by history to be living where we'd live now.

[00:15:37] - [Speaker 1]
We're in a miraculous age. The world is getting better really fast. Poverty is being alleviated very quickly. We're solving a lot of the biggest problems of of of the world from health care to environment to food in in many, many, many exciting ways, and AI is here very much to help. So I'm an optimist about AI.

[00:15:55] - [Speaker 1]
I'm not a doomster. And let me explain a couple of the points. Op, you asked about why is it becoming operational. Well, first of all, let's understand that all the revolutions in history have been been about productivity. And if we're worried about jobs, we we can say that in 1776, when our country politely separated from your country, 95% of Americans, and I'd say probably a slightly less percentage in England, worked in agriculture.

[00:16:22] - [Speaker 1]
So all of us now today, it's 2% in America. So that means that 93% of Americans, we have a much bigger population, lost their jobs. But they went into other things, and we live a much better life today, and we are able to farm much more efficiently. And who would go back to the days when we had to pull the plow by ourselves or with our oxen. You know, now we have combine harvesters that are guided by AI, and they plow in straight rows, and and it it's so much more efficient and effective, and our food of cost of food is much lower for that.

[00:16:50] - [Speaker 1]
Alright. Now operational. The AI that most people think about is from ChatGPT or Anthropic or Gemini, so on, or Grok. That is giving you a good answer. K?

[00:17:02] - [Speaker 1]
A very good answer out of a very complex set of data that's been scraped from the Internet and amassed as this the world's body of knowledge. Now it's not all of the world's knowledge. It's actually what's open source knowledge or that they have bought. What Lumber Capital looks for, and I think a lot of VCs, we're looking for proprietary datasets or mastery of a very complex data flow and and and automating that. And here's where we get to the operationalization of it.

[00:17:29] - [Speaker 1]
The ChatGPT algorithm essentially open the large language models are answering the fundamental simple query. What is the next logical word that follows in a sequence? So I'll give you the test. To be or not to Be. Again, you are just as good as ChatGPT.

[00:17:48] - [Speaker 1]
Excellent. I hope they're paying you equally. And now that is great. It's great to get a complex set of data to be distilled into a wonderful answer. However, if we can also then take that answer and operationalize it by automating a series of acts, a series of activities, actions, or operations that are traditionally manual, and even if they're not manual, they take some time to get permission and gating and and logins and all that stuff.

[00:18:19] - [Speaker 1]
Well, guess what? Agentic AI is here to automate the whole shooting match from soup to nuts or whatever metaphor we wanna use. The idea is that Agentic AI is going to understand your beginning point and your goal, and it will jump the hurdles necessary to achieve that pulling from different data sources, logging in when necessary, putting in passwords, doing what it needs to do to get through the steps, and then making the appropriate decisions, you know, with intervention if needed from a human, and then helping you solve problems. So what I say is to simplify it is that the ChatGPT anthropic Basic type of LLMs are solving answers, and AgenTic AI, the promise of it, and I'll give you an example in a moment, is solving operations and activities and automating workflows. Is that fair?

[00:19:17] - [Speaker 0]
Yeah. Yeah. Perfectly.

[00:19:18] - [Speaker 1]
Let me give you an example from just recently. One of our LPs runs a debt collection company, and it helps people reconcile their fact that they've been overspending, and and they they have too much debt, and they need to reconcile it down and and and so on. What this company does is it has independent individuals on phones or sometimes operating on chat to to clients and trying to reconcile and and resolve their debt situations. And a trained individual human being in America in 2026 can typically resolve about 12 to 15 of those cases per week. That's a that's a highly functioning, well paid, and and productive worker.

[00:19:58] - [Speaker 1]
About three months ago, they decided to train a bot, an agentic bot, to do some of the similar kind of project. They trained it out in San Francisco with a small firm of consultants, brought it into in house here in Florida, turned it on. And in the first week, it was able to resolve one hundred and twenty six cases, 10 x exact increase in productivity. We're seeing it in other companies where folks are getting a five x increase in productivity, the minimum I've seen is 50% better. So we are sitting on an edge of a time in human history much like it was seeing our ancestors work go from horse and plow to a combine harvesters.

