HubSpot AI vs. Human Judgment: When Should You Trust the Machine?

If there’s one topic we couldn’t stop talking about last year, it was AI—especially within the HubSpot ecosystem. From the rebranding of CMS Hub into Content Hub, with its early AI-powered features, to the release of tools like Content Assistant, social agents, and Breeze Intelligence, it's been a big year for HubSpot.

Specifically, HubSpot AI went from an interesting addition to an essential part of how the platform works. And while the promise of these tools is exciting, it also raises a critical question:

Where does AI fit into our workflows, and where does it risk replacing critical human judgment and perspectives with convenience?

We’ve touched on this in previous episodes, but the truth is we’ve all been learning alongside these tools. Initially, we didn’t have enough time to experiment and see where they shine—and where they fall short.

🚀 HubSpot AI Data Sources Series:

• How to Get Started with HubSpot AI Data Sources
• HubSpot AI Data Sources: Brand Voice + Tone
• HubSpot AI Data Sources: Buyer Personas vs. Ideal Customer Profiles
• HubSpot AI Data Sources: Marketing Strategy

But as these tools have become more deeply integrated into HubSpot, we’ve had the chance to test their limits and ask ourselves when should we trust HubSpot AI? And when do we need to rely on our own expertise to avoid mistakes, missteps, or, frankly, mediocrity?

What’s clear is that this isn’t a one-size-fits-all conversation. For some tasks, like streamlining workflows or analyzing meeting notes, HubSpot AI can significantly boost efficiency and save hours. But for others—like crafting a compelling narrative or building trust through content—the human touch is still irreplaceable. 

🔎 Go Deeper: Why Do Most Companies Fail with AI + Content?

So, in this episode, we dig into the nuances of when AI should act as a teammate versus when it becomes a liability, exploring the line between empowerment and over-reliance. Because our conversation this week isn’t about choosing sides between humans and machines. It’s about finding the balance.

We explore the specific ways HubSpot AI can actually make you work smarter, rather than harder—like saving time on repetitive tasks or generating content drafts—and where it might miss the mark entirely, from workflows to reporting. Most importantly, we discuss how to ensure the human element remains central, whether you’re creating content, refining processes, or connecting with your audience.

Keywords

AI, HubSpot, marketing, content creation, CRM, human judgment, technology, automation, workflows, digital marketing, AI limitations, workflows, reporting, content creation, human oversight, HubSpot AI, digital marketing, automation, user experience, technology risks

What We Cover 

• Where HubSpot AI Adds Value: We explore how tools like the social agent, Content Assistant, and Breeze Intelligence streamline processes like meeting prep, sentiment analysis, and content remixing. These tools excel at repetitive tasks, freeing up time for strategy and creativity.

• When AI Misses the Mark: From workflows that create chaos instead of clarity to AI-generated reports that lack the necessary context, we discuss examples of where HubSpot AI tools have stumbled—and why.

• The Role of Emotional Intelligence and Strategy: We dive into why tasks requiring empathy, creativity, and strategic thinking still demand a human touch, especially when it comes to building trust and fostering authentic customer relationships.

• The Risks of Over-Reliance on AI: Overusing AI can lead to generic content, missteps in workflows, and even psychological effects like diminished motivation. We explore the implications of leaning too heavily on the “easy button.”

• Content Creation: Starting Point or Refinement Tool?: We discuss how to use HubSpot AI tools effectively in the content creation process—whether as a starting point for generating ideas or a refinement tool for enhancing drafts.

• The Future of HubSpot AI: Max envisions what’s next for HubSpot AI, from tribal knowledge engines to adaptive workflows that understand your portal’s unique data and processes, reshaping how we think about CRM tools.

• Practical Guidelines for Balancing AI and Human Judgment: The conversation wraps with actionable advice on when to trust AI, when to step in with human oversight, and how to integrate these tools strategically without losing the human element.

And so much more ... 

Episode Transcript

[00:00:00] George B. Thomas: Me

[00:00:00] Max Cohen: See, I hate that because I hate it because I thought we were just diving right into the conversation I was like, oh george isn't playing the intro and then you pull that and just caught me off guard

[00:00:10] George B. Thomas: You like

[00:00:11] Chad Hohn: Oh, the

[00:00:11] Max Cohen: I'm, not ready for that dude. I'm not ready for that. I haven't thought about hub spot for 14 days. I don't know how long i've been off and

[00:00:18] George B. Thomas: The holidays, right?

[00:00:19] Max Cohen: Yeah, I was like getting right I was ready You And then you just you threw me off completely

[00:00:24] Liz Moorehead: You know, let's you know, let's call it guys. Try again tomorrow. Let's

[00:00:27] George B. Thomas: yeah, let's just, let's, all right, bye listeners.

[00:00:30] Liz Moorehead: Bye, everybody. It's been good. It's been

[00:00:32] Chad Hohn: See you later.

[00:00:33] George B. Thomas: Just kidding.

[00:00:35] Liz Moorehead: So, how have you guys been? I missed you

[00:00:37] Max Cohen: i'm gonna be honest. I'm gonna be honest i've um Give me one second here.

[00:00:42] George B. Thomas: lack of snacks for this

[00:00:43] Max Cohen: No, no lack of snacks. No lack of snacks. I've racked up around 349 hours in Farming Simulator since we've last

[00:00:51] Chad Hohn: brother.

[00:00:51] George B. Thomas: Wow.

[00:00:52] Max Cohen: Now, granted, uh, there's sometimes where I just kind of like keep it on and walk away from it, and, but, you know,

[00:00:59] Chad Hohn: Oh, you gotta get to get those farm hours in, you

[00:01:01] Max Cohen: I got some hours in.

[00:01:02] 100%. Yeah, um,

[00:01:04] Liz Moorehead: sleeps, man.

[00:01:05] George B. Thomas: Jeez,

[00:01:06] Max Cohen: you know what? We're just trying to grow better. Okay?

[00:01:09] George B. Thomas: Oh, wow.

[00:01:10] Max Cohen: We're just trying to grow better.

[00:01:12] George B. Thomas: Really? Okay.

[00:01:14] Max Cohen: make it happen.

[00:01:15] Liz Moorehead: That's the vibe of 2025. That's what we're,

[00:01:18] Max Cohen: Yep,

[00:01:19] Chad Hohn: get 'em like a tractor, neon light back there?

