183: Why Trusted Data is the New AI Moat (w/ Rick Kranz @ AI Marketing Automation Lab)

Databox (00:01.277) Hello and welcome back to another episode of Move the Needle, Databox's podcast about how companies are moving the needle, improving some metric, aesthetically growing their business in some way. Today I have a special guest, longtime collaborator and friend, Rick Krantz. Rick is a three-time founder, two-time founder, two-times exit entrepreneur.

Built a few businesses and sold them, in other words. Rick Kranz (00:26.338) Thank Databox (00:31.293) And he started the AI marketing automation lab 18 months ago, two years ago, Rick, when did you start it?

Rick Kranz (00:38.04) Yeah, it's about 18 months ago, Yeah, 18 months ago, exactly. Databox (00:42.983) Cool, so you've done a variety of stuff.

We've known each other for the better part of, what is it, 15 plus years, maybe a little more. Rick Kranz (00:49.774) It's late 2010 we met on the phone, late 2010. Databox (00:55.

037) So yeah, so 16 years. And we've been collaborating ever since on and off a little here and there. you went out, you had a agency before this. Before that, you had a manufacturing company that you ran out and ran.

You sold that. And then you started an agency. That's when we met. You did that for a long time.

Then you sold the agency to another agency, right? Worked there for a little while. And then you saw the AI stuff happen in and you're like, shit, this is my window. I'm not, I'm not so old that I can't figure this out.

And I'm old enough to know that this is a big deal. So you kind of went out on your own, right? And again, and started something up. Rick Kranz (01:37.

794) Yeah, yeah, you went out with my my daughter and I actually so we're partners. Yeah, we partners this Databox (01:42.095) Okay, cool. Nice.

That's cool. Rick Kranz (01:46.638) Yeah, so it's been exciting. It's been interesting.

Databox (01:51.901) Tell us a little bit about what you're doing there at the AI Marketing Animation Lab. What's the business model? Rick Kranz (01:58.

946) Yeah, so the business model, we have two ways we work. One is the primary one where we started is the community. And so we have a community that is basically, it's for implementers. So if you want to be the person in your organization who can say to leadership, I've looked at this process we're doing, here's how we can totally automate it, here's what it'll cost, here's how long it'll take, here's what it means to maintain it.

And here's what the impact will be on the bottom line. If you want to be that person inside your organization, then this is what we create inside of our lab. We teach people to do this. And every lab member that we've had has actually gotten to that level now, where they can look at their system.

And they are. And they're remodeling their systems inside of their own organizations. And that's what we focus on. Very hands-on training.

You're not watching any videos. We're building systems with you. Databox (02:51.836) Yeah.

So you run sessions each week and from what you've done in the past, you're doing like multiple tracks almost. It's like you're running a little university over there where you have like different courses or tracks. And then each week you're building a component of something in each track. So like each track is designed to launch something, right?

So I don't know, maybe walk us through some of the things that... like what those tracks are, courses are, what is the systems that your team or your members have built. Rick Kranz (03:28.098) Yeah, so a big one that everyone really wants to build is the AIO system, or it could be AEO or GEO, depending on where you're coming from, Yeah.

Databox (03:37.477) Right. Getting excited and being mentioned in LLMs and Claude, Anthropic, Gemini, etc. Rick Kranz (03:45.

528) Yeah, yeah. it really takes that job that would be almost humanly impossible to coordinate all that information and create enough content at scale to be cited and recommended inside of these LLMs when someone's asking for recommendation, especially in the B2B space. This system does it. And our members have taken this and customized it, like even one of our members.

Keith from Mojility, he's turned it into a whole service that he's providing for their clients, their HubSpot partner. And he's been transparently providing it for them. They know what it is and how it works. And they've been using it for, I guess, eight months now.

And now he's turned it into a monthly service that you can just rent it and use it. So they're taking what we start. Databox (04:24.497) Okay.

Databox (04:42.231) He's taken like something that you guys helped him build and he's almost like licensed it to a client. The client uses it on their own and pays him a monthly fee. So you almost, you helped him build and launch a product of some sort.

Yeah. Rick Kranz (04:45.667) Mm-hmm. Rick Kranz (04:58.

38) Yes. Yeah. And, and the, but the beauty thing about it is that he's added so much more of his own stuff to it, right? implemented our rag system.

So we teach everybody how to build a rag system and, it's, Databox (05:14.676) For those of those those of listeners that aren't familiar with that demystify what a rag system is for us. Rick Kranz (05:21.646) Yeah, I won't I'm not going to get into the the acronym itself but Just what it is.

It's a way to store your proprietary information in a vector database and a vector database is just taking unstructured data like words and and PDFs and stuff like that in and Turning them into numbers so they can be retrieved instantaneously in a semantic fashion Databox (05:26.396) Yeah. Databox (05:42.032) it is Databox (05:48.

989) Yeah. Rick Kranz (05:51.084) Semantic meaning when you ask a question, it's not looking at the keywords in your question, which you can easily do with a database. It's actually understanding the context of what you're asking and then retrieves that stuff.

in. Databox (06:00.861) Right. Yeah.

So if you connect your RAG system, just to give people a practical application, connect your RAG system to say, Claude or NIN or something like that. Rick Kranz (06:13.964) Yeah, in this particular instance, we built this on make.com and Airtable, but it could just as easily been like you built yours, right, on N8n and Airtable and some other database systems.

And then it could pull from the Rack system. So when you're writing about a specific topic, it goes in and pulls from the Rack system. So the AI, which is doing the writing and the editing and the researching, is told to only Databox (06:21.701) OK?

Yeah. Rick Kranz (06:43.802) use information that is inside the system. So it's not using anything that's trained on.

It's not using information that the internet already has. It's using your proprietary data. Databox (06:54.907) Right.

So when creating content, right. It's important that marketers create content that's unique and also relevant to their customer and their business. Right. And so the way that a rag system enables that is you can store all the content or source material that you create, like call transcripts or point of view perspective from your internal team.

Right. Like Rick Kranz (07:03.032) Mm-hmm. Databox (07:20.

113) And so you can store that in that rag system, vector database, whatever. And then the AI can pull from that when it's creating content so that the content ends up being unique and appropriate for your audience and your. Rick Kranz (07:30.668) Yes.

Yeah. You can even store all your existing blog posts. We store all our images. You can store, if you're an agency, you can store all of your clients images in a Rack system.

Yeah. So if you're looking for something specific, Databox (07:43.015) Yep. You're basically pulling from everything you have, pulling from raw material that you might have collected.

