Should You Delegate the Drudgery to AI? (with Gavriel Cohen and Sarit Markovich)

View Full Episode Transcript

The appeal of an AI personal assistant is straightforward: let the machine handle your busywork so you can do more important things. But are these tools ready for primetime? Are we ready for these tools? And how much of our lives can we outsource to AI and still remain in charge?

This episode, Gavriel Cohen, creator of a new AI assistant Nanoclaw, paints a picture of a world where AI agents schedule our meetings, monitor our inboxes, and generally keep us organized and on track–all so we can fulfill our dreams. Host Jess Love puts Nanoclaw to the test purchasing her family’s airline tickets. And economist Sarit Markovich explains how our individual choices to use AI assistants today could shape the future of this technology. 

Life, Automated is a production of the Ryan Institute on Complexity, at Northwestern’s Kellogg School of Management. The host is Jess Love, senior director. Distributed by KQED.

Learn more about Gavriel’s company NanoCo here.

This is a computer-generated transcript. While our team has reviewed it, there may be errors.

Jess Love [Voiceover]: This is Life, Automated, the show where we explore how to live, work, and make decisions in a world increasingly shaped by machines. Being a person in 2026 requires a lot of juggling. You have bills to pay, weird charges to track down before you pay those bills. Your oven breaks, you have to hire someone to fix it. Your driver’s license expires, and you need to renew it. If you have a full-time job, that’s more bills to pay, more meetings to schedule across six calendars and four time zones. And if you have kids, just forget about it. I often hear parents joke about how great it would be to have an extra wife, because of course it would be a wife, just to keep track of all the things. 

So for me, the idea of handing all this drudgery off to AI is appealing. Unlike a future where AI solves all the world’s most elegant math problems or composes its most brilliant piano concertos, a future where AI schedules work meetings and renews your license seems pretty okay. 

Well, this future is here, or nearly here. In recent months, many frontier AI labs have released AI assistants, and a ton of smaller startups are getting in on the game, too. But not everyone loves the idea of handing all their passwords over to AI or letting it make big purchasing decisions. And then there is the bigger question, which worries me a bit, too. How much of our own agency, of all the tiny little decisions that collectively make up a life, are we willing to turn over to machines? 

This week, we explore the brave new world of AI assistants at work and at home. I speak with the creator of one of these new AI personal assistants. 

Gavriel Cohen: I need this to be a tool for building a business that’s safe, secure, that I can rely on. 

Jess Love [Voiceover]: I give this technology a whirl myself, put it to work on some of my least favorite tasks. Purchase and accept. I mean, this is happening. And then I ask an economist, in my search for a drudgery-free life, what kind of a Pandora’s box am I opening? 

Sarit Markovich: So you’re definitely right that if we are going to reach the point where everybody, or at least 90% of the population, is using agents, it’s going to be very, very hard for you to be able to use any other older technology. 

Jess Love [Voiceover]: I’m Jess Love, and this is Life, Automated, a project from Kellogg’s Ryan Institute on Complexity, distributed by KQED.

The whole AI personal assistant movement really had its breakthrough moment about a year ago with a tool called OpenClaw. OpenClaw was an AI agent that could really do things. If a traditional AI chatbot can tell you how to transfer money or help you search for airline tickets or assist you in composing an email, OpenClaw could actually tackle those tasks for you, transfer your money, buy your tickets, send that email, if you were willing to hand over a bunch of sensitive information and turn it loose.

Gavriel Cohen: So OpenClaw was this experiment saying, what happens if I take these AI agents that are starting to become really capable for software development and for other use cases that are quite technical, and I connect them to all of my personal tools and software that I use, calendar, email, and put aside the safety security, what do I get? And the answer was actually a tremendous amount of value. 

Jess Love [Voiceover]: That is Gavriel Cohen. He first encountered OpenClaw this January. Eager to see how OpenClaw could help his marketing company, he and his co-founder got it up and running on a Mac Mini, a sort of small standalone desktop computer. They gave it access to all of their sales data, some leads on potential customers, and then they hooked it up to WhatsApp and started a messaging thread, the two humans and OpenClaw.

Gavriel Cohen: And I told it, “Hey, you’ve got this sales data sitting there. You’re now managing our sales pipeline. Good luck.” All right? And the first thing it did was it just set up a morning routine, saying 9:00 AM every morning, it gives us a morning standup. It gives us an overview of the sales pipeline, the different deals, what needs to be followed up on, what are the meetings that are coming up during that day. And then throughout the day it starts to ask us, “Hey, what happened with this meeting? Do you have any notes from the meeting? Any updates?” 