[00:20:43] - [Speaker 1]
But it happens so this is happening so fast. And if I can double down on this speed and the acceleration of change, I would cite to you that I think it was ChatGPD OpenAI's version 5.3 and Anthropix version 4.6, which were relatively new, were the first time in human history where the software of the new version was mostly written by the old version. The AI improved itself. Yeah. Now that is both exciting and slightly scary because, wow, where will that end?

[00:21:17] - [Speaker 1]
We don't know. But we do know that the it never sleeps. It can just keep working all the time and improving itself and, you know, optimizing. And so we have a future in front of us with it which will definitely boost the productivity. We're already seeing it in The US.

[00:21:35] - [Speaker 1]
I'm I'm curious about The UK statistics. The US, the last two quarters, we've had productivity gains of over 4%, four point one and four point nine, I believe, if I'm precise. And and that's very, very high in in in in in recent history. Won't go into boring detail about that, but productivity is why is it important? Well, obviously, in a person's life, if you're more productive, you get more out of a day's activities.

[00:22:02] - [Speaker 1]
But the productivity is also tied in economics to wages. Many people are concerned that, oh, people are gonna be having wages go down. Well, no. It's just the opposite. When we inject capital into a factory and build new machines for the workers, the workers are more productive.

[00:22:17] - [Speaker 1]
That is the only way in economics, if we're trying to be fair about it, that workers can be paid more and the business can still make money. So if the workers are more productive, then the boss can pay them more, and everyone does better. Now you might say, well, but they might not need as many workers. That's fair. But then I go back to the fact that America in 7076, 95% of Americans worked in agriculture, and today, it's 2%.

[00:22:39] - [Speaker 1]
So all of us live better because we've done different things. And the wealth that's going to be created, in this and the productivity savings are gonna get rid of a lot of the mundane jobs that copy and paste, the entering forms. This morning, I was at a doctor. I had to fill out about, I don't know, 10 different forms for one very simple X-ray.

[00:23:00] - [Speaker 0]
Yeah.

[00:23:00] - [Speaker 1]
I thought it was crazy. Now I should have a little personal agent that will follow me around. It'll be on my phone, and I can just hand it to the woman or, you know, it'll beam it aboard, and and it'll have all of my data and answer the insurance questions and my date of birth and and all that kind of stuff and verify and validate. We do so much wasted work.

[00:23:22] - [Speaker 0]
Yeah.

[00:23:22] - [Speaker 1]
Most of us do a lot of wasted work. If you think of your average day, well, it can be made a lot more effective and a lot more efficient in both in our personal lives and in our work world. Now I will tell you the downside. What I'm worried about is not so much the inequality within a country, like within The United States, because I think that most people will be brought up by this technology. The price of goods will go down because of we're so much more productive at making them.

[00:23:51] - [Speaker 1]
Our supply chains will get better and more efficient. That will help as well. What I'm a little bit worried about is that this technology is really concentrated in a few places. The US, Israel, UK, Singapore, Japan, China, few places on earth. There are a lot of countries that don't have much of anything in this, and now they'll benefit as users.

[00:24:13] - [Speaker 1]
But in terms of producers that are that are gonna be accruing the value of the companies that are built here, look at the magnificent seven that are, you know, are big companies on the stock market today. They're all American, and they're all relatively new companies. They didn't exist in 1950. None of that. I don't think any of them existed in 1970 or 1980.

[00:24:33] - [Speaker 1]
They're relatively new. So what I think is gonna happen is that The US and other companies like countries like this, like Israel, Canada, UK, are gonna rise very quickly in GDP per capita, in wealth, in power, and an ability to project force as we're seeing today in in the battle with Iran. This is a technology driven war. The AI is assuming an incredibly powerful part of the winning strategy, if I may say so. So I think that countries, if we're talking to nation national leaders or if we're talking about business leaders, get on it.

