I Tested Grok Bot vs Claude on 12 Daily Use Cases

0:00 There's a battle going on right now between Grockbot and Claude, and they're both competing to get you to use their tool. So, which tool should you actually use? Well, today I'm going to break down

0:08 why these two tools are fundamentally different, tell you which tool to use for the 12 most important professional use cases, and then I'll address an elephant in the room that I guarantee

0:17 you're ignoring. So, part one, you're not picking a model, you're picking who does the work. Grockbot and Claude actually fall into two completely different categories of tools, [music]

0:25 an AI team member and an AI workbench. So, what is the difference? Well, an AI team member is something you hand work to. You ask it to go do something and it goes and does it. Whereas an AI

0:34 workbench, this is something that you work inside of. You sit down and you build the final product in real time. So, Grockbot is an AI team member. And from a product standpoint, this category

0:43 of tools makes everything feel as close to texting a real person as possible. So, with Grockbot specifically, you can name your bots and even drop them in group chats together. It literally feels

0:53 like you're texting your team. And this is great. But just like when you're collaborating with team members, you don't have visibility into what and how they're doing things which limits the

1:01 flexibility. Claude, on the other hand, is an AI workbench. This category is designed to feel like a tool in your hands, not a person that you're speaking with. So you're in control on every

1:09 step. And so with Claude specifically, it runs on your local machine, works directly with your files, and lets you see the entire process step by step. And the thing to consider with these

1:17 workbenches is to get the most out of them, it asks for more upfront because you're not handing something off like to a team member. you're actively building something. And if the distinction isn't

1:25 clear yet, here's a chart with specific features and how these two categories are different. Starting with what computer are they working on. So cloud co-work, right? This can run on your

1:34 local machine whereas Grockbot runs on a shared virtual computer across every bot that you spin up. Do you have visibility into its actions with Claude, you see more of a step-by-step process, and

1:44 whereas with Grockbot, you get minimum visibility where most of it is abstracted away from you. Then who owns the files and conversation history? For this, let's talk about Claude code. So

1:52 Claude Code works directly on your machine with your own files. Whereas Grockbot lives entirely in their ecosystem. That simplifies the experience, but it also locks you into

2:00 their account. So the next time that you're deciding which of these to use, ask yourself, [music] do I want to work on this myself or hand it off entirely? And that's a framework I use to decide

2:09 which tool fits which task. And it's a mental framework that's stuck when I've taught it to other people. And throughout these use cases, I want you to start more thinking about this AI

2:17 team member versus AI workbench concept because that's important going forward no matter what tool you use. So part two, which tool wins each use case? We're going to go through 12 important

2:26 professional use cases today and they fall into three buckets. The first operations, which is your day-to-day, then growth, which is how you expand your role and business. And then the

2:35 third is innovation, which is how you experiment with new things that might eventually become something bigger. And the use cases that we cover get more interesting as we go. But if you can't

2:43 solve the operational task, you'll never actually have time to grow or innovate. So, we're starting with operations. Use case number one is reviewing your inbox and messages. The question that I ask

2:53 myself for this use case is, do I actually want to work through this myself or is this something that I would hand off to a team member? And for most inboxes and messages, I don't want to do

3:02 this myself. So, handing it off to an AI executive assistant just makes sense. So, in this case here, I'd create a new Grockbot, name it executive assistant, and then I'd tell it analyze my inboxes.

3:12 And for me, that's my Slack and my email. And then I'll say, "Flag anything I need to respond to by the end of the day." It then walks through connecting your accounts and sets up a routine to

3:21 run on its own. And for a test like this, I bias towards whatever is the easiest, least friction to set up because I like tweaking it over time. And I like constantly iterating on this.

3:30 Honestly, I've yet to find something that I truly truly love. So, I just like iterating and going by feel. And what Grockbot does is it handles this pretty seamlessly for smaller tasks like this.