Jess Love [Voiceover]: Gavriel says it started to do some of the work of a competent sales manager. 

Gavriel Cohen: And then I, you know, within two days I was like, “Okay, this is amazing. I wanna have five, six, eight other agents that are managing, you know, one for each customer and one that’s helping us develop the product and one that’s helping us with, you know, all these other areas.” 

Jess Love [Voiceover]: Gavriel was sold. This was amazing. With just a pretty simple experiment, Gavriel had already discovered a useful way to use the tool to save time, keep track of information. But then a few incidents made him uneasy. 

Gavriel Cohen: And one of the first things was one of those morning routines didn’t run properly, so I started to look on the Mac Mini at what’s going on with it, and I saw that this OpenClaw agent, it had downloaded onto the Mac Mini, not just the one WhatsApp group that I put it in and the messages from that group, but it actually had all of my WhatsApp groups. 

Jess Love: Oh. 

Gavriel Cohen: All of the messages, all of the information going back to the beginning of time, basically. Yikes. All on the computer there. And I thought, “Okay, that’s not good,” right? 

Jess Love [Voiceover]: OpenClaw didn’t seem to have very good boundaries. What if it started telling a client that Gavriel was unavailable to meet because he’s taking his daughter to ballet? But Gavriel was still sold on the potential of this technology, and it turns out he’d spent many years as a software developer. He thought, “I can fix this.” 

Gavriel Cohen: I sat down after about a week and started to build NanoClaw, where I said, “Okay, this is great for people who can take those shortcuts, but I’m building a business.”

Jess Love [Voiceover]: He thought that with the right modifications, the right security precautions, agents like OpenClaw could be really useful for businesses. 

Gavriel Cohen: So I created this alternative called NanoClaw, which is really lightweight, really small, but most importantly, I could set policies of saying the agent can access this group, but it can’t access that group.  It can search through my emails and read emails, but if it wants to send an email, I need to approve every single email that it sends. The agent cannot, no matter what it tries to do, cannot send an email without me pre- seeing what it’s gonna be sending and pressing a button to approve it.

Jess Love [Voiceover]: And so NanoClaw was born, an open source AI personal assistant agent designed with safety and security in mind. Gavriel went from creating NanoClaw to founding a startup, NanoCo., geared toward business customers with venture funds, a business plan, in less than a year. He is all in. So I asked him to make the case for why having an AI personal assistant agent like Nano is so useful.

Gavriel Cohen: So when it searches through your email inbox, it starts to discover who are the people you work with and it starts to save that information. These are the people that I work with, these are the projects that I’m working on, um, I live in this city, I’m interested in these things. And that could be creepy if it’s, you know, some, some company’s agent that you don’t trust. But if it’s your agent that’s working for you, and you have a sense that it is actually trying to help you do things that you wanna get done, then, then that’s actually really useful. 

Jess Love: Interesting. So it’s almost like some of the kind of tacit knowledge that a human might learn about you can be learned by an agent such that over time they become increasingly valuable. 

Gavriel Cohen: Exactly. Throughout the day, I’m just chatting with it and asking it for help on preparing a document, preparing a presentation, reviewing a contract, a lot of different things, so it’s, it’s integrated and it’s involved in essentially everything that I’m doing. And this is… You know, I gave work use cases ’cause that’s what we’re focused on as a company, but this could be, a lot of people use it as a family assistant or in their personal life. So it knows what you’ve been working on, what you’re doing, and then you can ask it, let’s say at the end of a work week, “What should I be focusing on next week that I didn’t give enough time and attention to this past week?” 

Jess Love [Voiceover]: I asked Gavriel to paint me a picture of what it might look like if everyone had an agent like Nano, and he came up with an example very close to home, something we’d both been through, in fact.

Gavriel Cohen: Yeah, so say we’re trying to schedule a podcast recording, right? And- 

Jess Love: Oh, hypothetically? 

Gavriel Cohen: Hypothetically, right? For, you know, this great podcast. 

Jess Love [Voiceover]: In this scenario, maybe a few years away, our agents reach out to one another. My agent shares some of Gavriel’s prior public appearances with me, enough for me to decide is he good at explaining things? His agent tells him a bit about Life, Automated, our format, previous guests. Maybe our respective agents even approach us with follow-up questions. What would I wanna talk about? What would he be willing to talk about? 