[00:25:10] - [Speaker 1]
Lean forward into AI. You cannot avoid this. This is like the change of buggy whip and horses to the automobile, but it's gonna happen in five or ten years, not, you know, half a century. And I just urge people to really jump into your personal life. Start it.

[00:25:29] - [Speaker 1]
Learn with your children because your children are gonna be much more intuitively interested and capable than some of us older folks. But if you're a CEO, have all of your employees working with it, trying to improve their own day and submit ideas to your team how to make your business better. It's happening right now, right here, and it's highly productive and as as a change agent.

[00:25:52] - [Speaker 0]
And I guess, predictably, I was reading, predictably, I was reading that around 90% of the companies that you evaluate today are grounded in AI. And a question I've got to ask, with so many founders entering this space, so many different companies spinning up almost every week, what separates the startups that genuinely stand out from those that are just simply adding AI into an existing product or maybe claiming to have AI or don't have AI at all. It must make your job incredibly challenging with so many out there.

[00:26:23] - [Speaker 1]
Well, I don't think so. Respectfully, I think it actually because we have a philosophy, and I'll explain it in a moment, we stick to our philosophy, and therefore, we're able to weed out a lot of what I call weed from the chaff. Yeah?

[00:26:34] - [Speaker 0]
Yeah. Yeah.

[00:26:34] - [Speaker 1]
We do it is again, I think I said this earlier, but I'll reframe it and and reinstate restate it. We look for companies that are driven by amazing teams because a great team with a mediocre technology will improve. A great technology run by a mediocre team will probably not succeed. So we look for great people, first of all. World is about people even in the world of AI.

[00:26:57] - [Speaker 1]
And I like to also throw in this. There's a professor at Berkeley, I forget his name, wonderful man, who said that the artificial intelligence is a very poor choice of words. We should really call this augmented intelligence because the purpose of all of this, the purpose of what we do should be about human flourishing. Yeah. And and that's what the profit motive is at its base.

[00:27:16] - [Speaker 1]
If I can serve my fellow human being, I will be rewarded for it. Whatever your domain, whether you're a nurse, a doctor, a basketball player, a programmer, if you serve more people with value, you generally receive more. And so what we're trying to do is show that this productivity gain can accrue to all fields, and everyone will be better off. How do we choose companies? It was your question.

[00:27:40] - [Speaker 1]
The fact is that there are a lot of people applying it, but often people are doing what's called a wrapper, a simple sort of, oh, we're dot AI now. We used to be basic. Now we're dot AI. So that's fine. They might be getting some operational improvements.

[00:27:52] - [Speaker 1]
But unless they are native AI, if they're building their product to be an AI based tool, they're probably not gonna get the same kind of stellar, results. And what we particularly look for are proprietary datasets or proprietary algorithms. The big foundation models are open source or relatively cheap and available, and those big companies will be very successful, the big ones. So I would employ a sort of a barbell strategy. You have to own those big companies that have those models.

[00:28:23] - [Speaker 1]
And then these I think the one of the great some of the great winners of today are being built right now in the private sector. They're they're I mean, private companies. They're small. They're startups, but they're latching onto a certain database with that they own or they construct with their customers. And I can give you a couple examples.

[00:28:40] - [Speaker 1]
One is a company called Telen AI, which is automating auditing. Now many people think that auditing is a very boring subject, but good thing about auditing is that all the data is highly structured. So it's very accessible for these models to add and divide and subtract and and and put into what an auditor needs for their reporting. And so this they've devised that there are about 85 modules to do a complete audit for a large comp corporation, and there are about 30 done. They've they've heard knocked off about 30 of them.

[00:29:12] - [Speaker 1]
And they worked with midsized accounting firms, and they make these accounting firms much more productive. A senior partner can handle many more clients. We're we're applying the same thing with law firms. We're applying the same thing with, genetic testing, same thing with mining engineers. That's an interesting application.

[00:29:30] - [Speaker 1]
One of our companies called Verdash AI uses US geological survey, Peruvian, Canadian, whatever the government survey is, which is public, and private data that is very proprietary. It's only owned by the miners and the explorer explorer people themselves, the geologists. They take this data. They analyze it against existing ore bodies underground, and then they will help you find ore bodies where it's it's under what's called covered terrain. There are no outcroppings.