3:39 And that's when it starts feeling like magic. At times, Claude can feel a bit overwhelming for something like this. And because Grockbots's connections carry across chats, I can then actually

3:48 create a second bot. I call it video sales manager. And it uses the same Slack connection and email history. I'll then just ask what outstanding deals are currently in the air. And based on how

3:57 past history, estimate how much of that I'll close. That then helps me budget going forward. And then I can make hiring decisions for my team. And this is important, but it's the kind of task

4:06 an executive assistant would handle. It's a nice to have, not necessarily business critical, but it saves me a ton of time in my day. So, for this use case, Grockpot is a winner because I

4:15 actually prefer the abstraction for more straightforward use cases. Use case number two is creating action items and pulling insights from meeting notes. This is a similar question to use case

4:24 number one. Am I working on this or handing it off? So, I personally love recording calls, and it has nothing to do with just creating action items and follow-up messages after the call, but

4:31 instead, it's because I want the context of that call sitting in my system so I can work with AI to improve how I run calls, build products, and solve problems for my customers, as well as

4:41 improve my team's experience. And this process of iterating on the context is something I want to do myself. And so, part of the problem for this specific use case for Grockbot is that it doesn't

4:51 store or save anything on my machine. by design, out of the box, everything lives in their virtual system. Now, Claude, on the other hand, lets me pull meeting notes in through my connections, store

5:00 that context directly on my machine, so it can learn over time, and then I can work through it to pull out the action items that provide value to my business. So, for this use case, Claude wins

5:09 because this is a task I want to stay in the loop on, not handoff. But in order to actually do this, you do need to record the calls themselves, which brings us to today's video sponsor,

5:17 Granola. And if you've been around this channel, you know that I love this tool, which is why I'm so pumped to be working with them. So Granola is an AI notepad that transcribes your meeting straight

5:26 off of your computer's audio. So what that means is that no bot has to join the call. It's just you and the other person and it still records the meeting. This feature alone is what made me first

5:36 start using Granola. But what keeps me using it every day is what happens after the call because the reality is that the meeting is never all that much work. It's the follow-up. It's making sure you

5:46 don't forget anything and things fall through the cracks. And Granola helps me solve these problems through three different ways. The first that I stopped writing action items for my team. So

5:53 after a call, I just share the granola link and every person asks their own questions in the meeting. They'll ask what are the action items. What did Austin say about the deadline? It has

6:02 honestly changed how I run meetings and how I structure them. The second is that the integrations do the follow-up for you. So Granola connects with Slack, Zapier, and Claude Code directly through

6:11 their MCP. So for example, you can finish a meeting, open your laptop, and the recap is already sitting in your team Slack without you having to do anything. And the third is that every

6:20 meeting becomes something you can search and act on. So how this works is literally during calls if a person says something that I'm not sure what it means, I'll ask Granola a question about

6:29 it while I'm still on the call. So for example, I'll be like, "What does MSP stand for?" And it will give me a brief overview instantly so I can feel educated while I'm having the

6:38 conversation. Or after the call, if I totally forget what was discussed, I click into the call on Granola and I ask questions directly there to refresh my brain about what we discussed. So, I'm

6:48 excited to officially partner with Granola and offer my community 100% off their first month. Click the link in the description and try it on your next call. You are going to be happy that you

6:56 used it. Okay, getting to use case number three, completing online tasks for you. So, yesterday a friend of mine had Grockbot negotiate his insurance. It got him 3 months free and then lowered

7:06 his monthly rate, which saved him $500 in 22 minutes. I then tried the same thing with my internet bill. It went back and forth with the customer service rep and I didn't actually end up saving

7:15 anything, but at least now I know I'm not getting ripped off. On prompt is a screen to help you renegotiate any prices, which is pretty powerful. So whether you're using Cloud or Grockbot,

7:24 you should just go and do this. And both the tasks that I mentioned fall squarely into the completing online tasks for you. Going back to the executive assistant question, would I have my