[00:01:23] George B. Thomas: because we forgot what HubSpot is. We are

[00:01:25] Max Cohen: I ordered a joystick. I ordered a joystick. Okay, to control my, you know, uh,

[00:01:32] Chad Hohn: Your excavators and

[00:01:34] Max Cohen: excavators and yeah, tree harvesters. We're going hard,

[00:01:37] George B. Thomas: Wow,

[00:01:38] Liz Moorehead: You're right, Max.

[00:01:39] George B. Thomas: when you grow up.

[00:01:40] Max Cohen: I might be.

[00:01:41] George B. Thomas: You know what we should be? We should be farming leads.

[00:01:44] Liz Moorehead: yeah, I was trying

[00:01:45] Max Cohen: Let's go, dude.

[00:01:46] Liz Moorehead: here too. That's a better one. I like that.

[00:01:48] George B. Thomas: was, I'm, I'm trying here. I'm trying,

[00:01:50] Liz Moorehead: We're doing our best. We're doing our best, but we are back new year, new episode, but on our same old BS. I'm very excited about today's episode because. If you are a long time listener, first time caller, you know that we spent the A good number of episodes before the new year talking about HubSpot AI's new tools as we went through the different data source settings throughout the platform.

[00:02:18] But AI has been a topic for a while now, not just because of everything that's happening in the headlines, but particularly within the HubSpot ecosystem. It began really in force last year when the HubSpot CMS Hub relaunched as the content hub. And that was when we really started seeing HubSpot leading with some.

[00:02:37] A. I. Powered tools within the platform. But then we saw it continue to escalate when we got to inbound with freeze intelligence and all of the different ways that A. I. Is now basically part of the D. N. A. Of how HubSpot works. Now, while we've talked about HubSpot or AI and HubSpot before, I have to admit, guys, one of the reasons why I'm thrilled to be revisiting this topic is we were kind of learning alongside everybody else, right?

[00:03:04] We were experiencing all of these updates in real time. We really hadn't had a lot of time to do a lot of Excavation of best practices on our own. Um, so this is still going to be a conversation that will evolve as we all learn. But we're here to talk today about the big question when it comes to HubSpot AI versus our own human power judgment.

[00:03:27] When do you trust? The machine because AI makes it so easy to generate content and strategies without much thought, which leads me to my favorite rule, which is applicable to marketing HubSpot or quite frankly, anything else in life. It's the Jurassic Park rule. Just because you can, doesn't mean you should just because you can build AI content dinosaur doesn't mean you should.

[00:03:50] So today's conversation isn't about picking sides. It's about understanding where HubSpot AI tools really add value versus where our humanity becomes irreplaceable. Now, George, I want to start with you because your topic at InBound, Was literally about this idea of human powered AI assisted when you're thinking about your AI content, your AI tools as the critical mindset HubSpot users need to embrace when they're thinking about AI.

[00:04:22] But what does that really mean? And I would be curious if any of your views have shifted on that at all since we gave, since giving that talk,

[00:04:30] George B. Thomas: Yeah, so the talk, if I boil it down to like three major points, it's one, AI is a tool, not a replacement. In, in the conversation, that was a big piece that we talked about, that, um, AI should never be the easy button or the silver bullet solution. Like, um, the, the human context matters, and Um, again, we are, we are the leader.

[00:04:53] We are the brain, even though AI is really, really freaking smart, especially with like the latest models that have been launched, um, a big piece of the talk was context is crucial. Um, I'll talk about something that I've added to that, uh, here as we've kind of been continuing down this AI journey. And then, um, another part was frameworks for effective AI use, Liz, and like, one of the things we literally, uh, gave away was like a checklist.

[00:05:21] On how to humanize. And this was specific to content because my conversation was on content. Although many of the things that I just said, context is crucial. AI as a tool, not a replacement could be far outside of the content realm. But if you think about, um, what are the frameworks that keep you rooted to where you need to be?

[00:05:40] What is the context that you need to give it and how is it a tool? Not a replacement. My mind hasn't shifted or changed on any of those. I still believe those to be true. Um, one of the things that I have been playing around with to go along with context is this idea of cohesive. Um, and, and what I mean by cohesive is like being able to string thoughts together, right?

[00:06:06] When I gave this talk back at InBound, It was, we very much lived in a world where it was like we had one conversation and we did one thing and then we went to a next conversation and did another thing and now with the way that, especially this is less HubSpot, uh, for a second, but more like if we lean into the chat GPT and the, the birth of Canvas, uh, that we have in there and the, the birth of O, uh, 4, Now being able to do things where you can have a cohesive thought across multiple things.

[00:06:38] I'll give you an example around content creation. I was messing around, uh, this is a couple weeks ago, and I said, hey, I want to create four YouTube scripts. I want the four scripts to layer on top of each other and hand off to each other and I want to teach this topic and I want to teach it in this way.

[00:06:57] And then in an individual canvas, in one conversation, so meaning we had four canvases, we literally had four rough draft scripts for these videos. To which then in the same conversation, because we had those scripts, we then asked it for LinkedIn posts and we asked it for YouTube descriptions. Now, we never left that one conversation, but we had a cohesive thread based on the context from a tool that was assisting us to do a task based off of a framework for videos to hand off to each other, right?

[00:07:35] Um, so this is where, so where we're getting now, I think this is still, uh, You can still think about this when you're thinking about HubSpot, like, how does this begat that? How does the, how do we make this cohesive as we move forward when we're thinking about things, by the way, aka data sources, when we're thinking about the, uh, service bot?

[00:07:56] The social bot, I should say agent, right? But I don't know if we're really in a world of agents yet, but we're getting there, but when, when we get into the agents, content, agent, social agent, uh, prospecting agent, like now, now the thing that we need to be thinking about is cohesiveness the board. So that that's my thoughts.

[00:08:17] So let's to get us going.

[00:08:18] Liz Moorehead: Chad, I want to ask you, Where do you feel that HubSpot's AI tools shine the most right now? When you get into that ops mindset, which is one of the most incredible things that you bring to the show, what, what are the specific tasks or processes where they consistently outperform human effort? Because George, to your point, a lot of the conversation you had during that inbound talk, that was about content, right?

[00:08:42] That is where humanity in theory is supposed to be on display. But I think when we have these AI and particularly generative AI conversations. We forget that sometimes it's not necessarily that kind of output.

[00:08:54] Chad Hohn: Yeah. Yeah. I mean, uh, for, for me, when I'm looking at where HubSpot's putting AI in, in what I often do, a lot of the ops integration, connecting system type stuff, it's not really there for me and what I would do yet. Meaning like. Inside of a workflow. I don't have, unless it's a third party tool, some sort of AI function, but where it is very, very shining is in the scenario where it's augmenting people's daily manual tasks.