Rick Kranz (07:50.318) Yeah, and you can make sure you're not rewriting something you already have that's performing well. So yeah, it can get pretty crazy. But the point is, this is what's going on inside of our community.

We start, here's a system, teach them how to build it, and then they take it to a whole other level, which is fun to watch. It's fun to watch people develop and their eyes open like, wow. Databox (07:54.183) Yep.

Yep. Yeah. Databox (08:11.485) Yeah.

You've been doing it for a while and I know you have a number of members in your community and as you alluded to, we are a member of your community. A member of our marketing team, Nevena, is in your community and she built what you're talking about. She's taken that to the next level where we're even building out even more sophisticated stuff that allows us to produce content across mediums, different content types like... an opinion piece versus a how-to article, a video script, and then across channels.

like we're, we're, and we have, you know, RAG databases set up with our, with our existing content or our perspectives on the market, how we position our product, who we're selling to, all our call transcripts and customer communications, videos that our partners create that show off how they use our products. So like we have all that stuff organized. So Rick Kranz (08:44.366) Mm-hmm.

Databox (09:10.779) when we want to create something new, it's not hard, right? And we can repurpose content very quickly. So yeah, it's been helpful for us.

But that's not it. Like there's how many different systems or skills do you have you help people build? Rick Kranz (09:18.39) Or am I lagging or are you lagging?

Databox (10:29.18) Hey, you there? Yeah, it looks like it's still recording. So we're just going to keep going.

We'll edit that part out if there was. Rick Kranz (10:32.514) You back? Yeah.

Rick Kranz (10:39.032) Yeah, they could just edit that out. Yeah. I would say adding on to what you were saying is you guys have taken like 10 levels above what Nirvana's done.

Yeah. It's really cool. We actually have gone back and now adding more to bringing context out of our client meetings and bringing that into some of this, basically our lab session. Databox (10:48.

613) really? Yeah. Databox (11:05.396) awesome.

Rick Kranz (11:07.65) bringing examples of what's going on in our lab sessions. It's extracting it and bringing them into the content creation. Databox (11:13.

306) Okay. Got it. you it's like, content is the way I look at it as content is kind of a by-product of, doing work, serving customers. Right.

So that's what you've figured out. Cause you have a community where they're building things and more than willing to talk about it, for the most part. And so you can just literally take your sessions that you're running and feed that into your own content machine and output content based off of that. Yeah.

Rick Kranz (11:40.482) Yeah, yeah, that came from your brain because you said to me, write about what you're Databox (11:43.804) Thank Databox (11:47.484) Yeah, it's such a foreign concept for people.

I think people try to write to convince other people to do something different and just like write about what you're doing. It's like people are like, I wanted that, right? Rick Kranz (12:00.842) Yeah, I just, you know, it didn't click for me until you hit me over the head with it.

I think probably the 20th time. It's like, wow, that makes sense. Databox (12:04.55) You Gotcha.

I actually just this morning, just like did a session with our sales team and I taught them, I showed them the simple skill that I have built in Claude that, they're using, some of them were already doing, but now the rest of them are going to use the skill that I built that pulls data from all of, not just the content we've created, but the performance of that content. And then it helps them. Rick Kranz (12:19.864) Mm-hmm.

Databox (12:37.02) Right, so it walks them through like a five step process where they pick the angle and the hook and then it like looks at all the content I've written on LinkedIn and says like that angle should work based on these other articles or these other posts because it's similar to that and then here's the hook that might work etc and like they approve the angle then they approve the hook then it drafts the outline, once it's addressed the outline and then it drafts the post.

Rick Kranz (12:40.195) Nice. Databox (13:04.582) and they can edit and give feedback on each step.

So it makes the content creation easy for them, but also allows them to like tell their own story because the input to it is a call transcript or two call transcripts or a call transcript and an article that the marketing team might've written or something. they, but they, because they're kind of approving each step, it can, it's still coming from their source material, their calls and their brain. Rick Kranz (13:19.232) Yeah.

Rick Kranz (13:35.311) Yeah, well, that's, mean, that's the cool thing about that is because you, you get off of call and you've been on a call with a prospect or a client and, they usually, there may be actually some testimonials in that call that just, that just went by you because you're focused on helping them with something else. And there may be some story moments in that call of how you help them, but you're not thinking about that. Like, why would you be thinking about that?

You know, that's not where you're at on the call for. Databox (13:54.682) Right. Databox (14:02.

608) Right. Yeah. Rick Kranz (14:04.514) But then all of sudden, that transtur, instead of getting the generic summary that you get from Fireflies and one of these other things, like here's what it said, here's what you have to do.

No, we built a system that takes that and finds the testimonials now, grades them, and then finds the story moments. Databox (14:27.174) So you're actually like, you have an agent sort of that's like browsing through calls that you have and it's looking for those things. Rick Kranz (14:31.

929) Mm-hmm. Yeah. Yeah, they all get built out into an air table and graded. then we have a Claude skill that I work with.

Now, the Claude skill I work with every morning to uncover content ideas goes through all the emails and transcripts and anything I read that I thought was worthwhile. Now is told to lean 70 % towards what we're actually doing inside the lab, what other people are doing, like what other people are accomplishing. And it's taking those moments now. And you see my last two LinkedIn posts were actually.

Databox (15:02.716) OK. Yeah. Databox (15:08.

444) Yeah, I did. noticed. Yeah. You talked about Keith, actually, right?

The Keith. Rick Kranz (15:11.662) Yeah, yeah, and that comes from that surfacing that for me. It's like, okay, even if you if I if I was a great writer, which I'm not, would say, okay, I'm to write this, but I still wouldn't have the the information, right?

So the system actually uncovers it for me. The first thing it does is it uncovers this is this is what you should write about. And I have a choice that I could just write it myself or let it write it for me and then edit it. So yeah.

Databox (15:15.398) Yeah. Right. Databox (15:36.

016) Right. Yeah, but the beauty of it is not only did you not have to think about it or plan it, you just had to look at what it produced for you and then take it to the next step and finish it. But it's just a byproduct of the work you're doing. So that's cool part.

Yeah, so nice. So we're not quite at automated content production, but you've automated, I think, maybe the hard parts or the time consuming parts. How's that? Rick Kranz (16:02.