7:32 assistant sit down and argue with a cable company for me? Absolutely. Would I do it myself on a busy day? No. In that case, it makes it a perfect task for an AI teammate. And now, I have used

7:43 Grockpot for other online tasks, and I was impressed. It found me restaurants in Florence, and then it booked a wine tour for me and my now fianceé after we got engaged. Don't tell her that I had

7:53 Grockbot do that, but I did. And so, you can also do this with Claude. So, I wouldn't say this alone is necessary to switch to Grockbot, but I found that Grockbot's just a little faster and the

8:01 interface, the reduced friction, just I find myself doing more things like this. But the reality is that you can do it on Claw. The one thing that Grockbot is unique here is the group chat component

8:10 of it. So let's say you have a grocery shopping bot and then you have a personal assistant trainer bot as the user. You can have the shopping bot talk directly to the trainer bot before

8:19 ordering anything. And that way you'll never actually have to manually update the grocery bot. You could just talk to the trainer and it'll automatically fill in the grocery shopper. This type of

8:28 cross collaboration is where Grockbot has an edge for this task primarily because it's more flexible and the optionality that it comes from. And I'll touch on the ability to invent new use

8:36 cases more later in the video because I think this is so powerful about the interface. Use case number four is creating reports and identifying solutions. So, Grockbot is a winner of

8:45 use case three by a slight edge. Not a massive game changer, but use case number four, creating reports and identifying solutions. This has a clear winner. One thing I foundationally

8:54 believe is that the best users of AI don't just use it to move faster, they use it to make the final product better. And so, sure, you could tell Grockbot, create a monthly report based on

9:03 customer feedback. And that's a great start, but there's really no great collaboration happening here. Grockbot is genuinely strong at aggregating and capturing data, but to actually take

9:13 that raw data, analyze it, and then shape it into something that you're confident in, Claude is the obvious choice here. Back to our core question in this video. Are you working on it or

9:21 are you handing it off? If a report is worth doing, it's worth doing, right? And if you're watching this, you've probably already set up your Claude system to have an extensive LLM

9:29 knowledge base. So, it already has a key context about your day-to-day, your business, and how you like reports structured. And on top of that, you're going to actually want to work through

9:36 the report with the AI as a collaborator, which is exactly what Claude paired with Obsidian gives you. So, a clear preference here for Claude, but Grockbot is great for those quick

9:45 analysis, but anything that deep dive requires collaboration, you're going to need Claude, and that's likely any report that's worth doing. And so, that wraps up operations. And as an operator

9:53 myself, I genuinely do love that stuff, but it's not where the fun starts. That's in the growth and innovation part of your professional life. So bucket two is growth and use case number five is

10:02 finding new customers. This one comes down to two strengths. One, Grockbot's ability to let you experiment quickly with low friction. And two, Claude's ability to help you iterate,

10:11 collaborate, and think deeper. Here's how I think about it. You want to experiment and then harden the flow. And so what I do is I'll run a bunch of loweffort experiments to see if there's

10:20 any real signal through Grockbot. And when there is, I come into Claude and actually build out a real workflow around it or investigate the situation more. So, for example, at one point, I

10:29 had no idea who was actually reading my newsletter. So, I spun up a product growth engineer bot and I had to talk to an audience expert bot to figure out what kind of audience I had and how to

10:38 speak to them directly through our products. That was the experiment phase. Cheap, fast, and pretty low stakes. Once we actually had a direction worth committing to, I moved the whole thing

10:47 into claw to build out and harden the real process and final product. So getting customers can be a whole other video, a whole other deep dive, but high level, here's how you can put it to work

10:57 in two steps. The first is if you don't already have a target audience file, use either tool to interview you and just build it out. Very simple. You can just say interview me to identify my exact

11:06 target audience. And then once you get that, use a follow-up prompt to figure out where you can actually find those customers and how to reach them. And so for getting customers, I give a slight

11:14 nod to Grockbot just because of how fast you can iterate. There's no clear winner here, so you can do it all end to end through Claude if you have to. Use case number six is creating marketing

11:23 content. This is where quality matters more than quantity. It is trivial for any AI tool to pump out infinite content. The real question is which tool actually gets you something that's good.

