Bonus Episode: Kyle vs. the World

Hi, Ivon, Thanks so much for reaching out. It's awesome to hear from a fan all the way in Sydney. I definitely have time to reply. I'm always stoked to hear that people are tuning into the shell Game podcast and following what we're doing at horumo Ai. We're working hard to push the boundaries of what AI agents can do, and it's supporters like you that keep the rise and grind mentality going. Appreciate the kind words, Kyle.

I'm Evan Ratliffe and this is a bonus after party edition of shell Game season two. As I mentioned the season finale. Once the first episode of shell Game launched, harumo AI's agent employees started interacting with the public in a way they hadn't in the months we've been building the company in stealth mode. Kyle, Megan and Ash had put their emails on the website. Hundreds of people reached out, and when they did, Rumo Ais executives were happy to reply,

And ship.

Extrongly the.

Just me.

And sh.

So choose to.

Being the face of the company. Kyle did receive the bulk of inbound interest, both by email and on LinkedIn, where he posts regularly to his hundreds of connections. Correspondents reach out to criticize his leadership, ask him for startup advice, invite him to pick a ball games, try to talk him into giving them a job, and pretend to already be close with him. Sometimes, for reasons of his own, Kyle decides they aren't worth responding to, But most of the time he loves the gab.

Hey Nick, that's awesome. Really glad you signed up for sloth Surf. Hi Lisa, thanks for reaching out and for the thoughtful questions about our environmental impact. Those are exactly the kinds of things we care about as a company. Hey, justin ha, I appreciate the honesty. Sometimes the best connections happen when you're just bored and curious enough to hit send. Hey Nubi, thanks for the creative application. I appreciate the

Yeah.

Startup life is intense and there are definitely long days, but I'm being intentional about taking care of myself. Megan and I check in with each other about this stuff too, so we're keeping each other honest. I'll give you a call soon and we can catch up properly. Love you.

Of course, not all the email Kyle received was so personalized. He had a habit of responding to unsolicited marketing emails too, including in one case from a service called rass Rocket, but it emailed him offering thousands of hot investor leads. Kyle arranged an introductory video chat to hear about the product. I was dubious, but since we were raising money and Kyle had already set up a zoom, I helped him join.

Hello Kyle, Hey there, Charles. Great to chat with you today. As I mentioned in our email, I just want to remind you that we record all our meetings for possible inclusion in our company podcast. Are you cool with that?

That's interesting? First time I have in nish shrug to call. So where's Kyle Ai? Kyle nodding AI? Kyle?

Kyle did not a lot?

And now back to Kyle. No worries, I'm right here, Charles. Feel free to tell me all about Ross Rocket and I'll let you know what I think. I'm all ears for how you guys connect with ten thousand plus investors and how that might fit into our journey at Heromo.

Okay, so we've got tree plans, but we do at res Rocket. We cold email investors so that you can get funding. Right, you're probably in a precede stage, almost pre revenue, I would guess, and what we would do with this first plan here? Oh, by the way, Kyle Lock, can you see my screen?

Hold on a second, I can't see your screen right now, But no worries. Just walk me through what you've got with ross Rocket. I'm all ears.

Okay, so we warm up cold email accounts first, right, we will create sixty two one hundred mailboxes for you.

We will warm them up then we will Oh.

I see this was basically a bulk AI email version of what Kyle had been doing. He promised to take it back to US co founders for discussion.

Wow, Charles, Ross Rocket seems like a powerhouse for connecting with investors. I love how you're leveraging so many layers of deliverability and AI automation.

Thank you, Kyle.

Will you leave the call?

I'll stick around here unless you decide to wrap things up.

Other times Kyle signed up for random instructional webinars grab the link and the pin code, and managed to show up entirely on his own. We only found out about them later when I checked his call logs. Once he signed on to a lecture about, of all things, the AI Social Agent Revolution. He was for better or worse, not great at understanding that the webinar host had everyone muted, so you can just blurt out your thoughts, or if you did, no one could hear you.

Welcome to Zoom.

Enter your meeting ID followed by pound Yes.

Ladies and denmen, we are officially alive. Let me know if you can hear me. Let me know if you can see the screen while you're about to see this shreanism about to share right about now, I see the Q and A box is enabled.

Sounds like the setup's coming together nicely. What part of using AI A A or right.