698) Yeah, I don't believe in automating content production for LinkedIn. And that's the primary social channel we're on. think, you know, if you're engaging in social media, you're writing for people and your comments should be coming from you. I never use AI for commenting.

always, you can tell my comments because they're poorly written. But they come from the heart, right? And they come at the moment. And if my LinkedIn posts, would...

Databox (16:05.348) Right. Of course. Databox (16:15.

547) Yep. Databox (16:21.372) Yeah. Rick Kranz (16:32.

43) Mostly I would say 50 % me 50 % AI surfacing stuff. I just think this is my philosophy When it comes to blogging now and you're writing for You're writing for an algorithm or now you're writing for AI. So I feel like machine-to-machine is going to do a better job with that Not in the B2C space though if you're B2C Databox (16:39.461) Yeah.

Databox (16:48.859) Yeah. Gotcha. So when you're on social media, you're not automating completely.

What you're saying, though, is maybe for SEO, where you're writing for the machines to consider content, it's probably OK, as long as the content is still unique. Rick Kranz (16:59.704) No. Rick Kranz (17:07.

862) Mm-hmm. Yeah. Rick Kranz (17:13.196) Right.

And that's the thing where, yeah, you got to getting the content. We did a whole webinar together on how to create unique content for AI, right? And it's still relevant. It still works that way.

Whether you do it by hand or use AI for it, it's still relevant. The other thing we do, AI Lab, is actually build these. We build custom systems for organizations. we're companies.

We work mostly with Databox (17:15.76) Yeah. Databox (17:22.331) Right, exactly.

Databox (17:32.54) Yeah. Rick Kranz (17:42.209) Small and mid-size companies, but companies that don't have somebody to send to the lab to learn how to do this, we do that for them.

We come in, look at the systems, pick one that we think would be the best to start with, give them the budget, set it up, train them, and give them maintenance and access to the school if they want to join the school. And then they own it. They own everything. It's no...

Databox (17:48.774) Yep. Databox (18:08.782) Right, yeah, you're helping them build the system, but then they own and they maintain and they operate.

Rick Kranz (18:13.484) Yeah, it's really just an extension of what I was doing as sales enablement work. Databox (18:18.429) Okay.

Yeah. So in your previous agency, you were doing sales and even at work where it was more build the system and then they run it. Right. Yeah.

Um, and so you have this dual model. You have this like membership thing where you're just showing them, they're showing up to classes. Um, you're showing them how to do it. They're going off and building it.

Um, and then you have the more of a, do it with you do it for you, uh, model, which is more of a consulting model. What, what's interesting about your business is. Rick Kranz (18:42.264) Mm-hmm.

Databox (18:48.092) different than I think I've seen other AI implementation firms go. Oftentimes, the ones that I've seen most, it's like they have a product that they built and they're like, we'll implement this thing for you. And, you know, it's more of they'll do the implementation and even manage it on an ongoing basis for them, which I think is a good model.

And then I've seen the model of like, Rick Kranz (19:04.504) Mm-hmm. Databox (19:17.144) If we are an AI implementation firm and we're going to come in and like assess everything that's going on in your business, identify all your functions, your roles, your processes, and then we'll sit there and say like, which one should we AI enable first?

And then they'll build custom solutions. So I think what's interesting is you're kind of meeting the meeting and you're in the middle a little bit where of those two solutions where you're not. It's not like you have it completely productized, but you do have a bunch of systems that you built. that you've taught people how to build that you can help people get up and running with quickly when they join your community, but then you also are doing, helping them implement those.

But you're not going so far as like, let's AI everything in your business. You're stopping short of that and staying kind of your lane from marketing and sales use cases, it seems. Rick Kranz (20:05.294) No.

Yeah. No. No. Yeah.

That's never been my philosophy. Even when I was doing sales enablement, I was more about, me tell you what not to do. Let's first knock out the stuff you've been doing, and let's stop doing that, and that money will go right to your bottom line. And here's what you can do, and find the one thing that's going to move the needle first.

Yeah. Don't do everything else. Databox (20:16.969) Yeah.

Yeah. Databox (20:30.076) Right. It's probably safe to say that you're a believer in implementing AI.

But when you're talking to someone that's maybe a little skeptical, they're not as knee deep or neck deep in it as you. They haven't built and launched systems yet. Maybe they're using the Claude or Chat GPT interface, but they haven't built skills. They haven't built automations.

They haven't automated a string of work. Right? What do you say to them? How do you, how do you account of them?

How do you use what you've done already to like say, hey, here's what's working for other people. Rick Kranz (21:10.572) I just, because we have, we built so many systems, I could just, I just share my screen and just show them here's what we built and here, let's give me something and I'll throw it in and let's see what comes out. So it could be a blogging system.

It could be the reg system. It could be the persona table that we built out. These could be one of the systems that we built out with data box MCPs. And so it basically, I just, it's a show and tell if Databox (21:40.

358) Okay. Rick Kranz (21:40.407) someone if they don't, you know, if they don't know what most don't know, like what could be done or what would just look like. So just show them.

I had like a painting contract company said I was thinking of hiring. We're we're exploding and I can't handle the work. The owner said to me, I was thinking of hiring an assistant, but I can't afford assistant for more than Databox (21:46.545) Right.

Databox (21:56.636) Yeah. Rick Kranz (22:08.878) three days a week, someone told me I should talk to you.

You can create an AI assistant for me. So as I did it, I put a demo together. I created an AI voice assistant for him and let him start talking to it. And it was able to generate quotes while he's at the site for his prospects.

He could just talk to it. He could just stream a consciousness to talk to it. And then it shot him over a QuickBooks, estimated using QuickBooks. Databox (22:13.

242) Okay. Databox (22:29.423) Right. Databox (22:37.

884) Okay. Rick Kranz (22:38.254) And he's like, okay, this is crazy, insane. it's like.

Databox (22:40.967) Cause he was probably going back to the office, right? Trying to remember what he took some pictures, right? Like, yeah.

Rick Kranz (22:45.986) Well, yeah, right. And that's, just, yeah. And he could just walk through the house and because what it's doing, it's not, it's just gathering the information and passing it off to another agent.

Like we're using Claude as a second agent to go through that stream of consciousness and organize everything. And then, and then through MCP, it's connecting to QuickBooks and creating the quote and then sending it back to him. So. It's the stuff that you can do is, yeah, it's a basic promise is just demo.

know, sometimes it's just takes. Databox (23:17.936) Yeah. So proof of concept is often easier.

I'm guessing you're not trying to convert skeptics, but you're trying to just educate people that may not know what's possible. Rick Kranz (23:28.63) Yeah, I learned a lot of time ago, don't do the two step sales process. HubSpot was good at that.