11:32 And there's a clear nod to Claude here for several reason. The first is you can build far more elaborate context around your voice, your writing style, and past work. then you get better iteration

11:42 experience inside tools you're already collaborating with. Frankly, the underlying model does matter in this case. One problem with Grockbot is Grock. The model is just not that good.

11:51 And in this video, I'm focused on the bigger product differences that won't change much over time, but today this is still a real factor to consider the model behind these actions. So, if

12:01 you're creating content and marketing content, Claude is a clear winner here. And the best way to do this with Claude, there are a couple of lessons that I've learned while growing my YouTube. The

12:09 first is clearly break down each step in the process. It's not write me five short form scripts. It's research, outline, scripting, revisions. Treat each step in the process as its own

12:19 distinct problem. The second is aim for middle to middle, not end to end. You want AI compressing the time each step takes, not removing the step entirely. Now, for what it's worth, content

12:29 systems are hard. So, if you are a founder or operator who wants to work directly with me to help you with this, there's a link below if you are interested in working together. But for

12:36 this use case, the clear winner is claw. Use case number seven is writing sales copy. To understand who wins here, you have to understand how AI works. AI will bias towards the statistical middle of

12:46 what the output should be. And I've talked about why in length in another video which is linked below. But sales copy that lands in the statistical middle does not stand out by definition.

12:57 So you have to actively push AI away from that instinct and away from that average response. and giving the stakes here which is closing new business. This is another clear AI workbench use case.

13:06 And so once you're using these workbenches like Claude, my favorite move here is bringing in actual experts, meaning their books, their podcasts, their videos, all of that, and combining

13:15 it to create training data to steer AI towards sharper, more specific answers instead of generic ones. My go-to here is using Clawed Code with Obsidian and iterating back and forth until I get a

13:26 product that I like. And if you don't want to go through pulling all this information, training your system with experts, my free build partner.AI plug-in can help you do that with

13:34 our/expert advice skill, which will call these experts and give you direct feedback as if you're talking to them. And so this is with your workbench, Claude. If this is with Grockbot,

13:43 working with that kind of dense contextual material just isn't as smooth. Now, once you do start tackling this use case, one thing that I like to use is an abstraction analysis prompt.

13:52 In sales, abstraction kills conversion. So, the goal is to speak clearly and directly enough to move people to buy. And this prompt on screen will analyze your sales copy like a curious

14:02 12-year-old, asking questions at every vague point. Again, abstraction is the enemy of conversion. And so, the more questions you can answer with the copy itself, the better it's going to

14:10 perform. A clear nod here is clawed, specifically using clawed code for non-technical tasks. Use case number eight is creating client proposals. Claude design is my go-to for this one.

14:18 So, a client and I were recently talking about the best ways to build proposals, and he was going back and forth with Claude trying to design them, but the iteration cycle just wasn't working.

14:27 Every small change meant generating the whole thing again and hoping it got it right. And the solution here, which is what I use for all my proposals, is using Claude Design. And there are two

14:36 specific reasons why this is so helpful. The first is you can give inline feedback. So, you click directly on what's wrong and either fix it yourself or tell Claude to fix it. And it knows

14:44 exactly what you're talking about cuz you already pointed it to it. The second is it makes it super easy to make general changes. So you can take an old proposal and build an entirely new one

14:54 using a client's branding. Same structure, same content, but it reformats it around their colors, their fonts, their logo in minutes. And so this is a much more effective AI native

15:03 way to edit proposals than regenerate and prey cycles. You're not waiting on full rebuilding every time you want to make one simple change. And so the reason that Grockbot loses here is it

