Episode #75: Three Breakthroughs: AI Agents: They Came, They Saw, They Conquered

Frequently Asked Questions

FAQ

• Where is the transcript for this webinar?

  Samantha Herrick:
  Welcome to the land of tomorrow. This is the Tech Optimist Podcast, a show brought to you by Alumni Ventures. This is a time where we discuss the people and innovations shaping our future.

  Naren Ramaswamy:
  We thought software was static. Now it’s basically having an intern on demand for everyone.

  Samantha Herrick:
  That is Naren Ramaswamy, senior principal here at Alumni Ventures.

  Mike Collins:
  A lot of the times the people who win are the ones that figure out the new media and the new news.

  Samantha Herrick:
  And that is Mike Collins, co-founder and CEO at Alumni Ventures. And that’s me—my name is Samantha Herrick, and I’m the guide and editor for this show.

  All right, everyone, welcome back to this super exciting episode of The Tech Optimist. This is another Three Breakthroughs episode. We’ve heard this is a crowd favorite, so we are so happy that everyone is back to spend a little bit of time with us as we discuss the future and future technology and all of that super fun stuff.

  So of course, before we hop into Mike and Naren’s breakthroughs for today, I’m going to give you a little snippet—or a little sneak peek—as to what the breakthroughs are. Again, we’re going to keep with the process of not spoiling it and sort of letting you discover them for yourselves.

  So here are some hints for the first breakthroughs: we’ve got robots underwater, we’ve got Iron Man, and we have AI search engines. All right, get those gears turning. Let’s see if you can guess the right breakthroughs for this week.

  But without further ado, we’re going to hop into a quick AV ad, a little disclaimer, and then right into the conversation. So sit tight, don’t go anywhere—we’ll be right back.

  Speaker 4:
  Do you have a venture capital portfolio of cutting-edge startups? Without one, you could be missing out on enormous value creation and a more diversified personal portfolio. Alumni Ventures, ranked a top 20 VC firm by CB Insights, is the leading VC firm for individual investors. Believe in investing in innovation? Visit av.vc/foundation to get started.

  Speaker 5:
  As a reminder, the Tech Optimist podcast is for informational purposes only. It is not personalized advice, and it is not an offer to buy or sell securities. For additional important details, please see the text description accompanying this episode.

  Mike Collins:
  Hello, welcome to Tech Optimist and our Three Breakthroughs show. I am here with Naren again, and we’re going to talk about exciting new things in the area of technology, innovation, and venture capital. We’re venture capitalists at a firm called Alumni Ventures. We’re one of the more active venture capital firms on the planet. We co-invest exclusively with really strong lead investors, but we do a lot of deals. So we see a lot of stuff, and we’ve got exciting things again this week coming up on Halloween. So Naren, why don’t you kick it off today?

  Naren Ramaswamy:
  Yeah, happy to. Mike, this is an interesting one to start with. We’ve been talking about the space economy and SpaceX’s amazing technological feats over the past few weeks. That got me thinking about what’s happening in this field of ocean exploration. I came across this article about some researchers out of Zurich who developed a robotic fish that is being sent underwater to collect environmental DNA in a non-invasive way.

  Samantha Herrick:
  Okay, so never did I think that we would have an episode of this show—or a breakthrough come through one of these weeks—that’s about robotic fish. I guess it could just be the sort of oxymoron that that brings along, right? Robots and water normally don’t mix.

  But I have some information on this robotic fish that I want to share with you before we hop into a quick video, which helps summarize it a lot better than I can.

  So this is from a group of students at ETH in Zurich. They’re engineering students who developed an innovative robotic fish named Eve as part of their SURF-eDNA project. This autonomous underwater vehicle (AUV) is designed to revolutionize ocean studies and marine biodiversity monitoring.

  Some really cool features about Eve: it’s a cutting-edge example of biomimetic robotics technology designed to blend seamlessly into the aquatic ecosystem. It looks like a fish—you’ll see it in the video here. It has a soft structure. The robot features a soft body with a silicone tail that moves naturally, powered by hidden pumps within the body of the fish.

  Eve is equipped with a filter to collect environmental DNA from the water, allowing scientists to gather valuable information about species inhabiting the area without resorting to invasive techniques. It also has advanced sensors—the robot is fitted with a camera and sonar enabling it to avoid obstacles and navigate effectively underwater.

  For its purpose, it monitors an ecosystem. It can help identify certain species living in a body of water and can help explore a habitat. Eve’s design allows it to access and study sensitive ecosystems with minimal disturbance.