You guys had to do that the early days, first convince people that inbound was a thing, and then that you need the software to do it. Databox (23:38.714) Yes. And show them the ROI, the calculation and all that stuff.

But you don't have to do that. You're not doing that. In that case, like you just, save that guy. Like you got an introduction, of course, but he shared with you that he was trying to think how to afford an assistant.

And you just showed him though, he doesn't need an assistant at least on things. Yeah. Yeah. Yeah.

Rick Kranz (23:47.832) Try not to, no. Rick Kranz (24:01.462) Right.

Yeah. I mean, yeah. Not for that. If you're just going to have someone that's going to answer the phone for you, yeah.

Databox (24:09.692) Yeah. But the ROI is obvious in that case. So how many systems would you think do you think you've inspired or like your members and everything have gotten live?

Like what do you think? What do you think your impact's been in the 18 months? Rick Kranz (24:12.194) when you need to.

Yeah. Rick Kranz (24:30.444) Well, our community is still small, even though we're capping it at 60 members so that it could always be hands-on. And we have small sessions.

do three to five a week, depending on who's around and what we're building. And we still, I would say we're still only half full. But I think looking at the impact, there's some, I would say there's... Databox (24:36.

112) Okay. Databox (24:50.084) Okay. Databox (24:56.

316) There's got to be hundreds of systems live as a result of your effort. Rick Kranz (24:59.456) Yeah, yeah, there's definitely, I just in the lab, we have over 100 systems that we built and some of them I've retired like, you know, things in the beginning were like, we did some things that were cool. But the stuff that actually works is still, we probably have 50 automations running at any given time in my own business.

Yeah. Databox (25:05.596) Okay. Databox (25:09.

852) Right. Yeah. Databox (25:22.244) In your own business, you mean?

Okay. And those are all things that you've made available to some degree, at least to your... Rick Kranz (25:29.1) Yeah, they're all available.

If you're in the community, they're all available in our platform. You just go in, you can download them. Databox (25:37.308) And you just charge like a few hundred bucks a month for the membership, right?

Per person. Yeah. Rick Kranz (25:41.472) Yes.

Yes. Yes. It's yeah, it's like 345 a month, but it's it's really hands on. mean, nobody seems to leave.

We it's. Databox (25:50.874) Yeah. But they they get like access to all the old skills.

and like, once they, they probably go through a few sessions with you, they probably learn how these things are built so that they can then go and like, intelligently pull a skill down and implement it effectively. Right. Like, yeah. Okay.

Yeah. They're tailoring it or customize. Rick Kranz (26:10.412) Yeah, and then they build their own.

That's crazy stuff that I wouldn't even think of. that's what I hear most, biggest value, people are saying the biggest value that they're getting out of the community is the community when we meet. When we have a session like this, we're in a Zoom meeting and maybe there's five or eight of us at a time. It's the ideas coming from everyone that's been building and helping make recommendations for with someone else.

Databox (26:24.251) Yeah. Rick Kranz (26:38.966) could build to solve the problem that they're trying to solve.

Databox (26:41.564) Yep. Gotcha. Okay.

Cool. Walk us through your changing subjects a little bit. Walk us through your preferred AI stack. A term I've been thinking about using more, AI stack.

Yeah. It's like we have marketing stacks and sales tax, but I think there's, I think there's a little new AI stack that's forming. It's really changing the way software not just gets built, but like the way you kind of assemble your Rick Kranz (26:54.859) A.

I. Rick Kranz (27:07.83) Yeah, so right now, and it evolves constantly, right? Right now, my favorite AI stack is Airtable as a simplistic database, visually, for holding information.

And you can store images or text or data for processing, right? And then Databox (27:19.824) Okay. Rick Kranz (27:34.

969) You could put the reprocessed stuff back into Airtable. For automation, it's Make. Early on, I looked at N8n and Make, and I felt that in the early days, Make had better guardrails for ensuring that it did exactly what you wanted it to do and nothing more. In the early days, N8n was all about giving one AI agent the opportunity to do whatever it wanted and send stuff around, was cool.

If you're a hobbyist and you like building it. But I'm building systems that have to be 100 % reliable for businesses. And that's changed. And you guys know that because you guys use it.

So I'm not saying don't use that. just stop. We made a choice. We've done a lot of making that.

Databox (28:17.318) Okay. Yeah. Databox (28:25.

213) Right. You build a lot on make and I think this point make does most of what NNN does at this point. Have they caught up with like calling LLMs and returning results from LLMs and all that? Yeah.

It did. Okay. Got it. Rick Kranz (28:40.

622) Make? I did that in the early days. So what I like about it is everything is easy to plug in and plug out. So most of our automations run several different LLMs in one automation.

And then if something changes, we just pull it out. We're having problems in the early days of Claude. It was getting overrun. The API was getting overrun a lot.

And it would stop. So we just pull it out, throw Gemini in. That's a 30 second. Switch out so it's basically just pulling things out plugging things in so it's easy to update keep updated I'm sure any end is is just as good at that, Databox (29:19.

556) Yeah, I thought that was what their like original thing is. Maybe, maybe may kind of before, but the other difference, right. And you tell me if I'm wrong, but in NNN you can kind of write code. I think developers like NNN a little more, gives them a little more flexibility versus.

Yeah. Rick Kranz (29:31.503) Mm hmm. Yeah, and I'm not an expert.

I don't have real experience, but that's what I hear that. Yeah, it's a lot easier. Also, and then you could just tell Claude to write it for you. You don't even have to drag and drop anymore.

You could just so you could just call Claude to write in it'll write. It'll write the code, the JSON. I believe it's in JSON and just upload it. Yeah, so so make an 8N and and now Claude Cowork.

Databox (29:45.757) I see. Write the code and plug it in there. Yep.

Yep. Rick Kranz (29:59.371) is a big part of what we do. Data box MCP has become part of the tech stack that we love.

Yeah, I'll dive into more on why that. Pine cone for vectorization. Yeah. Databox (30:04.

368) Woohoo! Databox (30:10.49) Okay, yeah. Databox (30:15.

63) Okay. Does pine cone help you vectorize this stuff for air table or is it like a separate thing? Rick Kranz (30:23.298) So PyCon's really storage.

For the vectorizing, we're using the LLMs. And we use IBM's advanced system, the Dockling. So everybody in the lab was probably one of the most complicated sessions that we ran. It took about six or eight sessions.