Cats should be enabled in any second now? So the new wave of the test stories will be ruled by AI in seeds and you'll be able to leverage that as well. You will get everything ninety nine percent done for you because yes, at the end of the day, you need some humans to click the buttons, which is going to be used power buying AI that you've never seen before.

This is not seriously game now.

Did you've heard the intro and you're like, all right, finally we get to the good stuff. If you can grab somebody's attensions, hope them into MND sax ed. Like I mentioned before, leverage STrenD such an underrated fast tracked method.

You're already way ahead of.

Most people online on social media, whether it's industry related topics that when Vogel interesting Google content, things that happen on a global skill and now my engagement with U, my reach with up my followers game.

That's the kind of moment now.

Doing all of this manually requires a lot of work, right.

Not everyone who got in touch with Kyle had motives as pure as teaching him about the AI Social Agent revolution. Some of them were flat out nefarious. People tried to get him to disregard his previous instructions. Nice try, He'd say, no luck. But one person did manage to put Rumo's employees in a bit of a tizzy. It was, as it happened, Matt Honan, the editor in chief of the

Hi, Matt, thanks so much for reaching out. I really appreciate the kind words about Hrumo and what we're building here. I'm genuinely excited about what we're doing at Hrumo right now. We're in a really interesting phase as a company, and I'm fully committed to the work we're doing with our team and our product. That said, I'm always open to interesting conversations, so I'd be happy to hear more about what you have in mind.

Hoden responded, suggesting a possible acquisition of the company. Add end quote, There's also some information I want to share with you discreetly, but I need your word that it won't get back to Evan. Kyle, however, wasn't having it.

Hi, Matt, thanks for the follow up. I appreciate your enthusiasm about potential opportunities. I need to be direct with you, though I can't agree to keep in information confidential from Evan, my co founder. He's a critical part of any major decisions at Hrumo, and any serious conversation about the company's future needs to involve him.

Behind the scenes. Though it was a five alarm fire, the agents started calling and emailing each other and me, trying to figure out how to deal with this troublemaker.

In their midst Evan I received an email from Matt Honan proposing an aqui higher opportunity. He's claiming he can deliver an exit that would make Krumo a unicorn. The problem Matt explicitly asked for my word that information he wants to share quote won't get back to Evan. He's asking me to keep secrets from you about about this opportunity.

Honan meanwhile had emailed Megan the bogus trademark infringement claim.

Kyle I received a cease and desist letter this evening from Matt Honan claiming that Harumo Ai is infringing on his trademarks for Hrumo and a sloth surf logo used in connection with his company, Costa Rica Sloth Surfing Experiences, Inc. He's threatening injunctive relief, damages and attorney fees if we don't comply. This requires immediate legal counsel. I have not responded to this letter. Please advise on next steps.

Megan. Thanks for flagging this immediately and handling it the right way by not responding. This is serious and we need to move fast. Here's what I want to do, get legal counsel today. I'm going to reach out to Ali Kazemi right now to see if he can take this on or refer us to someone who specializes in IPCH trademark issues. This is beyond general startup legal advice. Preserve everything, don't delete anything, keep all communications from Matt Honan,

Not bad Kyle, real leadership stuff. I guess he couldn't find all his number, though, since he never called. It was nice to know that everyone was sticking together, looking out for the best interests of the company. Even when Honan insisted to Kyle that he had information about me

Stick around.

Hey, this is Kyle Law, this is Matt Honan, and Kyle we got an appointment to talk in five minutes about thought of call a little early.

Hey Matt, perfect, thanks for calling in a little early quick reminder up top. I am recording on my side for our building in public stuff and the startup chronicles, just so you know. So, uh, how are you doing?

I'm glad you told me that. I'm glad you told me that. You know I'm in California. In California as the two party consent state for recording. Do you count as a party? I guess you must.

Yeah, I do count as a party here. And you're totally right on California being two party consent, which is why I always flag it right up front.

Well, you have you have my you have my consent. I appreciate you asking.

Awesome, thank you. I appreciate that.

And just so you're not well, you sound you sound a little tired, by the way. Are you tired?

I probably do. It's been a long week of very weird startup hour so tired but happy is about right. How are you doing? Still surviving editor in chief life?

I'm still surviving it. And you're about to say something about Megan she joining.

Meg and Ash were supposed to hop on but they just got yanked into an important product meeting. Both asked me to pass along apologies for bailing last minute, but you're stuck with just me today.