[00:29:59] - [Speaker 1]
You can't, like, chip at the rock and find the gold in the quartz vein. It's under a big desert, and you'd have to drill thousands of holes to find out where the veins of quartz and with the gold in it are running. So this is an amazing breakthrough because instead of a thousand drill holes, you can drill 10 with the same productivity, and you can do it faster and cheaper and much less environmental degradation. So those are just a few range of things in the genetics realm. I'll give you another example of a company from Israel called Imogene, where, a lady was diagnosed with stage four lung cancer, and it had spread up into the brain, and and the MRI scans could show the lesions in her brain.

[00:30:38] - [Speaker 1]
And she had a four to eight weeks left, you know, of life expectancy at one of the doctors. And she said to them, is there anything we can do? And they said, well, the problem is that lung cancer has 28 varieties of genetic types. Three of them have very well developed chemo treatments or immunotherapy that can be life saving to you. And I think it's immunotherapy, not not chemo.

[00:31:02] - [Speaker 1]
And the she said but the doctor said the problem for you is that the traditional genetic testing takes three to four weeks. But there is this new little startup company called Imogene. It's not yet FDA approved yet, but they have gotten very, speedy results. They can give you an answer in a minute. And she said, let's try it.

[00:31:21] - [Speaker 1]
And they tried it, and she got back the answer that day. And it was your this is your lucky unlucky day. You have stage four lung cancer, but you have one of the three for which there is probably a cure. Would you like to try the go right now. That lady is still walking around years later.

[00:31:36] - [Speaker 0]
Amazing.

[00:31:37] - [Speaker 1]
You know, while people are rightly afraid that some dangerous aspects of AI out of control can be bad news, it's definitely saving lives already. It's definitely helping us be more productive. So, you know, we should really lean forward again, be excited about it. It's a positive future. It's gonna lift most all boats.

[00:31:57] - [Speaker 1]
I don't know of any technology that is completely, you know, benign and positive on every score, but this has overwhelmingly more positives as far as I can see than some of the negatives, which are which are valid, we should address them as well.

[00:32:11] - [Speaker 0]
And you're clearly passionate about technology and AI in particular, but one of the things that really stands out for about you to me as well is you also place a heavy importance and focus on founders who truly understand the industries that they're serving here. So why does deep expertise matter so much when building AI driven companies, and how are you seeing it shape the the kinds of solutions that ultimately succeed? Because we'll we'll see many solutions that pop up and they don't fully understand and they tag AI on and off they go. But this this focus on truly understanding the industries that they serve, it it feels like you've really hit something here.

[00:32:50] - [Speaker 1]
Well, thank you. I I think it's true. I think there are several angles. The first one is to build a better mousetrap. One generally has to know the domain.

[00:32:58] - [Speaker 1]
Sometimes we can, by serendipity, stumble into it, and good good on you if you do it. But most of the time, it comes from sort of a deep background understanding of the alternatives and finding something new. Another point that is probably overlooked by technologists, but it's very important on the sales and marketing side, is knowing the people in the industry, knowing the lingo in the industry, having credibility among peers that you're going to as an as a first time entrepreneur with a new product. What is the the the the fear of the buyer? The fear of the buyer is, uh-oh.

[00:33:31] - [Speaker 1]
This might be a dud. I might be buying a lemon. This might this company might fail. I'll look bad. So if you are from the industry, if you have credibility, if you have domain expertise, if you can talk their lingo, if you've done something of note and created something or, you know, shown that you've written an intelligent paper about why this product might work or shown a demo, that goes a long way to helping you get over that hurdle of early adopting skepticism.

[00:33:58] - [Speaker 1]
Does that make sense?

[00:33:59] - [Speaker 0]
Yeah. Those

[00:34:00] - [Speaker 1]
are couple reasons. And we another important thing I wanna add is that we also want people who have deep, positive, ethical behavior. And domain expertise often means you've lasted in an industry because you weren't a jerk and you didn't cheat and you didn't, you know, steal money from from the from the folks. That goes a long way too. It's not exactly technical expertise, but if you have, you know, grounding in a field and have respect from your peers, that goes a long way as well.