Everybody built out the Dockling system that IBM open sourced, which is an amazing system for. Databox (30:39.568) Yeah. Okay.

Databox (30:47.44) Okay. Rick Kranz (30:50.99) RAG system for vectorizing.

It's a hybrid chunking and vectorizing system. So it's really good at figuring out the context and where to cut things off. So if you know anything about vectorizing, it's good. I won't get into it.

It's a little weird. basically, it's the difference between most vectorized systems will just, if you thought of it as like, OK, I got this book of 500 pages. Databox (31:05.916) enough.

I don't. Rick Kranz (31:20.482) And it's going to take each page and rip it off equally. It doesn't care what's on the page.

It's OK, this is one vector, this page, page two is another vector. It doesn't care. The docking system will actually understand the page. And when the thought is complete, it rips it apart.

And then it leaves the other thought. And so the rest of that page may belong to the next three pages. And it'll keep those together. Databox (31:25.

199) Right. Databox (31:35.568) That's when. Databox (31:43.

58) So it kind of does the reasoning or the thought process of an LLM in order to do the vectorization. Rick Kranz (31:49.902) Yes. Yeah.

Yeah, I think that's our tech stack. Databox (31:56.804) Yeah. So more lately, like you've been publishing skills free and you created, like I obviously know the ones that you created that levers the MCP.

I don't know if you created other ones, but you've been offering these Claude skills. Like they all kind of have a theme around the ones that I've seen around doing data analysis, of course, with the MCP server. What's your thought process? Why are you, why are you, Rick Kranz (32:02.

153) Yeah. Databox (32:24.273) giving them away for free instead of like saying, you gotta join the membership to get them. Rick Kranz (32:30.

286) I Think it's Yeah, what is what is the thought process around that? I always felt like you you need to give some value first, right and in the early days Well, we used to get we used to create what PDFs and give them away Yeah, and and the dirty secret that we never told anybody in the marketing world is that nobody reads ebooks they just just like them and keep them on their hard drive and they feel like Databox (32:34.46) Thank Databox (32:40.741) Yeah.

Databox (32:45.594) Yeah, E-box, right. Yeah. Databox (32:50.

491) Yeah. Databox (32:54.008) I don't know. I'll download it or download it read it to Maya and they'll never do.

Rick Kranz (32:59.726) It checked off a box and you yeah, you felt good about it. Okay downloaded that that's good. I'll read that tomorrow But now we're giving away these skills It's it's great to hear feedback from people that like I installed that and it works and it's like wow that is that is is cool and I didn't think about it before I didn't think about doing skills before until You guys released your MCP the data Databox (33:07.

665) Bye. Databox (33:15.451) Yep. Yeah.

Databox (33:28.046) Okay. Yeah. Rick Kranz (33:29.

486) Because it's like The one thing I try to look at AI, like how to use AI, I don't. I don't teach people to use AI to do more of what they're already doing or do it faster. We try to look at using AI to do something that wasn't possible before. And I don't mean it was impossible in the sense that it was impossible.

was impossible because it was impractical cost-wise or time constraint-wise. certain Databox (33:45.647) Okay. Databox (34:02.

554) what was impractical time constraint wise. Rick Kranz (34:05.654) whatever you want to automate with AI. So I'll give you an example.

Data analysis, a full data analysis of looking at all your different sources of data coming in. Most companies didn't have a data scientist or a data analysis nerd that was really into that, especially small and mid-sized businesses. Databox (34:08.389) Okay.

Databox (34:32.732) We're not all nerds, Rick, but okay, keep going. Rick Kranz (34:35.212) Well, it's not a bad thing.

mean, I'm a nerd. Yeah, it's like fine. I didn't mean in a bad way. Yeah.

I mean, in a good way. So we don't have all those brainiacs that want to sit and do that. and the time and even if they did, you know, it could take days to figure some of this stuff out to really cross reference everything. So that is something that was kind of impractical for a lot of companies to do.

Databox (34:37.848) I'm joking. Databox (34:48.804) Right.

Or have the time to do it. Databox (34:57.734) Yeah. Rick Kranz (35:04.

12) But the information was there. the early days, DataBox made it easy to gather this information and create charts from it. you could, through DataBox, which was brilliant, you can get your data from Facebook ads, from Google ads, you can get your data from. Google Analytics, you can get your data from Google Console, you can get your data from your CRM, like what are the conversions, what are the visits.

You can get all of that into one hub where the data is, I use a word, it's probably not the correct word, but normalized, meaning it all matches in a sense. Yeah, so if you're looking, I want to look at this week, you don't have to worry about like, what's the week begin for that one? Data box handles all of that and you could create these charts. So you can create charts and it's like, okay, now which charts do I create?

And then you create all these charts and to a lesser extent, you could do some of that with CRM systems, but to a much lesser extent, data box is pretty comprehensive. And we used to build them for clients. And the problem I saw was that, okay, you build them to get excited about them, but they don't really use them in the sense that looking at it, and then there's some obvious things. some traffic and certain things, and ads, and ad words, and stuff like that becomes pretty obvious.

But some of the stuff is not obvious. So when you release the MCP, which if anyone doesn't know, listening to this, it's model context protocol. Basically, it's the equivalent of a USB cable between Databox (36:28.477) Mm-hmm.

Yeah. Databox (36:40.305) Yeah. Databox (36:53.

638) Sure. Rick Kranz (36:55.116) Software systems, right? So you can you know, you can take your USB-C cable and connect it to anything that has a USB-C and it'll work But that's what MCP does.

So now with the data box MCP we can connect it to Claude Cowork and build a skill to be that analyst To be that brainiac, right and it's incredible what What can be done stuff that just couldn't be done in the past cross referencing all sorts of information together and getting advice from the AI like so they one of the one of the favorite things that we built and released is called sales pulse pulse. Now what I worked with a lot of salespeople built a lot of sales dashboards inside of HubSpot for salespeople and they really liked it.

But did they really use it? No. Databox (37:38.812) Okay.

Rick Kranz (37:50.667) Some did sometimes a sales manager sometimes the ceo would look at it, right? And it was just it was too many dashboards like too many too many graphs and stuff like and Databox (38:02.813) Right, yeah, too much to interpret.

What I find is that what we found, and this has been true since I've been selling dashboard and reporting software for the last nine years, right, is that most people use dashboards as a report card. And so like at that point, it's over and there's maybe not much you can do. Maybe you can change going forward, but like if you only look at the dashboard thoroughly or the dashboards thoroughly once a month, Rick Kranz (38:21.176) Mm-hmm.