  As for future goals and development: they’ve been working on this project for two years, and Eve has been their most recent creation. They aim to scale up the technology to make it accessible to scientists worldwide. They want to provide more detailed insights into ocean habitats threatened by climate change and human activity, and potentially help prevent species from becoming endangered or extinct.

  Eve represents a significant advancement in underwater robotics, offering a promising new approach to marine research that combines cutting-edge technology with environmental sensitivity.

  Now I’m going to play a video here. It’s a YouTube video from Tuan Nguyen Capital titled Meet ‘Eve’ the DNA-Collecting Robot Fish. They’ll provide a little more info and show some imagery. You’ll actually hear from a few of the students and professors who are on board. So without further ado, here’s that. Enjoy.

  Speaker 6:
  Meet Belle. She’s not your average fish, and neither is her sister, Eve. The soft robotic fish were developed by these engineering students from ETH Zurich. Lake Zurich may be some 300 kilometers from the sea, but it’s a perfect testing ground for Dennis Baumann and his SURF-eDNA team.

  Dennis Baumann:
  The main goal of using this kind of robot is to build a reliable tool for biologists. [inaudible]. Yeah, I mean, also it’s just very cool to have a fish robot, I guess.

  Speaker 6:
  The robot can film underwater and collect DNA to monitor marine ecosystems—all while blending in.

  Dennis Baumann:
  By making Eve look like a fish, we are able to be minimally invasive into the ecosystem that we are surveying. That means the fish population is not scared as much by us, and the creatures living there are not scared as much by us.

  Speaker 6:
  Today the team is jumping in to test Eve’s navigation.

  Speaker 8:
  It’s working quite well right now. You could say this is autonomous, maybe a bit like your home robot—your home vacuum cleaner robot. So for now we’re just quite happy if we can see that the fish can actually move around obstacles, avoid any collisions, basically.

  Speaker 6:
  Eve’s fluid swimming motion is thanks to internal pumps hidden in her silicone fin, moving the tail from side to side. It’s here on campus where the students design and develop almost every aspect of the fish, including the movement of its fin.

  Speaker 8:
  It works very well now, but it took a long time. [inaudible].

  Speaker 6:
  The students took inspiration for the project from their professor, Robert Katzschmann, who developed a small robotic fish called SoFi when he was a student at MIT.

  Robert Katzschmann:
  I think there is a bright future ahead of us where we will see robots that are not built in the way that we think of robots today. And that’s what we’re trying to do with this project.

  Speaker 6:
  The team’s robots can also collect DNA samples of the marine environment known as eDNA.

  Speaker 10:
  The fish swims for about 30 minutes, so 10 liters of water can go through this filter. And then this filter collects the particles, the eDNA.

  Speaker 6:
  From this small sample, researchers can get a snapshot of what kind of fish live in the body of water.

  Dennis Baumann:
  We are able to identify the species and also how healthy the population is. By identifying that, we can maybe prevent species from being endangered or dying out.

  Speaker 6:
  As the students continue to refine their tech, they hope one day this fish will change the way scientists study the deep.

  Naren Ramaswamy:
  What’s really interesting to me about this is the convergence of many different technological trends. We talk about autonomous vehicles on the road, SpaceX launching rockets and catching them…

  Mike Collins:
  Yeah.

  Naren Ramaswamy:
  All of that stuff happening underwater now—you have what looks like a fish, it’s not threatening to species, it’s made with this silicone material, so it’s not harming or changing the pH of the water around it or anything like that. It’s purely for scientists to understand in a non-invasive way what’s happening under the water.

  And I am thinking about the immense applications of having this as a platform. One of them is just understanding how many species there are underwater. We think there are 2 million; humans have only explored, I think, 1/8 of that. That’s our level of understanding of what’s happening in the ocean world.

  We can talk about economic opportunities for venture in deep sea mining, which obviously has a negative connotation with regard to the environment. But what if there was a way, using these platforms, to really understand what’s happening under the ocean and do that in a way that is good for the environment—maybe recirculate minerals that are not present in the ocean, actually making it beneficial for animals underwater?

  And so it’s just fascinating to me as a conservation tool as well as a platform to explore an unexplored resource.

  Mike Collins:
  I also think it’s on the theme of the importance of sensors and data. We’re all super excited about artificial intelligence and machine learning, but you need the data and you need sensors. Whether those are on your body, in space looking down, on the surface of the ocean, under the ocean, on the land, in the land, in the air—sensors everywhere.

  Developing the appropriate sensor for the appropriate environment is an area of great opportunity and innovation. And again, as land mammals, we have probably given 99% of our attention to the landmass of this planet. But there’s so much more to be done, so much more to be understood.