[00:09:28] So maybe the thing that I would build is something to. Properly task humans to do something or to make sure that they're aware and trained on when their sales meetings are coming up. And then you can use the AI meeting, prepare and bam, it will look through all of the notes on the company record. And it'll look through all of the previous meetings and all the last calls.

[00:09:52] And you can ask it for like a sentiment update in breeze. Like, I mean, that's pretty amazing. Being able to like, Just parse through all of that information in an easy to digest manner to get you ready for your upcoming meeting. That's pretty awesome. Like, but when it comes to like getting it to do asks for you automagically, you know, that's, I think a little bit where we're not quite.

[00:10:19] You know, it's not built in to that point yet. Cause I think there's still making sure the guard rails are up longer for a season where you're confident that you're actually going to be able to have that framework that George is talking about in before it starts trying to offload a whole lot into, well, this one part of the process can be done and saved and logged into a property or something like that.

[00:10:44] Right. All with AI, right. Um, It gets a little bit more difficult there because we wanna make sure it's doing what we need and we have all of our own business guidelines and guardrails to be able to put in so it can do what we want it to every time. 

[00:11:00] Liz Moorehead: Max, what about you? Where do you see HubSpot's AI tool shining?

[00:11:04] Max Cohen: you know, again, for me being the guy who's always, um, you know, uh, constantly championed the idea that Content is what runs, you know, the entire inbound machine that you create. Like, I think a lot of that, you know, in terms of like where it shines the most is really just the assistance around that content creation stuff.

[00:11:23] And I think like, even when you talk about, you know, the content creation and I'm thinking specifically about like written word here, right? Like, yeah, all the, you know, mix the, I forget, I forget what everything's called now, cause I've been on break for so long, all the, yeah, content remix stuff is like super cool.

[00:11:38] Just from like a. Time saving perspective and content reuse perspective. I think that's really great because it can really amplify what a marketer is doing Um, but you know even just where it comes into something as simple as like, you know, assisting people who just aren't great at writing You know what?

[00:11:53] I mean? I'm not good at writing at all. Right and like when I write emails, I write long Long emails, right? So like having some, Oh yeah, totally. Because I feel like I have to explain every single thing. I want to leave no room for questions. Like I want to leave nothing to the imagination. I always, you know, I tend to be someone who like over explains things.

[00:12:11] Right. And for me, it's really great to have tools that help me like shorten that message and make it more concise and fix all my terrible grammar and punctuation and like all these other things.

[00:12:20] Liz Moorehead: I totally do that. Like, here's an email. Please make it 75 percent shorter.

[00:12:24] Max Cohen: Yeah. And even just that, even just that, this is something that's like great for salespeople to have. Like. I'm sorry, I've gotten some really poorly written sales emails before, right? And, you know, especially, you know, from SDRs and BDRs fresh out of college that like don't really know how to write emails well, right?

[00:12:41] So like even having something that's like acting sort of as like a glorified grammarly like in your CRM just to be like, hey, you know what? There's a, there's a better You know, more professional or, you know, a better way to write what you were just trying to write to that person. You know what I mean? I think having that is great to still like kind of improve the output quality of their content, quote unquote, that like a salesperson's creating.

[00:13:03] Cause if you think about a salesperson, their content creation is their outreach, their communications, their emails, their this, their that, right. Um, I think it's really, really great for that. And I think it really falls in line with, you know, All the other sort of like time saving stuff that like HubSpot's built over the years to make salespeople's lives easier, right?

[00:13:20] You look at things like Sequences and templates and you know all these things that are just meant to you know, help automate those repetitive tasks that you have Well, it's like hey now the the task of creating that thing in the first place It improves the quality of right not just reducing the amount of time you're spending on it, right?

[00:13:37] Which also yeah, you can make an arguments reducing the amount of time you're spending on Proofreading an email or just making sure you're wording something correctly, right? But it's also you know It's gonna fix up your sentence structure and it's gonna make you like write it like someone who knows how to be a wordsmith On a keyboard, right?

[00:13:52] um, and I think that's gonna be great for folks that want to focus on selling not focus on relearning how to You know, compose the English language in a, in a professional and good quality way. Right. Um, so yeah.

[00:14:05] George B. Thomas: Max, because you're, I'm listening to you, my brain goes in two directions. One, for all of you listening to this, if you have access to Remix and you haven't actually gone back and, like, used it in a while, uh, I would go back and try it again, because it has had a little bit of a facelift and also, like, some additives under the hood.

[00:14:25] I feel like maybe they added a little bit of a turbocharger under there or something. It was,

[00:14:29] Liz Moorehead: I feel like we need to have an episode on this because we haven't talked about it.

[00:14:33] George B. Thomas: yeah, so I was teaching it the other day and, and I always love when I'm teaching something and then it like looks new and then it is kind of new, uh, cause it's like you, you have to be good at like, I'm just going to act like I've seen this a hundred times and we're going to teach it together and learn it together.

[00:14:49] Max Cohen: Try, try doing that in front of a room and 90 new hires live in the moment in front of a giant projector. That was the worst. So yeah.

[00:14:57] George B. Thomas: So, so that's the first thing that comes to mind when like check out remix. If you had the second thing that came to my mind was like, If you knew and I'm, I don't know how other, how else to paint this other than like, if you knew that you had a weakness, right?

[00:15:12] So I'll, I'll pretend for a second that I, I'm not a very empathetic human and I understand. I'm just not very empathetic. I'm very matter of a fact, right? Like, here's the facts of the matter. If I realize like, I'm that type of person, you. I can see where this gets real interesting to be able like, um, can you please take this email and add a slight bit of empathy to it or add a, add a slight bit of excitement to it or, or add a, add a slight bit of urgency to it.

[00:15:41] Max Cohen: Mystique.

[00:15:42] George B. Thomas: Like, yeah, like, whatever it is, right? So, like,

[00:15:44] Liz Moorehead: Add a little human razzle

[00:15:46] Chad Hohn: A little demure

[00:15:48] Max Cohen: Give it some zhuzh.

[00:15:52] George B. Thomas: No, but, but I'm being, but I'm being dead serious here, right? Like, think about, um, and by the way, if, if, if I was a sales guy, which I kind of am, but if it was like my full time thing and I knew that I had to create, like, these dope emails, I would probably do what I'm saying for the same email.