[00:34:31] - [Speaker 1]
And we we look at that as well when we're choosing our entrepreneurs.

[00:34:36] - [Speaker 0]
There has also been a lot of focus on model size and raw computational power, and you've suggested that reliability and accuracy will become much more meaningful measures of success. I completely agree with you on that as well. How do you see that shift reshaping the next generation of AI startups in the future? Any signs you're seeing here or any trends?

[00:34:58] - [Speaker 1]
Yes. Well, I'm gonna start even with the precursor. Remember my experience in the nineteen eighties being on the board of a AI company that the our problem was that the hardware was eighty eighty six chips from Intel. Very, very weak little processors, and we just string them all together. And and they sort of worked, you know, in a in a infantile way, but we didn't have this parallelized software that could work run through each of them.

[00:35:25] - [Speaker 1]
So we needed to wait for the GPUs. Now a lot of people are worried about energy consumption. And I think that there's going to be probably a transition from the current copper and metals based, world to more of a optical type of hardware, and and that should probably drop demand. You know, business is generally about optimizing. And when some commodity goes too high in price, we're gonna try and find substitutes, you know, a better way forward.

[00:35:53] - [Speaker 1]
Because business is competitive, and you can't rest on your laurels and just assume that no one's gonna try and compete with you. If you have a good thing going and you're making lots of money, then other people are looking around and say, I wanna do that too. So you're gonna get commodities, you know, commoditization of most things or or or competitors. Alright. So now go back to your your original question, which is is about increasing reliability.

[00:36:18] - [Speaker 1]
Right? And you have

[00:36:20] - [Speaker 0]
two words, reliability and Oh, accuracy. Reliability and accuracy.

[00:36:24] - [Speaker 1]
Similar similar words. Well, the early models of AI did a lot of hallucinating, and they still do some. That's why a human in the loop is often, required for right now, and reinforcement learning, learning from our, errors is very important. And I can give you an example. We have a company called Theatore.

[00:36:46] - [Speaker 1]
They are, documenting the video stream for surgeries, using endoscopic, tools for urology, gynecology, etcetera. These are, thin wires that go with a video camera at the end that go in through the veins and so on, and they're used, you know, with a scalpel to cut and a suture to sew inside the body for various types of procedures. Now the human doctor, the surgeon, will document after they have done their surgery, and they're on average about 76% accurate. Using AI to watch the whole video, to video everything, to then summarize it and keep the important bits, they're about 95% accurate right out of the box. But when the human doctor is able to reinforce and say, you got this right.

[00:37:37] - [Speaker 1]
Here you missed something. Let me add this note. Then the model learns. That's fed back into the model, and it reinforces the learning ability of this. So we're seeing that, and that's going to have a very cumulative positive accelerating effect.

[00:37:55] - [Speaker 1]
Makes sense? And remember I said that the new versions of the big models wrote themselves their improvements. That's partly because they're taking in feedback of where they made mistakes, where they hallucinated, and saying, how do I avoid that? And I'm anthropomorphizing software. It doesn't exactly say, how did I screw up?

[00:38:17] - [Speaker 1]
But it does try and find the the best possible outcome. That's what this algorithm is is searching for. And so I think we're gonna be doing better and better. Never perfect. There will always be errors.

[00:38:29] - [Speaker 1]
Always be bugs. There will always be, you know, just we're none nothing is perfect. On the other hand, we're asymptotically getting better and better.

[00:38:40] - [Speaker 0]
And if we have any founders listening who are thinking about building an AI company right now, are there at various stage, and maybe they're listening and they wanna appear on your radar. What qualities, strategies, or early decisions will increase the chances of building something that they can endure beyond the current AI hype cycle? I love that question.

[00:38:59] - [Speaker 1]
Alright. Well, we've heard that a few times, you can imagine, and I like to make it simple with a mnemonic, the six t's. I learned from another wonderful venture capitalist. And the six t's that we use are, first, what is the theme? What is the problem you're trying to solve?