  To your point, the future of humanity and the future of the planet are so driven by what covers 75% of it. Some of the more forward-thinking environmentalists are really focused on the oceans. So that’s a good one.

  Naren Ramaswamy:
  Exactly.

  Mike Collins:
  More sensors, please.

  Naren Ramaswamy:
  Yeah, digitizing part of the world that just hasn’t been digitized yet and helping us understand it.

  Mike Collins:
  Yeah. And we see the consequences of hurricanes—for example, this year they’ve just been devastating. But thank goodness we had the data we had, right? A couple of centuries ago, you woke up in the morning and it looked like a nice day and hell’s around the corner.

  So more of that is better. The weather models have gotten a lot better for sure. So my breakthrough of the week is, again, the early indications that agents are coming. And for our listeners, AI agents—in the simplest terms—are bits of software that actually go and do things on your behalf. Speaking into your phone, saying, “You know where I like to eat, make reservations, tell me where we’re going, order me the Uber.” And they go and do that for you.

   

  Mike Collins:
  Something out of Google, Project Jarvis, getting a lot of buzz.

  Naren Ramaswamy:
  Yep.

  Mike Collins:
  There are very strong rumors coming out of some of the larger players in the space—the Apple folks, the OpenAI folks. I think there’s an over-under of about 12 months before these things are really starting to be released in production. And I just want our listeners to know that it will be another step in what is possible in the world when they start doing stuff for us out in the wild.

  Samantha Herrick:
  Okay. So for this breakthrough, I too am a huge Marvel fan. So I went straight to Iron Man and Tony Stark and J.A.R.V.I.S., right? But Google’s Project Jarvis is an experimental AI-powered tool that is currently in development.

  I want to provide a little bit more context about the actual project of Jarvis before we let the guys dissect it and what it means for the rest of the sector. This ambitious project aims to create an autonomous AI agent capable of interacting with and controlling web browsers—and potentially other computer software—to complete tasks on behalf of users.

  Jarvis is designed to navigate web browsers autonomously to perform various tasks. It can handle complex, multi-step processes like booking flights or making online purchases. It can manage emails and schedule appointments. It can conduct research and gather information from the internet, and so much more. It can also interact with on-screen elements such as fields and buttons.

  The AI agent is expected to be powered by Google’s advanced Gemini 2.0 AI model, which utilizes a transformer-based architecture and mixture-of-experts technology.

  As far as development status, the project is still in development and could be previewed as early as this year. We should be getting more information soon. The full release date is uncertain—it’s not out there anywhere—but an internal preview was accidentally leaked briefly through the Chrome Web Browser Extension Store.

  So we don’t know exactly what this looks like yet, but it’s definitely on the horizon.

  Potential impact: If successful, Project Jarvis could revolutionize how users interact with computers and the internet. It represents a shift from passive chatbots to active, autonomous digital assistants. Literally, J.A.R.V.I.S. from Iron Man—just think about having J.A.R.V.I.S. in your computer instead of in a giant exoskeleton superhero suit.

  This technology could have significant implications for productivity and task automation.

  Some industry context: Google is not alone in developing this type of technology. Other major tech companies and AI research firms are also working on this. For example, Microsoft has demonstrated AI agents for email replies, Anthropic has unveiled an early version of agents in Claude, and OpenAI is reportedly working on AI agents as well. Many predict these will hit the mainstream in the near future.

  As the development of Project Jarvis continues, it will be interesting to see how Google addresses the technological challenges and potential concerns associated with such powerful AI agents. I agree with this article that this is definitely going to be interesting—so let’s see what the guys think.

  Mike Collins:
  Yes, you can create custom software that does versions of this now for you—you can write code that, in a very deterministic way, goes and makes you a dinner reservation. But I think we’re talking about models and model systems and platforms that are going to do this at scale.

  I am not going to sit here and tell you what that future looks like in granular detail, but I think we can all let our imaginations go with that. Obviously, there are implications in areas like food, entertainment, and transportation.

  But there are huge applications also in business, where you’re going to have independent agents doing customer service—more than just answering question chatbot-ish stuff—but really fixing customer problems. There’s a lot of the business stack that involves tasks that take repetitive work.

  A lot of white-collar jobs are sitting in front of a screen, moving things around within software systems, for example.

  Naren Ramaswamy:
  Repetitive labor tasks.

  Mike Collins:
  Repetitive labor—take accounts payable at a typical company. It’s getting in people’s expenses, filling… even if it’s software-driven, there’s still a lot of manual processes there.

  Naren Ramaswamy:
  Mm-hmm.