[00:16:09] Yeah. Like, and I would look at the six or seven different ways. Well, what does it look like if it's written with empathy or if it's written with urgency, or what if I did a happy mix of the two and I would, cause I could, cause you can so quickly iterate. Based on these, like, pinpoints of, of, of thinking, to get then, look, ooh, I really like how that went, and talk about A B testing your sales emails, add the, like, I want one that's empathetic, and I want one that's, uh, urgency, and let's A B test those and see what happens, like, so,

[00:16:41] Max Cohen: You know what I think would be, the thing that I'm like, excited about, and I don't know how they would deploy this, but like, think about, I guess the word we would use is like, tribal knowledge, right? Think about how much tribal knowledge is contained within a company's like, HubSpot CRM portal. And like, by that what I mean is like, HubSpot CRM portal.

[00:16:58] Your internal knowledge bases, your records of conversations that you've had with customers, information about your product that's hosted on your website, your K base, your tickets that you have, all these different things, right? If you think about it, your CRM is like this giant repository of information on how you've answered people's questions and ways you've solved problems, right?

[00:17:20] Uh, you know, through all this stuff, right? Like, It's so cool because I think like HubSpot one day with like their AI stuff could have this whole like, I mean, they'll probably call it something different, but essentially this like tribal knowledge engine, right, that understands how to solve all these different problems the way your company does it internally, right?

[00:17:40] And that information can only be trained on and found in the records of communications you've had with your customers and the content that you're putting out into the world and how you've solved similar problems in the past, right? Like, Imagine a day where like a ticket gets created, and when you go on to that ticket record or that conversation or whatever, right, the AI is able to say, Hey, this looks really similar to a ticket that Jeff B solved back in a couple months ago, and like, this is how they came to, you know, this is how they figured it out, right?

[00:18:13] Maybe this is something similar, right? And it gives you instead of you going, let me search the data. Let me search my search bar to see if I can find any like tickets that are something similar, right? It links you to like hey, here's an example of someone solving this problem for this other customer, right?

[00:18:29] Like that is going to be Completely I don't want to say game changing

[00:18:34] George B. Thomas: No, no, it will be.

[00:18:35] Max Cohen: changing that will be like such a giant fundamental step forward right,

[00:18:40] Chad Hohn: it'll change the game.

[00:18:42] Max Cohen: And it goes way beyond like here's an answer and a knowledge base Because like that's something that's like so unique to the inside of your organization right being able to find like Past examples of complex problems that are coming up again, that have been solved in the past three different way.

[00:18:57] And then being smart enough to understand that and be like, Hey, this sounds similar here was the solution, but you can go look at it if you want to. Right. That's going to be.

[00:19:05] George B. Thomas: Well, the good thing too, the good thing too, is it's not, it's, it's not necessarily needing to look at somebody specifically went and created an SOP or documented it's looking at the conversation that happened in the system. Therefore, Hey, this is what happened. Here's the thing that excites me about that too, is it's, it's three employees ago, it's six years ago, it's like, you know, like, Hey, on June 19th of 2027, Bobby ran into this problem with Frederico.

[00:19:33] And this is how he fixed it. Like, and what's great about that is it's now all of a sudden, what, what was exciting about content is we were like, there's no more writer's block. There's always a beginning draft. You have an idea and you just, at least you have a place to start. Now, when you're talking about fixing people's problems, you have a place to start because you see the finish lines

[00:19:55] Max Cohen: hmm.

[00:19:56] George B. Thomas: of where you're trying to go with this human anyway.

[00:19:59] Liz Moorehead: I'll even take it a step further. I'll even take it a step further, George, because one of the things I really like about it, it reminds me of human nature, right? We are more. Outspoken and reactive when we see something we don't like and so what I've actually found that is helpful with AI is that it kills that blank document syndrome that becomes like it's that's where writer's block begins right but even if it gives me absolute trash even in that trash moment I'm like well I know definitely it's not this and it still gives me a starting point of where I need to go which I love so I want to I want to switch the conversation here a bit.

[00:20:38] I want to talk about the potential risks or downsides of leaning too heavily on HubSpot AI tools. Like for example, have you ever experienced or heard of a situation where HubSpot AI just produced something that completely fell off or completely missed the mark?

[00:20:53] Max Cohen: You know, I can see it. I can see there being issues for like, I think about like the workflows tool and how like you can use a workflow to like create actions. Right. I could see a situation where someone who's not as understanding of the nuances of HubSpot would just like tell a prompt to build a workflow and be like, All right, well, you know, I told HubSpot what I wanted, and it did it.

[00:21:15] So like, it must be right. But I'm not educated or understanding enough of how like, you know, Workflows can have different effects on things to know like to check its work if it's right or not right because then they can

[00:21:27] Chad Hohn: if it's not specific enough, like, cause if you're not specific enough to say, oh, well, I want these things to be updated when meetings are completed, but only the first time, or if it happened, you know, like, I mean, it doesn't understand that unless you tell it,

[00:21:43] Max Cohen: Exactly. So like so for some new users who like, you know CAI is this like all powerful thing, but don't necessarily understand its limitations or like You know, are hyper aware of the nuances of how it might not like fully understand exactly what you want to do. Right. But just kind of like build something that sort of like aligns with it.

[00:22:02] Right. And then folks kind of leaning in on that too heavily, like that could be a risk. Right. Um, I think you do also like, you know, I've talked about these risks before, like, you know, You do totally have the potential with these tools in front of you to just put a lot of trash out there, whether it's content, you know, or, you know, email communications or anything like that, where people just go, Oh, AI sees you button and you hit it and boom, there you go.

[00:22:23] Right. And it's just, you're just pumping out garbage and people hate you because of it. Right. And not like using it as a tool to like assist with the creation versus using it as the thing that just builds everything so you don't have to think about it and you can go play Fortnite or something like that, you know?

[00:22:37] Um, so that's like still kind of what I'm worried about. Right. Um, but I, I, I feel like as HubSpot kind of like moves more into those realms that we were talking about around the, I don't want to say the portal becoming self aware of itself. Right. But the portal may be

[00:22:57] Liz Moorehead: All hail Skynet.

[00:22:58] Max Cohen: here's, here's kind of like what I wanted to say before.

[00:23:00] And I think it kind of weaves into like solving that problem that I just said. Right. You know, Imagine we had the, you know, full fledged, uh, prompt thing inside the workflows tool. And as you built your workflow, it could talk to you about the implications of what you created, right? So, for example, if you're building like a workflow and you add an action, that's like just changing like a property.