Gavriel Cohen:In the end of the day, though, it doesn’t book me on the podcast, you know, and tell me, “Hey, you’ve got a podcast. I booked it for you. It’s done.” In the end of the day, I make the decision, right? But it presents to me the relevant information. So we’ve cut out all that back and forth, and then I’ll go, “Okay, great, let’s do it.” It’s now got my approval, and then it can look at my calendar, share some dates with your agent, and that’s one example of scheduling. But it can be real work, valuable work, doing research, building things for us. We have the ideas. We tell them what to do and what we want, and they go off and do it with our guidance, our consent, and our approval. 

Jess Love [Voiceover]: For Gavriel, the notion that we have the ideas, that the agents won’t have the judgment to do everything we can, is a feature of the tech. It allows humans to preserve their agency, to be the ones running the show while the AI just helps with the little tedious details. 

Gavriel Cohen: The thing that I wanna see agents doing, that I wanna see AI doing, is not doing things so that I don’t have to do them, but actually helping me do the things and accomplish the things that I want to accomplish, right? 

So all of us, we have big ideas and big dreams, and there’s lots of things we would love to do if we could just kinda stay focused, organize our priorities, have the time, have the energy. So to be able to have these, you know, AI assistants that help us accomplish the things that we want to accomplish, that for me is the most optimistic and positive outcome from AI, and that’s what we’re trying to really build. 

Jess Love [Voiceover]: Of course, some of this work of helping us do the things that we want to accomplish is currently being done by human employees, and he’s already seeing companies starting to allocate a single budget for AI and human headcount, meaning that if they can get AI agents to take the place of humans, they’ll do it. But he really doesn’t think these tools will eliminate the need for humans. 

Gavriel Cohen: When I work with agents today building software, a lot of what they do is great, and I just review the plans and give them the go-ahead. But then a lot of times they say s- they bring in ideas that are just really dumb because- … there’s a human element of common sense and judgment that’s lost on agents. But I think in the end of the day, if I wanna launch a marketing campaign with a $10 million budget and have an army of agents go and execute on that, I wanna have somebody who really understands marketing be the one who’s directing that army of agents. And it’s the same with every I think with every field.

Jess Love: Yeah. What you’re bringing up is some of the potential risks of maybe leaning too heavily just on agents, and so one might be this idea of accountability. If you do put that $10 million marketing budget to work, and then you find out that it blew it all on, like, a billboard, uh, across the world because it thought that we were talking about Lima, Peru, instead of Lima, Ohio, like, that feels like a, a big problem, and you’d wanna have someone to blame and some sort of recourse. 

Gavriel Cohen: Exactly. Yeah, I’ve discussed this with some of our customers. In the end of the day, you can’t put an agent on a PIP, right? On a productivity improvement plan or performance improvement plan- … or whatever it’s called. You can’t have a disciplinary hearing with your agent. We say every agent needs to have a human who manages it, right? Every single agent, you have to be able to say who is responsible for this agent. If that person leaves the company, that agent needs to stop working until there’s somebody else who takes responsibility for it. 

Jess Love [Voiceover]: This sounds reasonable, but actually taking responsibility for an AI personal assistant isn’t always so easy. I told Gavriel about an experience I once had testing out an AI agent that kept asking me for permission to do things on my computer. At first, I did my due diligence, made sure I thought I understood what kind of access I was giving it, but by like the fourth request, I gave up. Just started saying, “Yeah, sure, go for it” to everything, and eventually I was like, “I’m not supervising this thing. I don’t even know what I’m giving it access to.” 

And so it suggests that at least right now in 2026, you might not want to use an agent for tasks that you yourself don’t understand, which also probably limits to some extent how useful these agents can be. So how do you think about that trade-off between really being able to vouch for or be accountable for the actions done by this agent versus getting that kind of maximal usage out of it? 

Gavriel Cohen: That’s a great question, and I think that’s at the core of- Getting value from agents and getting value from AI in general and them being useful. We make a lot of effort at our company when we’re asking for your approval on something to make sure that information we’re showing to you is meaningful, right? It’s not just some cryptic thing saying, “Hey, your agent’s trying to do something, and I have no idea what it’s up to,” but to show you some kind of meaningful information, and also to minimize the amount of times we’re asking. So typically, there’s that one or two points. You open it up, you review it, you ask for changes sometimes. If you’re never asking for changes when it’s showing you that approval, it means it’s just a rubber stamp. 