[00:39:13] - [Speaker 1]
Tell us crisply what is the problem. Why is it important? Why have you found this insight about how to solve it? That's number one. And why is it a big problem?

[00:39:23] - [Speaker 1]
Next is the team. Who are you? Why are you the right team? What do you have in domain expertise? What do you have the right determination, the persistence, the grit to go the roller coaster of startups?

[00:39:36] - [Speaker 1]
Because it's not always love and roses. It can be very difficult, and you can endure poverty and eat ramen noodles for nights on end and work till midnight and so on. So you gotta have that. You've wanted to have the right ethics that you're gonna be fair and and treat your employees well and treat your customers well and treat your investors well. All that goes into the mix.

[00:39:56] - [Speaker 1]
So you've gotta have domain expertise, ethics, persistence, smarts. You know? And then you have to talk about the third t, which is the terrain. That's a t word for market. What we tell people is we don't wanna know what the market is today because the time we invest until the time we exit with you and hopefully an IPO or great sale to another company or a PE firm is probably gonna be six, seven, eight, nine, ten, twelve years.

[00:40:21] - [Speaker 1]
And so we need to know where the puck is moving as Wayne Gretzky famously said. We wanna look there. So we want you to show us the video of your market, not just the photo. Where is the market moving? How is it changing?

[00:40:34] - [Speaker 1]
What's the dynamics of the regulation, the technology, the socioeconomics of your marketplace, right, and the competitors? That's the third t. The fourth t is the technology. We wanna know how you're doing it, what's the secret sauce, if there are patents, if there's trade secrets, what is it that you know that other people can't easily copy once you're out in the market? And sometimes that will be now, in these days, datasets, proprietary datasets.

[00:41:00] - [Speaker 1]
Alright? Now the fifth t is traction. Where are you in your movement, from zero to something great? Are you a raw company with no team? You're a solo founder.

[00:41:11] - [Speaker 1]
You have no product, no code, nothing, or do you have a demo? Do you have a team? Do you have some customers? Where are you on that process? And that's a risk reduction metric.

[00:41:21] - [Speaker 1]
And every step further you take, probably the higher valuation you'll receive from us and the lower risk it is for us as investors. Okay? So that's the traction. And the last one is terms. How much money is this gonna take?

[00:41:32] - [Speaker 1]
How much time is it gonna take? Not just, again, the snapshot of the first raise, but what are you gonna raise in terms of money going forward so we can understand the dilution that we're likely to face going forward through multiple stages of financing, seed, pre seed, a, b, etcetera, etcetera. So those are the six c's. I'll repeat them. First is the theme.

[00:41:53] - [Speaker 1]
What are you doing? The team. Who is doing it? The terrain. Where in the market are do you fit, and why are you special?

[00:42:02] - [Speaker 1]
The technology, how does it work? The traction, where are you on the process and the and the stages? And the terms, what is needed to get you to be a success? And sometimes it's beyond money. So so those are the things that we look for in a simple way.

[00:42:17] - [Speaker 1]
Really, what we're looking for, if I can make it to just three points, we're looking for people solving big problems, hard problems with a great team. And if you come to us with that and show us how you not will win only in the first stage, but how you'll be a winner years to come because you have something sustainable as an edge, as an advantage, as proprietary, that really gets us going.

[00:42:40] - [Speaker 0]
Love that. I think that's a powerful moment to end on. If anybody listening that have been sat there ticking all these boxes off thinking, hey. I think we need to, look into this. Where where is the best place for anybody listening to to find you online?

[00:42:53] - [Speaker 1]
Well, first of all, I would have everyone, come to our webpage, blumbergcapital.com. We publish a monthly, newsletter, which is called the Blumberg Capital Compass or the Blumberg Compass, I should say. It's a newsletter about what we find interesting both in our portfolio and outside and trends and so on. We're often at events in technology, and please come up to us. Visit us in our offices.