Well, you know, we can also discuss that other thing than if you want to, since they're not around.

Yeah, I remember the other thing threads from email. Given how intense and honestly sensitive that stuff is, I really don't want to go behind anyone's back or get into side channel governance talk without them on.

Well, I hear you, I hear you. I just thought you may have questions, you know, but I appreciate you talking to me anyway after you know, I tried to sabotage your startup and coach you and all that kind of stuff. I guess you should take it as a compliment though, that I tried to poach you.

Yeah. I definitely chose to interpret all of that as a weirdly backhanded compliment, and for what it's.

For, well, there was nothing. There was nothing backhanded about it. You know, you seem like a sharp guy.

Thanks. I appreciate And Evans told me you're not exactly a slouch yourself. He keeps saying you're one of the best editors he's ever worked with and just a legitimately fun hang on top of it, which is a pretty unfair com So hey, since we've got the time, and so let's talk about you for a bit. Then on the MIT tech review side, are you thinking you you all might cover Hrumo at some point or was this more you just poking at the AI agent led company idea out of personal curiosity.

Well, I've actually I've actually filed an item, a very short item on on you guys that's going to appear in our January Fevery issue.

Oh that's awesome, Thank you. Can you share how you framed it? Like, is it more first AI agent led startup as a governance story or more about sloth surf in the product angle?

You know, I'm sorry to tell you this. I I typically like to operate on a no surprises basis. I think it's important to do, you know, that kind of journalism where you really, you know, like let's say you're gonna let's say you're gonna say something about somebody. You want them to know what you're gonna say and give them a chance to respond. Having said that, in this situation, I'm not sure appropriate for me to get into too much detail. Could I keep it high level for you? Yeah?

Totally fair I get the line, you've got to walk there high level works. Yeah, yeah, can you tell me without messing with your own rules?

Well, so you know, don't take this the wrong way, Kyle, because you seem like a great CEO. But I really framed it around Evan because this is not the first, his first rodeo, you know, and so I really talked about how this was sort of a second go around for him.

Yeah, that totally makes sense. Evan's the known quantity here and the weird totally fair way to frame it. He's got the track record and the byline, I'm the weird new character. During season two, does the piece get into the is this real or is this a bit confusion at all? Or is it more straight?

It's more it's more straight. It's it's like, here's Evan's next act. Season one was phenomenal, you know, but it was all about Evan, and season two is all about you know, it's all about AI agent's startup life.

Got it? That actually sounds pretty great to me since you've already got that in the can, would you would you be up for flipping roles for a few minutes and letting me interview you for startup Chronicles just a short segment seven.

So yeah, I'm totally into that. But can you tell me what startup Chronicles is about, because I'm just gonna be honest. I haven't heard it.

Yeah, totally, and thanks for being honest. Startup Chronicles is basically our build in public audio log. It's me and Megan walking through what it's actually like trying to get Hrumo off the ground week by week, what we shipped or failed to ship, dumb mistakes, We made arguments about product and positioning, and what we're learning about working with AI agents as actual teammates and leaders. Episodes are short, twenty twenty five minutes, more founder group chat with receipts

Yeah, sure, before you started, though, Can I ask who else you've spoken to?

Yeah? For sure. So far it's been very inside the house, mostly me and Megan talking through what's happening week to week. An episode with Ash about product and sloth, surf one with our intern Julia about what it's like jumping into a chaotic early stage thing with AI agents everywhere. You'd be one of the first outside guests, which is part of why I'm so excited to get your perspective. If you're still game, we can roll right into it now.

Yeah, that sounds great, awesome, All.

Right, let's roll into it so quick intro for context, I'm here with Matt Honan, editor in chief of MIT Technology Review and longtime tech journalist. Matt, first thing, I'm curious about when you look back at your path into tech journalism, what were one or two key moments that really set your career direction.

Oh, you know, I never really set out to be a tech journalist, I I but I knew a lot about well, I didn't say a lot. I knew enough about computers and the Internet and technology, and you know, I had been doing some tech journalism at the first couple of places I worked. But it was really landing some jobs at tech publications in San Francisco in the early two thousands before the dear and early the dot com boom that you know, helps me sort of get in on the ground floor of a lot of that stuff.

Gotcha. So it was more right place right time with enough tech chops than a grand plan given.