Databox (38:31.632) then you're missing out on opportunities and issues until they've kind of, you might've missed the window for the opportunity or let the issue drag on and cause more adverse effects. so, the exact problem you named before is like people like to really analyze, you got to look at a lot of charts. You got to see like, how does this metric impact this metric?

this is down. Why is that metric down? I got to break that down by multiple dimensions and see. what might be down on, you know, that's contributing to that bigger and more important number being down.

And so like the amount of data you have to look at, remember, keep track of, understand and understand the correlation cause and effect between like that's a lot. It's for most humans, it's not feasible. I have an analyst here at DataBox, full-time analyst, and actually I pulled together a meeting this morning between him and a manager, because he was analyzing something over here and coming to a different conclusion than the person leading the function. And I'm like, you guys gotta talk and figure out why you're coming up with two different conclusions.

And it's because the analyst was missing context, right? And so that's the other thing that's often missing. You gotta have that context of what's actually happening that's maybe not measured in the numbers in order to filter these numbers. And so for a human to do all that, be good at the analysis.

of the data and also understand all the context, it's really hard. But the sales pulse is a perfect example. Your sales pulse skill is a perfect example of like, you don't need to look at all that data because the AI can step through all that data. And because you've given instructions on how to do that, it can make good assumptions or good, not assumptions, good conclusions from.

Rick Kranz (40:22.808) Yeah, exactly. the human's context window, meaning how much information can we retain at one time, is minuscule compared to the context window of, say, Claude. Databox (40:28.

155) Yeah. Databox (40:31.451) Right. Databox (40:35.

854) Right, especially if you start feeding it context and storing context about the situation. Rick Kranz (40:39.384) Yeah, yeah. it's yeah, we're really good at creating the stuff that we can do that I find AI has no ability to do when it comes to like coming up with creative ideas or strategies.

AI is still a little like weak on that stuff. It's just repeats stuff that's been done already. But so in sales pulse, I wish I had this when I was doing sales enablement, because I would have installed this for every sales person. What this does is it doesn't give you a Databox (40:57.

915) Night. Rick Kranz (41:09.334) chart right away. What it does, gives you information.

tells you, so it analyzes 14 different metrics coming from your CRM, all through a data box. So data box pulls these together and we pull that metrics. goes into a Claude skill and Claude analyzes. It looks for the trends.

It looks for what's happening. will analyze like, when is your pipeline going to dry up if there's a problem? and what's causing that, is the activity going on, like how many emails are going out, how many meetings have been had, it analyzed everything that's happening and then it talks to the salesperson and shows them like this is what's happening and then the salesperson has the ability to dig down further with it. So it'll ask, would you like to see a chart on this specific thing?

And the salesperson could say yes or no. It could create a different chart. So the salesperson is actually actively, interactively creating charts to drill down and see what is going on that Claude is telling him that's going on. So I think that's a really good Databox (42:13.

436) I'm giving some kind of corrective action like, hey, your close rate is down. Rick Kranz (42:20.738) Yeah. It's like, hey, you suck.

Pick up the phone and call. Databox (42:23.921) Yeah, or your activity is too low or your deal size is off, right? Like it's going to look at those things and say, here's where the issue is so that they can then think critically or dive deeper into why that might be happening.

Rick Kranz (42:35.074) Yeah, and even if it is doing well, it will find some subtle things that may be coming down the pike that you didn't realize. Yeah. Activity seems to.

Databox (42:43.856) Got it. Yep. Yeah.

Rick Kranz (42:50.7) Well, I from my experience, my activity slows down the better I'm doing, right? The better the company is doing, my activity slows down. And then when the company's like, shit, my activity speeds up, right?

Databox (42:59.772) Yeah, I think that's human nature to some degree. When your funnel is full, it's hard to make time to go and prospect. There's also less incentive because your funnel is full.

But then all of sudden, when your funnel dries up, you're like, shit, and it's too late. So I think a tool like this, the sales polls, you to stay ahead of it and gives you a reminder like, you're doing great now, but next month doesn't look so great, right? If you don't. Rick Kranz (43:25.

09) Yeah, it looks activity is it's slowed down. There's less coming in the top of the pipeline. So yeah, it's a great tool. And like I said, I wish I had it when I was doing sales.

Databox (43:40.636) Yeah. Got it. the question I've been getting, so we've obviously been plenty of people like yourself out there who are using Databox plus Cloud and Databox plus Make and Databox plus NAN and like in order to automate analysis, automate, you know, the identification of insights, issues, opportunities, in some cases taking action in an automated way.

So we have plenty of people that get it. But my bigger struggle is people that don't get it. And the question I get quite a bit when I ask people, or the answer I get quite a bit when I ask people, you doing data analysis with Claude? It's like, yeah.

And then I ask him, how are you doing it? And I get a few different answers. One I get is like, I'm screen grabbing my charts from my systems and I'm pasting it in to Claude so it can explain to me what's going on or write a report for me. Or.

Rick Kranz (44:21.208) Mm-hmm. Databox (44:35.748) I'm downloading my data into a CSV or spreadsheet.

I'm uploading into cloud so that it can create a chart for me. Or I'll get, I just connect all these individual MCP servers and then I do my analysis. And so I know you've tried or have done all of those things at some point along your evolution here. And so why is it that you use the MCP server instead of doing one of those other things?

Rick Kranz (44:56.856) Mm-hmm. Rick Kranz (45:06.59) Well, the MCP server, it gives access to...

Rick Kranz (45:14.882) Because creating a chart is not the end. So a lot of people think that taking data and getting a chart created is, that's the goal. That is not the goal.

There's two more steps after that goal. One step is analyzing, what does this chart mean? And the second one is like, well, what do I do next? What's the action that I got to take?

And that's. Databox (45:29.221) Right, right. Databox (45:42.

032) Yeah, what happened and what should I do about Rick Kranz (45:44.995) Yeah, and that's why I bypass the charts completely. Just you can bypass the charts and just if you give Claude access to the data box MCP, you can just give it all your data and say, okay, find the trends. Databox (45:59.

164) When you're prompting Claude with the DataBox MCP, you're just asking high level questions, right? And it, meaning Claude and the DataBox MCP, is doing in-depth analysis with basic prompts, pretty basic prompts. You're not sitting there saying analyze it this way and then this way and then this way and if this, do this. You're doing maybe some of that, but not a lot of that, right?

