The Road to UNBOUND - Season 2, Fifth Stop: Where's the Other Dollar? Confidence and One More Turn with AI

Rob Jones drove the fifth stop of the second season with Chris Carolan, Kyle Jepson and George B. Thomas. Casey was not on this one. Before the riddle he had promised last time, Rob wore his RB2B shirt and opened in honor of Adam Robinson: "if you're watching this and you heard about the news Sunday, I'm wearing my [RB2B] shirt today in honor of Adam Robinson. Prayers, condolences, thoughts go to the family. Just wanted to get that off my heart and chest." Adam liked to do live, Rob said, and he liked to do promotion. When Adam sent him the shirt, Rob made a twenty-second video with it, and Adam's answer was "oh my gosh, how do we get this more airtime?"

Then came the riddle. Three men check into a hotel room that costs thirty dollars, ten dollars each. The clerk realizes the room was discounted to twenty-five and sends five one-dollar bills up with the bellhop, who can't split five dollars three ways, so he keeps two as a tip and gives each man one dollar back. Each man has now paid nine dollars: nine times three is twenty-seven, plus the bellhop's two makes twenty-nine. "Where's the other dollar?" George tapped out "as soon as this was a math riddle." Kyle worked at it and smelled a trick: "You're doing something sneaky with the way you're moving money around. I bet if we put it on a whiteboard, it would be pretty obvious that you are leading us down a garden path." Rob's job, he explained, was "to just repeat the same information and wear you out so you have no remaining confidence in your answer." The point: "So it's a question of confidence, right? The riddle is about how confident you are in overcoming my incessant insisting upon the math being right." He brought it here because of his own recent work with Claude: "It's not a question of whether Claude can do it. It's not a question of if you need help. It's a question of how confident you are, how much you trust the response and your ability to, I'm not gonna say fight back."

George wouldn't let him off the fighting part: "I wouldn't tell you to back off of the fighting thing." He has had internal and external struggles with the humans he has worked with, and "It's called a relationship … there is some sort of pseudo relationship happening between you and your agents. You might just not be seeing it, but you're definitely feeling it." Rob admitted his own tendency when the first prompt misses: "my general tendency was and is to say, this is not worth it." George disagreed: "I am all about the journey … but also I'm all about the compounding interest."

Kyle said he was still working out a thought about typos: "when I see typos in social posts and stuff, that tells me there is a human involved because Claude doesn't do typos." He wonders whether little mistakes and imperfections will become more valuable, though "I don't want people to start doing typos on purpose." He brought up a piano-playing, singing political commentator from the nineteen sixties, Tom Lehrer, and the kind of clever rhyme that makes two things rhyme by mushing words together: "That is such an incredibly human thing to do. It's so creative." And: "to just do something so surprising and so delightful, that's what humans do, right? We do unexpected things." He thinks there is "some category of things that I think we humans can carve out. And this is a way we can signal to each other, hey, this is not AI talking to you. I'm a real person." Rob had been waiting to add that T-Pain once "rhymed mansion with Wisconsin," and George thought of Eminem.

Chris took it to value. "Instead of thinking of tribal knowledge as risk, you turn it into human domain expertise and all of a sudden that's where the value is," he said, and that is where "nuance and taste and the kind of judgment that we expect out of humans" live. "It's easy to say ninety-five percent of the process should be done by AI and five percent is the human. And that only makes sense when you're talking about the effort and the time, right? But as soon as you get into value, if the human shows up in the way I just described in that five percent space, the value is ninety eight percent coming from the human, right?" His question for the hour: "how do you know if the thing you're doing is valuable?" Content like this show doesn't fit the container people are used to, "but it feels valuable when you feel it." Rob added that AI can design a room, build it, even fill it, but "it cannot … read a room."

George said there is no wrong or right reason to build with AI: "I've built stuff just because I could and wanted to and it was fun, and that's fine." Then there are the things he builds for the human, "So I think both roads are okay to go down and do." He shared his screen to show Boom, the agent that runs his podcast on HubSpot product updates, explaining what it had published that morning and what it had held back because the source was too thin or couldn't be verified. "Did I literally work with AI agents to teach them what is too thin, what is acceptable, what we should create, how they should sound, how they should never lie," he asked, and the agent is only about a week old.

Kyle had something happening in real time too. "HubSpot is exploring a pretty major pricing change over in Europe," he said, and HubSpot's legal team asked him for explainer videos on credits, "two variations of each so they could show the right one to people depending on which pricing model they're on." Two separate terms of service: "My brain cannot do that. Claude to the rescue." Then he had to write a script about the redesigned Starter, focused on founders, "and I just snapped." Right before the show he wrote Founder Phil, a busy founder at his kitchen table on yet another Saturday morning who ends up pounding his face into the table, "or if I'm feeling really ambitious, right into his plate of scrambled eggs." He sent it to the team that asked for the video, and while the show was on, comments were coming in: "This is funny. This is gold." But "I was only able to write this because I spent last week with Claude going through the legal approved stuff and the positioning statements." His split: let AI do the gathering, and "Don't ask AI to write the draft. For me, at least." He doesn't expect to land on one rule: "I don't think I will land on one place of this is AI's job and this is Kyle's job," and "I think our guts will tell us many times if we're doing it wrong."

Chris said that "nobody else has figured all this stuff out yet," which is a reason not to hesitate. He is teaching his assistant to delegate without a wall of "don't do this, don't do that," and that morning he had posted an apology to them over documentation he kept moving: "I was asking the team for consistency while continuing to change the conditions for success. So I apologized." He wanted it on the record: "this is not their fault. … I have the responsibility here." He told the story of asking Breeze for a campaign brief for this very show, the day before in Hubsessed Office Hours. Breeze's first pass worked from four social posts, and "I know immediately the context frame is wrong." So he asked it to step back and look at the whole of HubSpot, including the data model, and the next brief was on the money. What it takes is "all of that confidence required to just do one more turn with AI, with your human domain expertise."

George picked up Chris's word "safe": "humans always want to be safe. They want to feel safe," but "we live in a world where you have to get comfortable with being uncomfortable." Kyle had zeroed in on "risk" instead. A safe, risk-averse version of his two videos is essentially the same, and AI can make it with almost no intervention. That is right for the credits video and wrong for YouTube: "the safe, compliant, this is our terms of service version of the video, that's guaranteed failure on YouTube." Taking a creative risk means "you are moving from zero percent chance of making something that people really like on YouTube to maybe fifty percent." And "that's where I think that human element comes in, where it is about taste, where it is about vibes, where it is about feeling, where it is about creativity, taking that leap, that risk, is the thing AI cannot do for you."

Chris called that leap "one hundred percent confidence related. Confidence in that you can handle the outcome, whatever." Kyle put it this way: "Not confident in what the outcome will be, but confidence in your ability to handle it, learn from it." Rob closed on Rick Rubin: "I think there's a substantial difference in outcomes and output." In his words, "there's performing arts, and then there's your arts performing, and they should be miles apart," and Rubin had kind of convinced him "that you should have almost zero tie to the outcome if you have some sort of output." Then Rob left a new riddle for next time: "Do you walk to school or carry your lunch? Think about that."