15:13 just lacks that iterative workspace. So the clear winner is Claude. And so now before we get to the last four use cases, you know the drill. This is our anti-slap agreement. Everything in this

15:21 video is built by humans for humans. So as part of this agreement, subscribe to this content so it can keep reaching more people. I appreciate all of you and all of your support. And as a thank you,

15:30 I give away a CloudMax subscription every single video. This video's winner is Karen Huler EC, who is building a workforce ecosystem for professionals. To enter this video's CloudMax giveaway,

15:40 comment below with the one task you'd hand to a team member or something that you're building. I'll take real and fake answers. Let's have some fun. The next bucket, we've gone through operations.

15:48 We've gone through growth. Now, it's innovation. Use case number nine is researching ideas and competitors. This is where Grockbot's web browsing ability genuinely feels like magic. It's just

15:57 lighter weight for Claude for this specific job and I just use it more for whatever reason. So, what I do is if I see a product I like, I'll drop the link and I'll say analyze this. And I've

16:07 given Grockbot access to our build partner.ai AI GitHub repo and asked based on their funnel and their offering, what features could we add? Then it'll go through and turn that into

16:15 a proposal doc that I can build off of. And I do this a bunch and I've created a product growth bot that's dedicated to this kind of investigation. I never actually found myself doing this inside

16:25 Claude. It's functionally there and I can very easily do it, but it's not something I have time to iterate on manually, which makes it a perfect task to hand off to a teammate. The winner

16:33 here for me is Grockbot. And on screen, if you want to try this yourself, drop in a competitor's link and ask it to break down their landing page, their funnel, and what you should be doing

16:42 that you're not based on this information. And going to why I use Grockbud for this. It's where interfaces genuinely matter to me. Use case number 10 is building an app or internal tool.

16:51 At this point, the answer should be pretty obvious. Claude wins here. This is your workbench. This is where you're iterating, you're collaborating, and you're actually building the thing. Now,

16:59 this doesn't mean that Grockbot has zero value here, and we'll get to that. But first, my favorite and honestly most underused claude feature is artifacts. I was on a run today with an engineer at

17:08 one of the biggest fintech companies and he didn't even know what this was. And so I was like, "Bro, you have to use this." So an artifact is essentially a simple website Claude builds and hosts

17:18 automatically which is sharable with a single prompt. And it can pull live realtime information so it actually stays up to date over time. Like I actually tell everyone about this. So,

17:27 another buddy of mine who I run a small real estate investment group with built a landing page artifact showing our entire portfolio and its performance and and we have ownership across about 200

17:37 units in seven different locations. So, this was quite overdue. We just never got time to do it. Now, this is a real tool that our team will use and it took legitimately zero technical skills

17:46 because Claude abstracts all of the website complexities away from you. So, when you're thinking about artifacts, break it up into two types, single use and longunning. Single use is something

17:55 like a proposal. I'll use claw design for this and I'll turn it into an artifact, send it to the prospect and then it's done. A longunning one is different. You build it once and then

18:03 it's used consistently. So the real estate portfolio one that I mentioned, that's a longunning one. And then another example is a founder I work with, he did this with his restaurant's

18:11 drink menu. So now any bartender can pull up a recipe on the spot and he didn't have to use anything technical. It's just a useful persistent link that his whole team can access. And so that's

18:20 how you can use artifacts which is just another way to build things within the clawed code ecosystem. But what about in Grockbot? Personally, I've used it to curate and analyze customer feedback,

18:29 reading everything and handing me a single summarized doc, which I'll then bring into Claude to build with. I just like how Grock handles pulling and connecting data, but the execution for

18:38 me always happens in Claude. And so, the winner here is Claude. Use case number 11 is automating repetitive work. Grock is built to move fast and take more risks. Claude is built to be more

18:47 careful. Inside Grockbot itself, you'll actually see routines that sits right next to the conversation history. And this is great for building lightweight automations. For anything that I need to