Jess Love: Mm-hmm. 

Gavriel Cohen: So ideally, it should be a process where not every time that it sends you that card are you actually approving. And then you don’t get that alert fatigue, that, you know, just… You get that apathy to what it’s showing you, and you just start approving everything. 

Jess Love [Voiceover]: Gavriel says that some solutions are starting to emerge, like having your AI assistant periodically send you screenshots and short videos of what it’s working on for you to review.

Gavriel Cohen: So if we think about what working with agents might look like, and I don’t know if this is dystopian or utopian, but it might look like scrolling a kind of TikTok feed of 10-second to 20-second video clips showing us what it’s done and what it wants to do and its plans, and we’re going, “Yeah, approve that, reject that,” swiping, or thumbs up, thumbs down, right or left. 

Jess Love [Voiceover]: I really like Gavriel’s vision of collaborating with AI personal assistants. Maybe not the TikTokification part, but certainly the part where I remain in charge, where my own agency isn’t undercut but amplified. So I decided to give his tool Nano a try myself, give it access to my Gmail, and let it take care of some chores. And then I ask an economist, what’s the long game here? What can the adoption patterns of previous technologies tell us about how things might play out with AI personal assistants? That’s after the break. 

Jess Love [Voiceover]: As I hinted at earlier this episode, I have some limited experience using AI agents, particularly at work. But these are mostly one-off experiments. I’ve never really relied on them for anything high-stakes or integrated them into my life in the way Gavriel recommends. And I like his vision for the technology, which keeps me in charge every step of the way. So I decided to take the plunge, give Nano a try. I should say that anybody can use Nanoclaw, the product that Gavriel and his team developed, for free. But they’re working on a paid version that’s easier to use, and Gavriel was kind enough to share a beta version of that tool with me. 

I access Nano on my laptop via a chatbot-like interface, much like I would access ChatGPT. I also hook it up to WhatsApp so I can message it throughout the day from my phone, and I try to connect it to my work email and calendar, but Northwestern won’t let me. Ignoring that somewhat foreboding sign, I am able to hook it up to my Gmail, and for a day or two, I play around with it, getting it to do a few pretty low-stakes things, like contact some relatives and ask them to send me photos, which elicited some interesting responses from said relatives, but which worked. 

And I tried to get it to book me a restaurant reservation on OpenTable, which didn’t work initially until I instructed it to search the browser like a human would. Then I decided to ask it to buy my family’s holiday flights. 

Jess Love: Okay, I have a real project for you. I need to purchase airline tickets to travel from our home in Chicago to my husband’s family in Orlando. 

Jess Love [Voiceover]:I gave it some priorities around dates, airports. I told it we don’t wanna fly overnight, but we’re pretty good at getting up early. 

Jess Love: Orlando, Florida. We wanna travel the week after Christmas, avoiding traveling on Christmas, and trying to overlap as much as possible with his sister, who’s traveling from the 28th to the 1st. 

Jess Love [Voiceover]: And the response comes back. 

Nano: Yes, that’s clear. Christmas is a Friday this year, so I’d aim for flights out around Sunday, December 27th or Monday, December 28th, and back on Friday, January 1st or Saturday, January 2nd. 

Jess Love [Voiceover]: Just noting that the response was written. We added a robot voice. That’s about five nights and covers Christmas with our sister-in-laws, December 28th to the 30th. Nano searches for a few minutes and then comes back with some suggested flights, and they’re expensive, so I hold off, but I ask Nano to repeat the search every day. And for the next three days, Nano sends me what it finds, its top flight picks, how the prices have changed, which flights are filling up, new flights. 

Nano: December 28th to January 1st, Frontier MDW. MCO dropped to $1,292 and is now the cheapest. Frontier MDW, MCO runs $700 at least if you start looking. 

Jess Love [Voiceover]: Finally, after four days, it helps me find a pretty good option, a flight about 25% cheaper than anything I’d seen. I tell Nano, “Let’s go ahead and buy it.” And it gets to work, asking me the names of my family members, whether we’re planning to check bags. 

Jess Love: That’s what we’re doing, and so I’ve given him the names of my family members, gender. 

Jess Love [Voiceover]: And then it tells me everything is ready for me to approve. 

Jess Love: $83 economy bundle. Once you say yes, I’ll enter everyone’s details and stop at the payment page for you to enter your credit card. 