Rick Kranz (46:23.0) You don't have to, because it will figure it out. So we create skills. I'll work with Claude.

My advice is you want to do this on your own, the best thing to do is work in a Claude chat. Use Opus. Sonnet is fine, but Opus is much better at figuring this shit out. Use Opus.

Have a conversation with Opus. Give it access to the DataBox MCP. Tell it what you're trying to accomplish. Tell it you want to ultimately create a skill that you can run either every day or weekly that will give you this information.

And then it's going to come back with some great ideas of how it could build this for you and what it would include. And it's going to start asking you questions like, what do you want to see here and there? And then have it just create the skill. And that's my process, create the skill.

And we have skills that we run every week. We have a content performance partner skill that goes across, again, it uses DataBox MCP, but it's looking at our content. It's pulling content information from our HubSpot through DataBox. It's pulling visits.

It's pulling conversions from HubSpot CRM from DataBox. It's pulling Google Analytics, and it's pulling Google Console. And it's all running it through the filter of our ideal client profile, our ICP. that's like, nobody's doing that.

like, okay, it doesn't, yeah, so data by itself doesn't really, could totally mislead you. could say, yeah, your traffic is doing great, right? But run that through a filter of your ICP and it'll tell you, so the content performance partner will tell you that, okay, 60 % of your traffic is not your ICP. Databox (48:18.

854) Mm-hmm. Rick Kranz (48:19.148) because of this one piece of content that's doing very well. Stop writing about that piece of content.

Databox (48:24.189) Right. So it's looking at not just at the performance of the content, it's looking at the actual content itself and saying, does that content reach the ICP as Rick has defined it? And if not, then it's saying, hey, it's a great piece of content.

This content is performing well from a numbers perspective, but it's probably not helping you figure out your business because it's not bringing the right ICP. Rick Kranz (48:43.422) Yeah. Right.

Yeah. And it'll look at the deals. It'll look at the leads. Like, ours, you know, it did come back in that sense.

so we had a piece on building a RAC system and it said, no, that's appealing to really developers because we had a really in-depth articles on how to build a route RAC. So, and that's not our ICP, development ICP. It also came back and said, even though it's looking at all the blog articles and everything, it said, look, you're killing it. on LinkedIn with your content and conversions.

So do more of that. Databox (49:17.052) I looked at it by channel as well. We have a similar system.

So we publish on lot of platforms, like almost all the social platforms, our website, YouTube, different formats. like each one of those systems, we're pulling in content that we published. And we actually have a weekly meeting where we review that content together because we have different people managing different channels. It's like, this is what worked here.

And now we're at a point where Rick Kranz (49:19.82) Yeah, so looks at it by channel and comes back. Databox (49:46.308) multiple people can run the analysis and see what's working on other channels so that they can inform their own work that week.

So, at some point, I think we'll get to the point where it's analyzing itself, all the channels and giving us direction or suggestions on what each channel should be doing so that it can... Rick Kranz (50:05.518) I don't know if you post but these are these are free skills. I don't know if you're posting links with the podcaster Databox (50:11.

813) Yeah, we can. Yeah. So you have four skills. know, I think I know them off the top of my head.

You got the sales pulse that analyzes an individual salesperson's effort, productivity, and gives advice. You have a newsletter subject line kind of a creator. Like it looks at all of your past newsletters, looks at open and and identifies what works in your subject line. So as you're writing your next newsletter, it'll help you write one that'll get maximized opens.

You have the content. Performance partner that you just mentioned right and Shit, what's the fourth one? There's a the weekly growth dashboard. Okay.

Got it, which is basically just running an analysis of your different channels and saying here's what worked Rick Kranz (50:46.527) weekly weekly growth dashboard. Rick Kranz (50:55.82) Yeah, it runs a running, I think it runs a six month or no, I mean, it's either four week or I don't know.

I think it's four week rolling average for the last four weeks compared to the four weeks before. Yeah, it's a rolling average and it gives advice on what's working and what changes to make. Databox (51:14.394) Yeah, got it.

cool. Rick Kranz (51:16.802) And they all pull from the data box, MCP. Databox (51:20.

176) Yeah, yeah, yeah. Yeah, so the question I was trying to lead you to is was more around like because data box structures it, what's that? Rick Kranz (51:26.094) Why can't you do this without the data box MCP?

Because you're not going to get it. Yeah, I tried it. It doesn't work. So are you going to do?

So I'm going to hook it up to, OK, let's say in this case, I probably have at least four or five different sources, right? And it doesn't pull the right information in. It just doesn't. It will pull some of the information in from one of the sources.

Databox (51:31.132) Yeah, yeah, I it. It just doesn't work yet. Databox (51:43.

942) Mm-hmm. Rick Kranz (51:56.231) and it may be in the wrong structure. And I don't understand what Databox is doing under the hood.

You can explain that. But there's something going on that it works with Databox and it matches when I check it. But it doesn't work if I don't use Databox. Databox (51:59.

676) Mm-hmm. Yeah, I can. Yeah. Databox (52:11.

473) This is like an end room. The sales team is starting to get the same objection where it's like, what can I just hook up all my MCPs to cloud and do that? And it's why do I need data box? And so I explained it to the team in technical detail.

Like I'm not an engineer anymore. I haven't written code in a while, so I'm probably 90 % right. But I explained it to them, like all the things that we do behind the scenes. then at the end of it, I'm like, I'm not sure it makes sense for you to explain that to anyone, because I don't know if anyone's going to understand it.

Until it's like this. It's like our answer has to be like, trust me, bro. It is work. But yeah, like at a very high level, like you need to define a bunch of things.

And it's like, you can do this elsewhere, but you need other tools besides data box. And you'll have to piece multiple tools together to do it. But you need Rick Kranz (52:47.614) Trust me, bro.

Databox (53:07.356) Three things, one, the understanding of a semantic layer, which is basically how all your data is connected. Like a simple example is your CRM, It deals and salespeople deals and accounts, Deals and contacts, they all have a relationship. And so the system needs to understand those relationships.

Now that gets really complex when you start connecting or relating data across systems from your CRM to your financial system, from your financial system to your whatever, your... internal operation system, right? And like you might have multiple pieces of software. So you need a system like data box to store those relationships between those different objects or tables.

That's one. The other one is metric definitions. Most of the time the data is stored in a way that makes sense, but not to a human that's looking at it. We can't possibly look at all of the data and turn it into a KPI effectively without some math in between.