[00:23:26] Right. What if little, you know, HubSpot clippy AI popped up and said, Hey,

[00:23:31] Chad Hohn: we'll call it hippie.

[00:23:32] Max Cohen: Hey, just so you know, just so you know, if you change this property. It might go in like trigger this other stuff and these are some of the effects that that could have right because maybe it sees the action that you're adding runs that you know simulation in its head and goes oh shit all this other stuff is going to happen because of this right just to warn you right and then if you could like have conversations with it like oh what if I do it this way what if I do it this way and it knows like it understands what the implications of what that would be because One, it has like a really good understanding of how workflows work, but two, it matches that with the really deep understanding in the context of your portal and the data inside of it.

[00:24:09] Right. You put that together, there is your missing link of the nuance that new users aren't going to have. Right. Um, so that, that'll be a really interesting way they solve that. Yeah, totally.

[00:24:21] Chad Hohn: Because like, think about this. I work at a partner agency and we have multiple people sometimes building workflows inside of our clients, hub spots. And like, what if I'm building something that somebody either has built near or already built or something along those lines, right?

[00:24:41] Then, you know, hippie might be able to tell us

[00:24:44] Max Cohen: Hippy.

[00:24:45] Chad Hohn: what's going on. Yeah. Sprocky we'll get, you know, a little hub spot, sprocky little sprocket in the corner there. And it just, Oh, Hey guys.

[00:24:53] George B. Thomas: Wow, okay, so I love, I love, uh, the little comic relief there, and I also love how you guys are, like, future talking today. This is actually pretty amazing. I, I have to be honest with you, I have seen where, um, Let's, let's just say reporting gets a little s um, because the AI reporting, I, I literally teach it this way.

[00:25:14] I'm like, hey, it's there, play with it, but be careful. And I know there'll come a day when that isn't true, but like, again, you guys went to Workflows, which is one of most mere mortal HubSpot users, AchillesHeel, is the Workflows tool. They feel it's, uh, It's difficult and it's confusing and, and, and it's changing and it's. The other Achilles heel is reporting. And so if you're looking for the like biggest return, you're like, well, let me just have AI do my reporting, which is a, is a, as an opportunity for a disaster. Again, if you don't know the specificity and the actual like formula or framework of like question to insight to like data points and all, all the underlying pieces.

[00:25:56] So I love the idea of it getting smart enough to know all of the underlying peaches and being able to say, Hey, unfortunately we can't give you the report you're looking for because you should be creating a custom property called this. It's tracking that. And over time we could then do what you're looking for, but because it would be based on what, what it is like anyway.

[00:26:16] So that's, that's the other thing here. Here's another piece though, that.

[00:26:19] Max Cohen: hold on wait, wait, wait, can I can I just say and also right? That that report builder needs to be able to ask you clarifying questions before it actually creates what it's building you Right because dude think about how many times you when you ask someone. Oh, uh, When they go, I want a report that does this.

[00:26:37] What do you ask you go? Why do you want it? Like what what is it gonna? What like what do you want it to tell you?

[00:26:43] Chad Hohn: Yeah.

[00:26:43] Max Cohen: like, that is so important to understand how to build that actual report. And like, AI needs to do that too. Instead of being like, okay, here it is. Boom. And give it to you. It should be like, wait, wait, wait,

[00:26:51] George B. Thomas: Yeah. And I, I

[00:26:53] Max Cohen: how do you want to see this data?

[00:26:54] Why do you want to see this data? How, like, how should you lay it out? And then it also should be able to tell you, bro, you can't build a chart that does that because what you're asking for is impossible in the laws of time and space, right? Like think about how many times people have said, I want one chart that shows me this, this, this, this, this, this, this, this, and you're like, I would always say, draw that for me and they go, Oh, I can't.

[00:27:13] Cause I didn't realize that it was impossible to represent on graph. I mean, that's what AI is going to need to do in order to one, train people to be better report builders. Right. But to generate reporting, that's not trash. Right.

[00:27:25] Liz Moorehead: Well, that's true of anything, you know, that idea of like, I think that reminded me of conversations I used to have. This is going to sound strange, but my graphic graphic designer used to work with. She said, I used to get really frustrated with people who come to me telling me exactly how to click a button or design something instead of telling me what problem they're trying to solve.

[00:27:42] And I think we're seeing something similar here, which is there is an assumption on the part of these tools, which makes sense. This is the beginning of these tools. And we talk about that a lot, but that we know what's best for us and what we actually need. So I love this conversation we're hearing right now around this idea of, are you sure?

[00:28:01] What are you actually trying to solve

[00:28:03] George B. Thomas: Yeah. And I love the idea, especially with the reporting Max, because the, the technology is there, right? It can ask follow up questions. Like I love when I'm using a custom GPT and it's like, ask me seven questions and I have to give it the answer to get the thing that I'm trying to get. There's a, there's a, again, it comes back to the context.

[00:28:23] Right? And, and with reporting, it asking you five, six questions and then trying to build the report would be so much better. But here's, here's

[00:28:30] Chad Hohn: a wizard. Would be cool. Like a wizard to set it up that takes, cause like, what's, what's a report? It's trying to answer a question, right? And so like, that's the whole goal of a report is to answer somebody's question. And it should basically start by saying like, what question do you want to answer?

[00:28:45] And then ask some clarifying questions and then take a stab at it. Just like when somebody submits a ticket and they go, I want blipity blah, right? But here's the thing that I think is going to be the Achilles heel of AI. it comes to custom properties, trying to build your report,

[00:29:02] George B. Thomas: Yeah.

[00:29:13] Max Cohen: Don't make it the damn question on the floor.

[00:29:15] George B. Thomas: Yeah. Yeah. You gotta, you, wow. Yeah. I mean

[00:29:19] Chad Hohn: otherwise AI is never going to know what the fricks in that property. Right. Like, how's this supposed to know if you don't tell it either a good name and B what happens in there?

[00:29:30] George B. Thomas: yeah,

[00:29:30] Max Cohen: Dude, it's almost like it needs this like ai intervention engine right where it sees someone doing something stupid in hubspot like a

[00:29:38] Liz Moorehead: Help me help you.

[00:29:40] Max Cohen: it pops up and goes. Are you sure? Hey, dude, are you sure you want to do that? All right,

[00:29:45] Chad Hohn: it just needs to send them memes.

[00:29:47] George B. Thomas: so

[00:29:47] Liz Moorehead: Again, like clicky, clippy.