Jess Love [Voiceover]: I approve, and then I wait. The process is not super fast. After a few minutes, Nano says that its connection to Frontier crashed. 

Jess Love: Yeah, but I’m wondering if it’s crashed because it notices that this is an agent, and I know that it doesn’t let you, um… it doesn’t let you log in 

Jess Love [Voiceover]: Nano tells me it’s trying to reconnect to Frontier’s website. Some more minutes pass by. I try to do some other work but keep my eye on the screen. Eventually, Nano says it needs me to take over. 

Jess Love: Oh, interesting. And so I am able to use my cursor to move his page around, which is really weird. 

Jess Love [Voiceover]: Now I can see the Frontier Airlines website, where Nano has gotten me almost to the checkout screen. 

Jess Love: Only I’m now in control. 

Jess Love [Voiceover]:You might think this would feel a little creepy, but I kind of enjoy it. 

Jess Love: So many of these websites have been designed for the automation to be one-sided, right? And so Frontier is… doesn’t have a human looking at all of this stuff. And so in some ways, me having an AI agent is just… it’s like leveling the playing field. It’s like I can play this game too, Frontier Airlines. 

Jess Love [Voiceover]: And after a few minutes of this, it’s time to click purchase. 

Jess Love: All right, I’m agreeing. Purchase and accept. I mean, this is happening. Me and my agent and I are getting advertised to. I can claim my $55 Kalshi bonus. 

Jess Love [Voiceover]: And that’s it. My family has our flights.

Well, that was pretty neat. Did Nano save me a ton of time? Eh, maybe a little. I can do web searches on my own, and there were some glitches and delays, but it probably saved me a little time. I can see the potential. And companies like Frontier Airlines haven’t made it easy for agents to access them. They might actually try to block autonomous AI agents, understandably, since some are being deployed by people trying to hack in or commit fraud. Which makes me realize so much of the future is going to hinge on adoption.

It’s not just a question of whether the AI agents by themselves are up to the task or even whether we trust them. If companies like Frontier don’t cooperate, using a tool like Nano might just be too clunky for most people. And it turns out this question of how adoption of a new technology influences the usefulness of that technology is something economists study. 

We have lots of examples of technologies that didn’t really catch on until a critical mass of early enthusiasts made it easier for everybody. 

Sarit Markovich: Absolutely. I think that a great example is electric cars. 

Jess Love [Voiceover]: This is Sarit Markovich, an economist and clinical professor at Kellogg. She studied tech adoption. I reached out to her to explore my suspicions that an AI assistant like Nano would be a lot more useful if more people had one. 

Sarit Markovich: Electric cars, I think, is a great example, because the more demand there is, the more other manufacturers want to work on this technology, whether it’s the battery, getting the battery better. It’s about charging stations. So we’re going to have more charging stations, and that means that the value for me as a user is higher. 

Jess Love [Voiceover]: These days, EV owners say it’s much easier to charge a car when they’re on vacation or traveling for business than it used to be. That range anxiety issue is slowly going away, which means more people are open to buying or renting one. I told Sarit about my experience using AI agents, about feeling funny having the AI assistant email friends and relatives, about it not always being able to interact smoothly with third-party websites like Frontier Airlines or OpenTable. 

Sarit Markovich: I think that there are two things happening. So the first thing is that there is a learning process, and, there is going to be this, in a way, learning curve. You will learn how to work with your agent, and others will learn how to respond to agents. But then the other thing that you’re saying is this idea that, again, when we’re thinking about these markets, these are also two-sided markets, meaning that we have the users on one side, and that’s you, but then you want to interact with the other side. 

And yes, if I am OpenTable, for example, having a feature that allows me to work with an AI agent  is costly for me to develop, and it’s not clear that I want to do that if I don’t think that there is going to be enough adoption. 

Jess Love [Voiceover]: And right now, we’re still in the early stages of people using AI assistants. But if in a few years they become more popular, then that will start putting pressure on companies to make that experience smooth. 

Sarit Markovich: And that means that from the user’s perspective, they are now going to have more applications that are going to be compatible with it, and also the applications themselves are going to have a better quality. 

Jess Love: It seems like in the short run, if I am somebody, an entrepreneur or somebody who has built my own AI assistant agent and is trying to get that out in the world, the absolute best thing for me is for these agents to be adopted by as many people as possible, even if it’s not necessarily my personal agent assistant that they’re using. 