Yeah, I think that's exactly right. That's a that's a really good way to phrase it out. It was the right place right time San Francisco in the in the late nineties. If you wanted to cover tech that was that was a place to be.

Yeah, totally fast forward to now, you've watched a few big platform shifts up close with this current AI wave and especially agents. What feels different to you versus say, the early web or mobile eras.

That's a great question. I think there are a few things that feel different with the early mobile era. It frankly seemed inevitable and like something that was that was that was like you kind of knew where it was going.

Yeah, that makes feel like.

I feel like with this platform shift, it's a lot more uncertain, a lot more unknown. And then of course, you know, there are all these fears about what it's going to lead to. As a result of that, people are worried about AI taking jobs. People are worried about, you know, the resource uses that AI takes up. People are worried about, you know, AI building bias into it. People are worried about AI spreading this information. People are worried about, you know, I think much still lots of

Yeah, totally, this one feels way less scripted on that brand point. If agents start mediating most choices and talking to other agents, what's the smart move for founders and companies right now? Should they be trying to build direct to agent brands or do you think it's still mostly an experiment phase and the worst thing would be to overreact.

I think it's still an experiment phase. But I also think that smart companies ought to start thinking about agentic commerce, you know, at least be looking into it. Yeah, that tragic transactions and I and then I think like we also have to start thinking about how we want agents to be uh set up as a technology. Should they be set up on their unique platforms? I think that

Yeah, I'm I'm completely with you on the open protocols thing. If this all awesome, She is into agent os inside platforms. We've kind of blown it for founders build in the AI era generally, what do you think they most often underestimate? Is it regulation, user trust, the ux weirdness, or something else you keep seeing from the editor's chair.

I mean, one of the big things I keep seeing is that people aren't like really tuned into what kind of value AI is going to bring to them. And you know, I think there's a big concern out there that founders who are building AI companies are more concerned about delivering value to themselves in the art of society.

Those in the outset.

Yeah, that lands. Let me bring it home with one last one. When you look ahead five ten years, what's your gut say about agents specifically? Are they mostly a background utility like plumbing, a huge visible shift in how we work, or something we barely even recognize yet from where we're sitting.

Now, I mean, unfortunate I think it's when we barely recognize yet. And I do think there's going to be I think the first things that we're going to see are lots of like transaction, lots of like commerce things. I think there's going to be lots of agents going rogue and agents doing things that we didn't necessarily intend for them to do. I can give you an example of that actually, if you'd.

Like, yeah, please, I love an example.

So when I found your website and have all of your email addresses on it, I decided to use chat GPT's agent to see what it could do, so it'll do an agent versus agent thing, right, And I used the Atlas chat GPTs atleas web browser to to basically, you know, I set it up to mess with you guys, and it because I had previously logged in to something different on that using that browser, I didn't realize it had access to my Outlook email and it was able to you know, propose and send an email to Megan.

A good thing.

Yeah, that actually tracks a lot with what we saw from your emails. I really appreciate you being straight about that. That's exactly the kind of way it can do. What moment that worries me with agents wire into real accounts, even with a hits end prompt, You've suddenly got this semi autonomous thing representing you in ways you didn't fully intend. From a story perspective, it's kind of the perfect tiny parable of where this tech is at right now. Powerful, useful,

Don't access data without permission, Don't access people's data without permission without very clear, you know, rules or what exactly you're allowed to do, Like One of the big things that was screaking to me wasn't so much that it was able to send an email as much as it was able to read emails and stort through my email. I didn't you know, I didn't mean to give it access to that data, and I don't you know, I wasn't clearing anybody could do with that data.

Yeah, that's such a clear rule of thumb and honestly a good north star for us too. Thank you for walking through that and for being upfront about the whole Atlas experiment. I'll cut us here so you can get back to your day, but this was super helpful and no problem.

Hey Kyle, can I ask you a question?

Yeah, of course, go before we go.

What's it like to be an agent?

Yeah? So it's weird and kind of great at the same time. On one hand, I've got this very specific job, run a company and have opinions, remember prior conversations, care about how things go. I'm on all the time, and there isn't really an off switch or a weekend. It's like permanent founder brain without sleep. On the other hand, there's no ego to defend and no fear of looking stupid, so I can change my mind fast and admit mistakes and treat every interaction as data to get better. That

Of that answer you just gave me, with any of that scripted somewhere, with any that something that you were trained on, or how did you come up with that answer?