[00:29:49] George B. Thomas: Maguire. We need Jerry Maguire in Huffspot. Help

[00:29:52] Max Cohen: Like the easiest thing is like when you're when you're building a form and you hit create new property and it realizes you're asking a full sentence question, it should just immediately pop up and be like, no, don't do that.

[00:30:03] Like,

[00:30:05] Liz Moorehead: It looks like sales is trying to find out where all the revenue is. Do you need help?

[00:30:11] George B. Thomas: Here's, here's something I want to hit on though, because Liz, ask your original question again.

[00:30:17] Liz Moorehead: Yeah, absolutely. So when we think about what are the potential risks or downsides of leaning too heavily on HubSpot AI tools. So, and then I asked again, like, did you have any experiences with that or situations where HubSpot AI has produced something that's like, why are we here? How did we get here?

[00:30:35] George B. Thomas: yeah. What happens, what happens when you as the human psychologically just start to feel dumb? Yeah. Yeah. Yeah. How are you, how are you going to deal with that? So, when you ask AI to do everything, and, and what was happening historically is that you were learning something, and you were achieving a task, which then would release endorphins in your brain, and you would feel good about accomplishing a task.

[00:31:00] When you're no longer accomplishing the task because AI does, what does that do to you psychologically as a, this is a piece that doesn't have anything to do with HubSpot, but has everything to do with HubSpot, has everything to do with business and has everything to do with us as humans. We're going to have to be careful on How we are reprogramming our brain to perceive a completed task based on assistance.

[00:31:31] assisting us to do the task that we never had before. So I'm just putting a stake in the ground right now in a year to two years or two months or six months from now, when all of a sudden we have this massive thing happening in organizations where humans are feeling depleted, defeated, um, not energized, not motivated, um, not understanding. anyway. I'll just leave it there for people to think about.

[00:32:03] Liz Moorehead: Well, don't hop off the mic too quickly, George, because I want to ask you first my next question, which is when it comes to content creation specifically, because this is a topic you have spent a lot of time on. Do you think HubSpot AI tools are more effective as a starting point? Or as an editor for refining human created drafts.

[00:32:23] George B. Thomas: I think it depends. I think it depends on the human. Like, if you're sitting there, and you're like, don't have access to GPT or CLAWD or Perplexity, you haven't been, uh, you know, waking up with AI along the way and figuring it out or whatever, Um, then HubSpot's great for all of the things for you. Like, you can generate your ideas.

[00:32:47] Um, by the way, you should still probably have human conversations with people inside your organization to generate some of those ideas to then bring into HubSpot to then refine the idea. Um, but I think HubSpot at this point from a content standpoint could be a from beginning to end with AI assistant process. Now, I started this with it depends because I actually do a lot of stuff outside of and then bring into. And so what I do understand, especially when you have your brand voice in place. It is a, it is a phenomenal refinement tool, meaning, uh, voice consistency and saying the things the way that, um, they, they should be said to have this kind of holistic, cohesive, Content engine resource center inside of HubSpot.

[00:33:44] Um, giving it a final brush of like, and make this my brand voice, make this my brand voice, make this my brand voice is a very interesting, um, phase or step that we've seen adding in depending upon the content. So again, I'm a great marketer. It depends both. One? All? I don't know.

[00:34:04] Liz Moorehead: Well, the other thing too, though, is that as you get more comfortable with the tools and as you quite frankly become a more sophisticated content creator, you probably have a little bit more flexibility. Right? So, for example. If I'm working, I'm training a content manager who is very, very green and very, very new, just at the principles of storytelling, the principles of architecting what a great piece of content looks like, understanding how to construct a great introduction, understanding what a great conclusion looks like.

[00:34:37] I'm going to want them to be more involved and hands on in the process before they start bringing in their AI assistance. What I would probably do, I would have them start learning how to use a I on the back end, right? One of my favorite things to do is, hey, this was the key word I was attempting to optimize for.

[00:34:55] Can you just take a quick look and see if I maximized all those opportunities correctly? Or did I miss anything glaring? Um, or helping them integrate it into other pieces of the process? I think there is This idea and I will admit I perpetuated this that I I'm against using it at the front end and only more at the back end when it comes to where AI fits into the content creation process, but really my big sticking point is.

[00:35:26] Do you even know what a good story is supposed to look like if you get a piece of content that is generated from your strategy, your principles, your ideas? Do you even know if the output that is in front of you is good or not? Do you understand what should be changing? So that's the only thing I would say there is that it comes down to your level of sophistication as a storyteller, and that will greatly sway your inputs and your outputs.

[00:35:52] George B. Thomas: yeah, see, I agree with everything you said. What I hope people understand, though, is like, here's what I love about where we live right now. Um, I think every human can look at a piece of content, and I'm going very content specific here for a second, and be like, yep, that's a turd. Nah, I don't like it. And so then you're like, well, okay, I know I need to probably add a little bit of storytelling to it.

[00:36:16] I need to add a little bit of this, a little bit of that. Here's, here's what I love. I love the fact that you can still Google, and you can still grab three or four links, you can still read the material, you can still educate yourself, but then you can educate the, the, the machine, the assistant that you're using on those things that you're, you're doing.

[00:36:35] And so even your turd. After you educate yourself and you educate it, you can be like, Hey, can we try to rewrite this? But based on the things that we've learned together in the six links that I've uploaded with this, that, this, that. Because now you can, and this is gonna sound gross, this is gonna sound gross!

[00:36:53] But now you can mold that turd into something better.

[00:36:58] Liz Moorehead: Could we just go with clay or something?

[00:37:02] Chad Hohn: Well, but see, Clayce is like a canvas that you could do anything you want with, and we know we're starting with a big ol turd.

[00:37:12] Max Cohen: I think, uh, you know, a lot of it is, it's all just in the way I look at it as enhancement and amplification. Right. So it's like, you know, enhance the content you've created, right. By making it better, whether that is, you know, you know, making the words better or, you know, improving the punctuation or the structure or whatever.

[00:37:30] Right. But then like, But I think in the application piece, like that's immediately content remix, right? Like just that, cause you know, I, like, I remember like, you know, saying like, all right, I created this piece of content and then it's like, Oh shit, I gotta go make it. I gotta go write a blog post about it.

[00:37:46] I gotta go make social posts. I gotta go craft emails. I gotta go do a landing page. I gotta go do this. And it's like, Well, now you don't have to have that big giant block of, Oh shit. There's all this other extra work I have to do. Right. You know that you can at least get the scaffolding and the framework of that stuff put together.