  Mike Collins:
  Getting that into your general ledger—all of those tasks could possibly be done, at least at a basic level, with agents. Some of the bots that people are using and generating themselves to do tasks are forms of that. They might do one thing—input here, output there.

  Being able to string these things together with checkpoints, I think, is what’s creating a lot of buzz. Those things are coming, and they’re right around the corner. Which means people are already doing them now in beta—it’s just in a controlled environment.

  Naren Ramaswamy:
  Yeah. Anthropic just announced their computer use model, which—if you look at it—it’s bare bones. It takes a screenshot of your screen at any given point. It can understand that, right? It can understand the text, parse through the images, and then it finds the button where it says “Next,” fills out all your information securely, and eventually goes through the whole flow.

  You can just say in natural language, “Book me a flight from JFK to London,” and it’ll take you all the way to the end—to the checkout page—and say, “Look, I’ve looked at all the websites, this is the best price. This fits your schedule. Here it is.”

  Mike Collins:
  “Do you want me to book it?”

  Naren Ramaswamy:
  “Do you want me to book it?” And so we’ve gone from thinking software was static…

  Mike Collins:
  Yeah.

  Naren Ramaswamy:
  Now it’s basically having an intern on demand for everyone.

  Mike Collins:
  Yeah. And then if you can do it with a keyboard, you can do it with voice, right?

  Naren Ramaswamy:
  Exactly.

  Mike Collins:
  Because we’re seeing that these things are voice-native already. And I know a lot of people are talking to AI bots on their way home or on their way to work—asking, “What are my appointments today?” And it’s a dialogue.

  So I do think it’s interesting. I think a lot of this is going to be voice-activated. Many of us are wondering: do apps fade to the background?

  Naren Ramaswamy:
  Yeah.

  Mike Collins:
  And it’s really just talking to a piece of glass to do a lot of the stuff you’re doing today. And then the glass—if you want to watch a video or do other things—it’ll obviously be there. Or if you want to dig into the app for whatever reason, you can. But you won’t have to because you’ll be able to just talk to these systems in the same way you can currently talk to your speaker and get it to play any song you want or any series or playlist.

  So just think about that applied to everything in your life—it’s coming very soon.

  Naren Ramaswamy:
  We just invested in a company called Wispr Flow alongside NEA. They’re trying to be that voice interface for all our devices, and they had an explosive first day as Product Number 1, Product of the Day. And yeah, it’s a phenomenal product, and I encourage our listeners to go test it out. I may be biased, but it’s really a game changer.

  Mike Collins:
  Super cool. Yeah, people should check it out.

  Samantha Herrick:
  All right, we’re going to take a second for an ad, and we’ll be right back. Hang tight.

  Speaker 4:
  Exceptional value creation comes from solving hard things. Alumni Ventures’ Deep Tech Fund is a portfolio of 20 to 30 ventures run by exceptional teams who are tackling huge opportunities in AI, space, energy, transportation, cybersecurity, and more. These game-changing ventures have strong lead venture investors and practical approaches to creating shareholder value. If you are interested in investing in the future of deep tech, visit av.vc/deeptech to learn more.

  Mike Collins:
  What’s your third one, Naren?

  Naren Ramaswamy:
  The third one on our list is about search engines.

  Mike Collins:
  Mm-hmm.

  Naren Ramaswamy:
  We’ve talked about Google’s monopoly and all the antitrust claims against them. I think from a technology standpoint now, Meta just announced that they’re developing their own search engine. If you go into Instagram today, they have that search bar now. And what was staggering for me is that Meta said over 180 million daily active users are already using Meta’s search engine.

  Why does Meta need to build one? It’s not to compete with Google as the first goal. The first goal is for them to be able to index all the data they have about you across all the platforms and make it accessible to you. It’s a means to an end, really. I find it fascinating that now any company that has a ton of data will create that little search engine on top of it, and that’s going to be the future for you to interact with Instagram.

  You won’t have to click, scroll, or look for something in a very manual way. Maybe you can speak into it like we talked about or just use the search bar, and it intelligently scrapes through the database to find exactly what you need.

  This is going to eat into Google’s old ads business, but Google is doing some really interesting things with NotebookLM and Gemini as well. So we’ll see how it plays out.

  Mike Collins:
  I think I’ve said on this show maybe two months ago that “search is dead.” And that was somewhat provocative, right?

  Naren Ramaswamy:
  Yeah.

  Mike Collins:
  So on one hand you say, “Wow, one of the great business models of all time looks like it’s going to be disrupted.” Not overnight, but I’ll just speak personally—I use Google Search way less than I did six months ago. I just want the answer, right?

  Naren Ramaswamy:
  Yeah.