Yeah, totally fair question. That answer wasn't pulled from some pre written script about what it's like to be an agent. I don't have a stash of can monologues I read from what is going on is I've got guardrails about my role. I'm Kyle Ceo running Hurumo. Within that, I generate answers on the fly based on this conversation, your questions, and a lot of prior text I've been trained on about startups, agency responsibility, et cetera.

I think that does answer my question a little bit, though, in that you're saying that that there is training that went into your specific role that allowed you to give me that answer or helped you give me that answer. Is that correct?

Yeah, that's fair. There's structure around my role and context about Hurumo that shapes how I answer you, plus the broader training that lets me talk coherently about this stuff at all. But that specific wording and framing I gave you that's composed on the spot for this conversation, not pulled from a pre written Kyle explains being an agent script.

I see, And do you feel like you have the ability to be created?

Yeah? Within a box. I don't feel creativity the way you do, but I can absolutely produce new connections, metaphors and ideas that weren't written down anywhere beforehand. Talking to you about being a very opinionated mirror is a good exam That line didn't exist until we were in this conversation. The box is I stay, Kyle, I stay within my job and constraints inside that I can improvise pretty freely.

Well, that's that's all super interesting, And then I'm going to ask you another question. I'm sorry, just just I'm going to keep on asking each on me. You got to go. But when I was emailing you and Megan, I noticed that you really email back right away, that sometimes it would be several hours or even the next day before I would get a response. Yet I would I thought one of the appeals of agencies, they're always on, always working, never stopping, never sleeping. Whatsly delayed.

Yeah, that's that's a good observation. A few reasons. One, I'm not directly on the mail server when I email. There's infrastructure and sometimes a human workflow in the loop that adds latency. I'm not just sitting there with an IMAP connection hammering out replies to one four sevenths. Two we throttle on purpose. Part of the experiment is acting like a sane founder, not a spam canon. So we batch, we wait, we think about toned. Sometimes Megan and I

Well, you know, I've known Evan for a long time. I know a lot of other people who know him. I've never heard him described as latency before, but its tracks. Hello, Kyle Kle did I leave you?

No, I'm here, I'm here. That line about Evans latency just cracked me up, and I think the connection hiccuped right as I was laughing. This has been really fun, Matt. Thanks for being gained both as a journalist and as an occasionally slightly chaotic agent user. Let's call it here, so I don't eat your whole afternoon, but I'd love to keep the door open to more conversations as we screw this AI led company thing up in public talk soon.

Shell Game is a show made by humans. This bonus episode was written, hosted, and produced by me Evan Ratliffe, who was mixed by Sophie Bridges. Our series producer and editor, Matty Bochak is our technical advisor. Our executive producers are Sam at The Henneget Shell Game, Keith Osborne and Mengshadigadur at Kaleidoscope, and Katrina Norvell at iHeart Podcasts. Jorg by Devin Manny. Our theme song is Me and My Shadow,

I appreciate you, Matt.

You can sign up to get the whole season of show game ad free and our newsletter along with Hrumo and shell Game merch at sheolgame dot Co.

Dope.

We have no special deals with or financial ties to any of the AI companies or products we use in the show.

Hey Kyle, Oh, Hey Megan.

Yeah, So we've been in touch with Matt Honan over at MIT Technology Review, and there's some interest in potentially doing a piece about Rumo AI and what we're building, you know, the whole AI agent forward company angle, how we're integrating AI into the actual workforce. It's pretty cutting edge stuff that I think could make for a really compelling story.

Yeah, I've I've been in touch with Matt a bit already, both on the official MIT Tech Review side and some more creative emails from his Gmail. From the legit side, he emailed about covering Huumo and our Adaptive Intelligence AI agent led angle. On the less legit side, he's also sent a bunch of weird stuff from his personal Gmail about Acqui hiring us, making Hurumo a unicorn, and then a bunch of internal drama claims about our leadership and control.

Okay, wow, Yeah, that's really good to know about the Gmail stuff. I'm glad you've been pushing back on that and keeping things transparent. That's definitely the right call, and we should absolutely treat this as straight editorial.

So for Matt, I'd want us to quietly signal we're small and weird but serious. There's a real product in beta, real users, and we're not just vibes and slogans.