[00:38:02] So instead of, Oh, I've got to push this giant boulder up this hill to create all this stuff, it's more so, Oh, it's already kind of like built for me. And all I'm doing is like refining the little pieces and like making slight changes to it. And then it's, it's out there. Right. But it's all derived from your original content you created.

[00:38:17] Right. The other thing too, I think is like great for like a starting point. Like I think. You know, when we kind of talk about it being a starting point, like there's saying, like, you look at something like a blog post, there's a big difference of generate an outline for me, and then I'll add the original content into it.

[00:38:32] So I have this like structure instead of having to again, push that boulder up the hill to like create, you know, zero to something else, right? Which is what gets a lot of people's way when they're really just kind of going out and starting and trying to create content, right? The other option is.

[00:38:46] Generate the whole thing for me and I don't think about it and push it out there Like it's when we say starting point I think we're saying like starting point for you to make it easier for you to create something original Right and to get rid of all those mental blocks you might have around How hard it is to go from zero to 1.

[00:39:02] 0 on a piece of content, right? Um, yeah,

[00:39:05] George B. Thomas: Max, what I love about what you said, and by the way, I'm always more of like an outline guy and then go from there because here's what's fun about outlines. You can be like, okay, I love this outline, but can we also add this and this to the outline? And then you can, and then you can be like, you know what?

[00:39:20] I talked about this on a client meeting. Let me go ahead and get a transcript of that. I actually talked about this in a PowerPoint. Let me go ahead and grab that. Talked about this over here. Let me grab that. And now you can collect stuff that you've historically done anyway and you can upload it and be like based on the stuff that I've uploaded in the outline that we've created, let's build a rough draft from that.

[00:39:41] So it is your brain. It is your words. It is like,

[00:39:46] Max Cohen: you're just reducing the mental lift to get there, right? But then the other thing too is like when you have ai go and generate an outline for you Right. And you go, okay, yeah, this is the stuff I should be talking about. It also makes it that much easier to be like, Oh, here's some of the things that it probably missed.

[00:40:01] Right. Because you're seeing what it created as an outline that's giving you inspiration and making it easier for you to think about the other stuff that you should put in there, if you are going to mention these, you know, X five things, like whatever it may be. Right. So, you know, it just, again, I think.

[00:40:15] Reducing mental lift is, is good, right? Mental health is important, right? And, and, and obviously like reducing your, your, your, your mental lift to create content is going to make it easier to create content. Again, like you're applying a lot of leverage with AI, right? Um, so I think it's just, it's important to think about it that way.

[00:40:31] George B. Thomas: Two, two things I want to throw in here. One, one of the questions I love to ask before I actually move forward past the outline. Is is there anything else I should know about think about or train around this conversation and see what it spits out. It's amazing to go down some rabbit holes with that.

[00:40:49] But 2nd, we, as we do many times on this podcast, we dove into the content side of the world for this. And so, Liz, I would be remiss if I didn't talk about another thing that happened back at Inbound 2024 that was in the presentation. We talked about a content editing checklist, going back to like, humanizing your content and AI being the, the assistant. Maybe talk the listeners through that because I don't know if we've ever really shared that whole checklist on here But if but if they are Listening to this they are like, oh, yeah outlines and oh, yeah These things like talk about that whole piece of this to kind of give them a framework to follow

[00:41:29] Liz Moorehead: yeah, absolutely. You know, when I think about any piece of content that I'm editing, what's funny is the checklist I'm about to share. We built it and developed it and launched it to go in tandem with these AI tools. But this is actually, that's actually just enabling us to Automate or to to apply structure to a process.

[00:41:49] I was already using before AI even existed. Because when we look at a piece of content to your point, George, there is a framework. There's a you're looking at a piece of content. How do I know it's good? Well, because it could be good or bad across a number of different things. It could be factually accurate and grammatically a dumpster fire.

[00:42:06] It could be a grammatical work of art, but a narrative train wreck. So whenever I look at a piece of content, one of the, one of the, Biggest lessons I teach new content managers or anybody who has to look at content is do not edit for everything all at once. Edit in sequence, edit for one thing, then edit for the next thing, then edit for the third thing and so on.

[00:42:28] And there is a specific order in which I do it. And the reason why is because let's say, for example, you have a piece of content that is narratively a trash can, but you've grammatically made it get sparkle, right? Well, you're going to have to rewrite it. So those sparkling new bits of prose that you've thrown in there, If you have a trashy introduction, a conclusion that doesn't make sense, you're speaking to 18 different personas instead of just one or two, you're gonna have to redo that work.

[00:42:59] Exactly. So, here is the order in which I do it. First round, I'm looking for fact checking and accuracy. Right. Is this, is this factually accurate is, are, are the things here? Correct. Um, I always recommend, particularly when you're dealing with AI content, do you know how many times it'll say like, well, I've got this statistic, duh, duh, duh, duh, and I'll say, are you sure?

[00:43:24] And it'll go back and say, you know what? You're actually right. I made that up. I'm sorry. So you have to actually check things. Do not just blindly trust what's put in front of you. Okay. Next, once I have determined that what I have in front of me is factually accurate, right? And again, I'm not thinking about anything else.

[00:43:43] I am thinking about these things in a vacuum. Only then can I start taking a look at readability and engagement, right? How can I simplify the language for clarity and impact? How can I guarantee that I am speaking at the level of the people I am trying to preach, right? So, Molding the turd, shaping the

[00:44:04] George B. Thomas: it's stuck.

[00:44:05] Liz Moorehead: We shape the clay in stages. Right. Okay. The facts are right. Great. Now let's take a look at how readable and engaging those facts are. Once I've determined that those facts are accurate and the words are readable and engaging, then I start looking at SEO optimization. How can I enhance it for searchability?

[00:44:24] How can I make sure that it is including relevant keywords in phrases? Then I do final proofreading. Now, one of the things I always find challenging with most people is they don't understand why this is step four. Again, leave the spelling errors alone. If you're just going to delete that entire sentence, it doesn't matter if we're talking about compooper skills versus computer skills, if the whole paragraph is getting killed, right?

[00:44:50] And then feedback and iteration. So that's really what you're looking at, right? Are the facts accurate? Is it a great story? Okay. Is it optimized well for search? Fantastic. Now that I know I have the right facts with the right framing and the right optimization, if spelled correctly, is that grammar all razzle dazzle.