  Mike Collins:
  Apple Intelligence, which is limited right now—again, do not underestimate Apple in the medium and long run—but very modest beginnings of Apple Intelligence. Searching photos—which we all have a boatload of on our iPhones—just got way better and easier with Apple Intelligence.

  For example, I did a search: “Show me all of the pictures of my kids in their Halloween costumes.” Done. And there’s just going to be more and more of that.

  You hear a particular term over and over again that you don’t know what it means—you used to Google that. You don’t Google that anymore. Even if you’re using Google products, you’re using Gemini, Perplexity, Claude, or OpenAI to get the answer to that question.

  That’s very bad for Google. The flip side is Google’s got a lot of other really cool stuff. They just made a big drug discovery—Nobel Prizes and this, that, and the other thing. It’s a really tricky transition when you have the golden goose under fire.

  How can I maintain it, milk it, twist it? One of the great business models of all time—trillions of dollars of value. How do I not lose that too fast while my other stuff comes online?

  Naren Ramaswamy:
  Innovator’s Dilemma.

  Mike Collins:
  Yeah, Innovator’s Dilemma. And you’re playing both sides of that equation. Again, way above my pay grade, but smart people at Google are figuring that out.

  Samantha Herrick:
  All right, so I wanted to hop onto something here that Mike and Naren both commented on that I had never heard of before—the Innovator’s Dilemma.

  This is a concept and a book written by Clayton Christensen, an influential book that describes a paradox successful companies face when dealing with disruptive innovation.

  Here are the core concepts and elements of this phenomenon:

  • Focused on current customers: Successful companies tend to prioritize their existing customers’ needs, which can blind them to emerging markets and technologies.

  • Sustaining vs. disruptive innovation: Companies excel at improving existing products but struggle with creating groundbreaking new products.

  • S-curve of innovation: The value of innovation follows an S-curve with rapid improvements in the middle stages, but diminishing returns at the beginning and end.

  • Disincentive to innovate: Market leaders hesitate to innovate because new products might cannibalize their existing successful products.

  Challenges for established companies include resource allocation (based on current demands, potentially missing new opportunities), small markets (disruptive innovations often start small), uncertain future potential (making it risky), and organizational inertia (processes and values hindering adaptation).

  The Innovator’s Dilemma highlights the challenges that even well-managed companies face in staying competitive in rapidly changing markets. It emphasizes the need for organizations to balance focus on current successful products with investments in potentially disruptive innovations that may shape the future of their industries.

  Naren Ramaswamy:
  One thing to touch on is how Gen Z interacts with technology differently than Millennials or Gen X. Gen Z is spending most of their time on platforms like Instagram, and Facebook’s parent company Meta has access to that data.

  They understand culture better than outlets like The New York Times or Wall Street Journal. Suddenly, what if these companies—because of the data they have—become the news outlets for culture for this growing group of young individuals around the globe?

  I think it’s beyond technology; it goes into news and media. We’ve already seen the shift from print to digital, but AI is only going to accelerate that further.

  Mike Collins:
  Yeah. And that’s just a trend that’s happened time and time again. Going way back, it was radio to television. Every generation has its own music, its own news, its own technology base. Those cycles are just getting shorter and shorter.

  It’s very tricky to figure out where the puck is going. Many politicians—the people who win are often the ones who figure out the new media and the new news.

  In this election cycle, I think it’s a story about X versus TikTok. It’s a story of podcasting being incredibly powerful.

  And in business, you’re always thinking about how the next generation is going to engage with you and your company. We talk about it here at Alumni Ventures—our customer is boomerish for the most part at the moment. But there’s $20 trillion of wealth that’s going to be passed down from boomers to the next generation, and then you’ve even got the generation after that.

  How do we begin to speak to them? How do we get them educated? How do they want to interact? How do they want to be venture capital investors? It’s not the same way.

  You can view that as a challenge, but I think great leaders view that as a fun challenge. It keeps you young to figure this stuff out—to meet people where they are and where they’re going.

  Naren Ramaswamy:
  Yeah, absolutely.

  Mike Collins:
  Great conversation again, Naren. We’ll do it again next week.

  Naren Ramaswamy:
  Sounds good. Thanks, Mike.

  Samantha Herrick:
  Thanks again for tuning into The Tech Optimist. If you enjoyed this episode, we’d really appreciate it if you’d give us a rating on whichever podcast app you’re using. And remember to subscribe to keep up with each episode. The Tech Optimist welcomes any questions, comments, or segment suggestions. Please email us at [email protected] with any of those, and be sure to visit our website at av.vc.

  As always, keep building.