[00:43:15] - [Speaker 1]
We're in San Francisco, Miami, New York, and Tel Aviv. Nice places to visit. A lot of entrepreneurs hang out there. We're very well networked. We have the CIO Innovation Council, and we can help on diligence, and we'd like to help people.

[00:43:30] - [Speaker 1]
So, you know, if you have a question, we'll try and answer it on an infinite time, but we try and be helpful when we can. If we if it's not right for us, you know, ask us for a referral to somebody with whom we think it might be appropriate. That's a fair fair question. And I wanna just give encouragement to the entrepreneurs. I think we are living in a time that has never been better for entrepreneurship in history.

[00:43:51] - [Speaker 1]
And I'm gonna give a special shout out to The United States for a moment because they passed a new tax bill called qualify QSBS, qualified small business stock. It started in the Clinton era, it survived every administration on a bipartisan basis since then. It essentially avoids capital gains tax for early investors and founders up to $15,000,000 per investor or founder. And if you're married, you can double that if you split that with your spouse. And if you have two kids, you can split more with them and shield up to, you know, say, 60,000 for a family of four of gains, and that's per company.

[00:44:28] - [Speaker 1]
So if you're an investor in a venture fund and it has 40 companies and six of them are unicorns, you can shield all the gains from each company. So it has to it only applies to c corps. It only applies to companies that make things. It can't be a law firm, a venture firm, an insurance company. But if you're making software, you're making widgets or whizmo gizmos, then it generally applies.

[00:44:49] - [Speaker 1]
I'm not a tax lawyer, but, you know, check. But it's it's essentially a tax free world for early stage entrepreneurs in America today. And even if you're, say, a British company, you can create a subsidiary or a parent in The United States and get the advantage of it. Same for Israel. Same for Canada.

[00:45:07] - [Speaker 1]
US c corps get the benefit of this. And I didn't know we're gonna talk about this. It may not be interesting, but the technology is here. The regulatory framework is here. The markets are hungry for adoption of innovation.

[00:45:20] - [Speaker 1]
People want this new technology. They see how productive it is. And so if you're an entrepreneur, find a big problem, help, you know, create a solution, come to us as investors. We're ready, willing, and able.

[00:45:33] - [Speaker 0]
Awesome. Well, I will have links to everything that you mentioned there. I urge everyone listening to check that out. And and, also, kudos to you, and good luck on you for finding that next multibillion dollar enterprise software company. Best of luck on your journey.

[00:45:49] - [Speaker 0]
We'll stay in touch. It'd be great to get you back on again in the future. But thank you for sharing your invaluable insights today. Really appreciate your time.

[00:45:56] - [Speaker 1]
Neil, you're a great interviewer. It's a pleasure, and happy to come on again. Thank you.

[00:46:01] - [Speaker 0]
What I enjoyed most about my conversation with David today was that sense of perspective that he brings to the topic of AI and innovation. And, yeah, it's easy to get caught up in the headlines about new models, bigger datasets, or the latest wave of tech startups. But David's view cut through that noise because in his world, the fundamentals still matter. The strength of the founding team, the size of the problem being solved, the depth of industry understanding, and whether the solution can create real productivity gains. These kind of principles are what have guided in venture investing for decades, and they still apply in the age of AI, possibly even more important now.

[00:46:44] - [Speaker 0]
And I also loved his framework there of the six t's for evaluating start ups. Theme, team, terrain, technology, traction, and terms. And I think it's a simple way to think about building a company, but but behind each of these ideas sits a huge amount of discipline, of clarity, and indeed strategic thinking. So if you are a founder listening today, I think there were plenty of signals in this conversation about what investors are really looking for. And if you're simply curious about where AI is heading next, I think David's optimism about productivity and innovation also offers a fascinating glimpse into the future.

[00:47:24] - [Speaker 0]
But as always, love to hear your thoughts. What stood out to you in this conversation, and where do you think the next wave of AI driven companies will emerge? Techtalksnetwork.com. That's where you'll find absolutely everything. And if I don't see you there, hopefully, I'll be speaking into your ears tomorrow.

[00:47:41] - [Speaker 0]
Speak with you then. Bye for now.