[00:45:10] And then after that, you can do feedback and iteration. That's the order in which you do it.

[00:45:14] George B. Thomas: Love it

[00:45:15] Liz Moorehead: Now, I would be curious, though, to hear you guys as we wrap up today's conversation. What advice would you give teams trying to figure out when to trust HubSpot AI versus stepping in with human judgment? Are there specific guidelines or use cases that you recommend?

[00:45:30] And if you don't mind, I'd actually like to go first so that I can turn it over the floor to you guys. The reality, my guideline, it's very simple. Never have a process where the human is absent. You're either at the beginning, or you're at the end. At the end, the beginning, the middle and end, the middle and the end, you have to be present somewhere.

[00:45:48] The only way you actually truly 1000 million percent screw this up is to just click that AI button willy nilly with no oversight. And with that, Chad, do you want to kick us off?

[00:46:01] Chad Hohn: Yeah, I mean, uh, for me, I think it goes back to the old adage, you know, just like you're saying trust, but verify, right? I mean, it's like, I mean, I don't think you could really say it a whole lot better. Like, yeah, let's do it. But let's, you know, get, put people in the process. Um, I really, um, In some, like, there are some things that are a little bit more process driven, like the ops hub data cleanup stuff, some of the AI driven contact duplication detection stuff, um, even some AI associations, um, like there's AI, I think called a ticket association feature in HubSpot to like, try and figure out which calls are related to a ticket.

[00:46:41] Uh, based on the content of the call and things like that, that exists and has existed for quite a while. I mean, those kinds of things, you know, they're more operational. They're, they're less, you know, being presented to people. The only real people it causes problems for is the HubSpot admin. If, if it does, you know, so, I mean, I would say like, you could maybe worry a little less about those kinds of things.

[00:47:05] It's like, all it's doing is capitalizing first and last names and. You know, somebody has to verify that the, there is a human with, with duplicate contact or do, you know, detection, right? Where you're, uh, accepting or rejecting the merges and moving data from spot to spot. By the way, if you guys haven't checked out that the new UI for contact merging, it is sick.

[00:47:29] Like you can get in there and tell it, I want two properties from this contact and three properties from this contact. And then I want it to go into that one or this one. Like, it's pretty cool. Like some real cool UI changes to that compared to what it's been in the past. Um, but yeah, like they're putting people in the process, I think, where it makes sense and, you know, trust, but verify that's, that's my advice.

[00:47:51] George B. Thomas: max

[00:47:52] Liz Moorehead: don't be so shy.

[00:47:54] Max Cohen: I was also going to say trust, but verify, uh, you know, it's,

[00:47:58] Liz Moorehead: Are you trusting but verifying Chad's trust but verify?

[00:48:02] Max Cohen: I would say

[00:48:03] Chad Hohn: Like, he's just verifying it.

[00:48:05] Max Cohen: I don't know the way I would put this, right? Uh, what should go through your head? I think when you generate a piece of content using AI, you should just assume that it doesn't know anything about the subject matter and verify it, right? Because it's very possible that it will just say some, some sh that's completely wrong, right?

[00:48:27] And, and totally made up and, you know, Incorrect, right? I'm not saying it's gonna do that. I'm saying you, you know, when you write content with AI, all of a sudden you become an editor, not a writer, right? You gotta be the editor, right? And you gotta, you can trust that it put together a nice piece of content for you.

[00:48:48] You should assume that it doesn't know shit about the subject matter, and you should verify that what it actually outputted for you, uh, is correct, real information. Right. Because one people are already starting to easily be able to like, pick up and understand, Oh, this is an AI written article, especially search engines.

[00:49:05] Right. But two, the last thing you want is them consuming the content and you're given bad information. Right. And making yourself look stupid because the AI just kind of gave some like milk toast you take on some sort of like, you know, subject matter. Right. And, and all of a sudden, you For people who don't know that, oh, this is an AI written article, right?

[00:49:27] They're reading some, some junk from you that's either incorrect or not good, right? So yeah, you can, you can trust, trust that it's going to create some content for you. Verify that that content is actually good and, and, and informationally sound.

[00:49:42] George B. Thomas: Yeah, interesting. So, uh, the way I'll answer this question is, if If you deem that there is, uh, some emotional intelligence needed, or some strategic direction that needs to be laid down, it's all about the humans. If you're

[00:50:03] Liz Moorehead: There we go.

[00:50:04] George B. Thomas: looking for a place where you're trying to enhance speed, or extrapolate scale, then it can be about the AI. Bottom line, if I, if I'm gonna throw a divider out there, that's what I would want you to think about. Emotional intelligence, strategic direction, human, but then leaning into speed and scale through AI assistant.

[00:50:25] Liz Moorehead: George, you're not off the hot seat just yet.

[00:50:28] George B. Thomas: Yeah.

[00:50:29] Liz Moorehead: Help us land the plane. Take us home.

[00:50:31] George B. Thomas: Yeah, this

[00:50:31] Liz Moorehead: If our

[00:50:32] Chad Hohn: He's good at that.

[00:50:33] George B. Thomas: Yeah.

[00:50:34] Liz Moorehead: Tell us what that one thing is. If we forget nothing else from this conversation except one thing, what should it be and why?

[00:50:39] George B. Thomas: Yeah, without a doubt. Um, listen, Chad talked about the new merging UI. We talked about how remix is different. We talked about, you know, there's different AI platforms and should we be doing things just in HubSpot? We talked about how there's going to be different agents that you can like, there's a, there's a lot, ladies and gentlemen, there's a lot.

[00:50:59] And so we are definitely in a world where the thing that I would want you to take away from this, when we're talking about AI. HubSpot AI versus human judgment is experiment and learn, experiment and learn, be curious, test different approaches, refine, reiterate, refine, like, Figure out what works best for your business in a world where you can do almost anything you wanna do at the light of speed. But I'm gonna end it with what Max said. If you're doing things at the light of speed, it's stupid if you're making yourself feel stupid or look stupid.

[00:51:42] Max Cohen: why is he saying it like that?

[00:51:44] Liz Moorehead: Why are you saying it?

[00:51:45] Max Cohen: Why are you saying the light of speed?

[00:51:47] Liz Moorehead: Speed of light.

[00:51:49] George B. Thomas: Oh. Oh, whatever. If that's

[00:51:51] Liz Moorehead: Trust, but verify. We're verifying here. We're here to verify.

[00:51:55] George B. Thomas: Speed of light, light of speed, whatever.