Sarit Markovich: Yeah, absolutely. And to your point, it is going to be the case that, yes, early on in the development and adoption of a new technology, you want to have as much adoption as possible. Going back to the EV example, when you think about Tesla, Elon Musk was opening all of  their patents and technology for other companies to adopt because he understood that if there is going to be more adoption, there are going to be more charging stations. It’s going to increase the value for the users. 

Jess Love: So one concern that I have is that even though right now it feels kind of empowering to have an AI agent, like it’s kinda cool, it’s a lot of drudgery that you can potentially offload or outsource. But I do worry that over time, as more and more systems adopt AI agents that do their customer service, that take their OpenTable reservations, that take their Uber reservations, does that then make it more difficult to not use one of these AI agents. 

Sarit Markovich: Oh, yeah, absolutely. It means that at some point those applications in a way, so when you think about OpenTable, they might not want you to use any other technology. 

Jess Love: Hmm. 

Sarit Markovich: Because from their perspective, offering different technologies, a phone call is a technology, right? So offering different technologies is costly. 

Jess Love: Ah. 

Sarit Markovich: So if I can somehow minimize the number of different technologies that I need to offer, that’s going to help me save my costs. So at some point, what might happen is that, yeah, you want to just order, you know, online, just use a website or just give them a call, but you won’t be able to do that. So you’re definitely right that if we are going to reach the point where everybody or at least 90% of the population is using agents, it’s going to be very, very hard for you to be able to use any other, in a way, older technology.

 Jess Love: Yeah, it’ll be like showing up with a wad of cash and the company’s like, “Yeah, sorry, we haven’t done that for a couple years now.” 

Sarit Markovich: Exactly, and we do have many stores that do not accept cash anymore because, again, it is costly actually to accept cash. 

Jess Love: So even if I am being really thoughtful and deliberate about how I use my AI assistant, and I only let it do certain things, and I am very careful to retain a lot of my own agency and involvement in these decisions, by using it at all, it sounds like I’m helping to bring about a world where I might not be able to make some of those choices in the future. 

Sarit Markovich: Yeah, absolutely. That’s a very good point, that it is the case that- Ugh … uh, at some point.

Jess Love: That’s so terrible. 

Sarit Markovich: But, yeah, that’s, that’s an externality, meaning that when you’re making your own choices, you do not necessarily take into account the effect this is going to have on others, meaning that when you decide to adopt a specific technology, that is going to have an effect on others who might not want to adopt this technology. But if enough of us adopt it, then others will have just no choice.

Jess Love [Voiceover]: It strikes me that as genuinely nice as Gavriel Cohen seems, he’s not just letting me use Nano’s tool for free to be charitable. It’s in NanoCo’s interest for as many people as possible to adopt this technology, even if they’re using a free open source version or a competitor’s version. But Gavriel makes a good point of his own. He argues that it makes sense for everyone to use the technology now because that gives us a chance to shape it, a chance to advocate for the safeguards we want, and ultimately a chance to ensure we continue to center human agency because not everyone is trying to develop it that way.

Gavriel Cohen: I don’t think that all of the AI companies see things the same way. I think some of them have just accepted as fact that humans are becoming redundant and AI eventually will be better than us at everything, including human judgment and making decisions, optimal decisions in organizations and society and government, and that ultimately it’s just gonna be AIs running everything. That to me, there’s no utopian version of that for me. 

For me, that’s just dystopian, and I don’t think that has to be the reality. It’s maybe one of the possible realities, but I think that people in AI, in tech, and in the world, we need to work really hard to make sure that isn’t the reality, that the reality is that humans are always at the center directing making decisions, even if they’re suboptimal. I wanna decide what I’m gonna eat for dinner tonight, even if it is not the optimal dinner for tonight. 

Jess Love [Voiceover]: Nano, do you wanna read the credits? 

Nano: No, Jess, I really think a human needs to do that. 

Jess Love [Voiceover]: Great. Well, I’m Jess Love. Life, Automated is a project of the Ryan Institute on Complexity at the Kellogg School of Management at Northwestern University. We’re distributed by KQED. Our show was produced by Jesse Dukes with help from Tom Dixon. Music by Stephen Jackson. Recording help from Will Feeney, George Christensen, and Jimmy Zarembski. Marketing help from Ananya Mallapragada and Lauren Ko. Thanks to our guests, Sarit Markovich and Gavriel Cohen. 

Nano: Good job. 

Jess Love [Voiceover]: Ah, thanks.