18:55 run longterm, I either have Claude build the automation deterministically, which means it creates code so that it can repeatedly run or I turn it into a Claude skill that I can wire into a

19:05 routine, something that I actually own and I can share with my team as needed. So generally, Grock, again, low friction. Claude is a little more work, but a little more thorough. In this

19:14 case, the reduced friction of Grock creating automations is something that makes a big impact. I'm just running more automations now than I ever used to because the bar for should I even bother

19:25 automate this has dropped so low. So with all that, I'd say this one's a tie. Start in Grockpot, graduate to claude, but you can just do everything in either tool. Use case number 12 is doing things

19:35 you've genuinely never thought about. So this last one isn't really a headto-head. It's more about how Grockbot impacted how I think. What Grockbot provides is an interface where

19:45 you treat agents like teammates. And I don't think that's a small design choice. Grockbot has taken off and people are genuinely like using it that way and we're already seeing it show up

19:53 elsewhere. Meta's new Muse feature has a very similar interface and it's to mimic this team-based collaboration. And the vibe that you get when you use it is what I call the group chat effect.

20:03 Seeing agents interact like teammates can stimulate ideas a plain chat interface never would. So, an example is I recently put two bots in a group chat and had them review a business offer and

20:14 asked them to argue with each other about whether the business idea was actually good. One was instructed to find flaws and the other to find the upside. I wouldn't necessarily have

20:24 thought to do something like this, but the group chat interface made me think about it and after reading the back and forth, it surfaced a couple of objections that I hadn't considered and

20:32 I ended up throwing that into my products landing page. I equate this much like the first time chat GBT came out in 2023. Using Grockbot has genuinely shifted my mental model for

20:42 what use cases can even look like. And so in this video I talked a ton about Grockbot and there's Meta's new product Muse. This genre of tools your AI teammates are worth experimenting with

20:51 in my eyes. And depending on when you watch this Grockbot may cost money. I know Muse right now is entirely free. So just get the creative juices flowing. Spend 30 45 minutes on it. You don't

21:01 have to go crazy. But before we get to the elephant in the room, here is a recap of the different buckets and how you should think about it. Operations, Grockbot for the quick lowstake stuff,

21:09 while clawed for anything you want to actually collaborate on and get right. For growth, Grockbot for the fast experiments, clawed for building the real things once you know it works. And

21:18 then for innovation, Grockbot for wide, messy discovery, clawed for actually building what comes out of it. But if you walk away from this video just thinking about the 12 use cases, you're

21:26 going to miss what I think is the most important, which is part three, the elephant in the room. I guarantee you're ignoring. You just sat through 12 use cases and you're thinking about which

21:34 tool to use. But the tool isn't the limiting factor anymore. It's you. Yes, you. The last use case I talked about was the importance of stimulating your creativity as you use these tools. The

21:46 process of becoming more productive with these tools is as much creative as it is technical. The creative claw code Boris Churnney has a great quote. Give the model a task that feels harder than you

21:56 think it can do. And reading between the lines here. He's saying you need to be more creative with how you use these tools and what you use them for. That's the part I think a lot of you watching

22:05 haven't actually sat with. Grockbot specifically is built to stretch that creative muscle because the friction of trying something weird is basically zero. Naming a bot, dropping it in a

22:15 group chat with another one, seeing what happens, that's a creative act, not a technical one. And the people watching this, I know you guys are using Claude, like 95% of you. And there's no reason

22:25 you can't flex this creativity with Claude. You just have to be aware of this and start flexing the muscle. So, for either tool, here's a prompt to get you thinking differently about how and

22:34 where to use AI based on your past conversation history. The problem isn't necessarily about which tool to reach for. It's about knowing what to do with either one. And if you want a full

22:43 playbook on that, go check out this video where I walk through exactly how to build 10 times faster with Claude. I'll see you in the next video.