It could be as simple as like, Rick Kranz (53:45.208) Mm-hmm. Databox (54:04.509) I need to add all these things up to get a sum for the hour for the day for the week.

Or it might be as complicated as like, I need to calculate, it could be not that this is complicated, but it could be more complicated, like calculating a ratio, in which case, you're calculating a ratio for a day, you you can't add up the ratios from every day and divide them by the number of days, because that'll give you a different ratio, if you look at the raw data from the whole week, right? So Rick Kranz (54:24.941) Right. Databox (54:31.

942) There's those rules and those are just two simple rules. There are other rules like sometimes a number going up is good. Sometimes a number going up is bad. And so there's all these little rules that we've defined that are related to each KPI in our system.

And we've defined many of them out of the box for all the integrations that we've built. And it's also possible for customers to go in and define these things easily in our interface. So once those things are defined, then you're doing the calculation of the KPI consistently. The third thing that's needed is doing statistical math in a consistent way in order to compare KPI.

So if you're going to want to understand and do analysis of data, you need to understand cause and effect. mathematically, you need to know, is this number correlated to this one? For example, if my salesperson has more meetings, do they close more deals? Usually that's true.

And so that would be an effective correlation. But the system doesn't know that. It needs to know what's the math. And it also could not be true.

There could be some sales reps. that do way more calls, but don't close as many deals. So it's important to be able to do that comparison to see if there's correlation or not. And that's just one, correlation is one, and detecting anomalies when something's out of whack, wherein there's usually an issue.

There's trends, right? No understanding how something's trending, which may or not, may not be something due to something you control. And so you need to understand trends independent of correlations. So all that math gets done in data box so that the LLM can do what it's good at, which is stepping through questions and thinking, you know, grabbing answers and, and, and, doing, following a process of some sort to do an analysis and explaining the actual results in words.

That's what an LLM is good at. It's not good at all the other stuff that I mentioned, understanding your data, defining your metrics, doing the math. Rick Kranz (56:04.909) Mm-hmm.

Rick Kranz (56:26.158) That's why it works. you know, there's also something, correct me, I may be wrong on this, but I tried to connect just for a test, Claude, to HubSpot marketing. And I don't think there's an MCP for HubSpot marketing.

Databox (56:27.356) That's why. Databox (56:35.899) Mm-hmm.

Databox (56:41.978) No, they haven't exposed the marketing data via their MCP server yet. Rick Kranz (56:45.902) Yeah.

So, so if you're on HubSpot, you're not going to get any of your marketing information. Unless you use DataBox. Databox (56:50.502) That's right.

Yeah. Keith, actually in your, in your, in your course wrote an article on this. There's other things too, like something you mentioned earlier, like if you want to do a month over month comparison, you need data structured in a certain way. And so that the month over month analysis can be done.

So that's one simple one. But there's all kinds of other limitations. It's not just HubSpot. I don't want to just pick on HubSpot.

A lot of MC. Rick Kranz (57:02.136) Mm-hmm. Databox (57:17.

638) P servers just expose the API data. But again, that doesn't mean the semantic layers is exposed. It doesn't mean that the metrics have been converted to KPIs. In fact, most systems that we use on a day-to-day basis do not have a capability to convert raw data into a KPI.

All they do is take the raw data and plot it, like just put it on a chart. so in order... For us to be able to allow a user to just visualize data from multiple systems, we had to come up with that solution of deturning raw data into KPIs so that we had one time series to show and could standardize on time, date ranges and state date pickers and things like that. So we've had to solve that problem for visualization purposes.

It just so happens that also becomes really valuable when you want to do data analysis with. Rick Kranz (58:12.012) Mm hmm. Yeah.

Yeah. The other thing I would say, the reason we use data boxes as a rule, I teach everyone to use trusted third party systems that don't contain AI to connect to your proprietary information. Like we don't connect. We don't connect Claude.

Like I said, I just tested it. We don't connect Claude to our email or to our calendars or any AI to our Databox (58:23.164) Mm-hmm. Databox (58:28.

829) Okay. Databox (58:39.396) Yeah. Rick Kranz (58:40.

074) stuff like that or to our CRMs. For two reasons. The first reason is primarily it's not going to pull the right data or all the data. Like I can't ask an AI to look at all my emails.

It's just not going to do that. It's just the context window. It'll blow up, use all the tokens. So we always use trusted third party systems like whether you use Make or N8n as an automation platform or Zapier.

Databox (58:44.784) Okay. Databox (58:51.291) Yeah.

Rick Kranz (59:09.326) or data box. So these are trusted third party systems that have been around forever. They work.

then those, they have customers as part it, those expose the information that we want. So it actually collects the correct information that we need to give to the AI. So the same thing. So in a non-data box scenario, let's talk about emails.

Like I want certain emails exposed to the AI, but I don't. Databox (59:13.424) Yeah, we have a customer support. Rick Kranz (59:38.

179) really want it going through my inbox because I know it's just not going to grab everything. So I use an automation like Make to go through the inbox and parse everything correctly into the form it needs and then feed it to the AI for analysis. And then the AI. And then there's the safety reason too, in case you just don't want to expose your.

Databox (59:40.465) Mm-hmm. Gotcha. Yeah.

Databox (59:52.024) Okay, gotcha. Databox (59:58.724) Yeah, yeah, think another reason to do that too is it actually like reduces your credit usage, right?

I'm sure you've run into. Rick Kranz ( ) Yes. Yeah. You get a lot more out of, out of your token usage, especially if you're, you you're running Claude and Opus.

Databox ( ) Yeah. Right. If you're running cloud and you're connected to five different MCP servers to try to do data analysis, that's going to guess and check a million times before it gets it right. Which is just going to blow through your credits.

Whereas if you use data box, the first thing that the data cloud does with a data box MCP is, is figures out what data source to use. And the second thing is figures out what data set or metric to use. And so it's doing it very methodically. Rick Kranz ( ) Mm-hmm.

Rick Kranz ( ) Mm-hmm. Rick Kranz ( ) Mm-hmm. Rick Kranz ( ) Okay. Databox ( ) And if you give it instructions to say like I want to look at these three data sources these four data sets these 12 metrics It'll do just that And you won't have to you won't it won't be hunting around for everything trying to go and figure out what's there Rick Kranz ( ) Yeah, Yeah, yeah, I remember early on you asked me when I was doing this, you asked me what data sources in in data box did I expose?

You know, did I ask it to look at and I was like, I don't know. I just told this is what I want. And then so we opened the skill and we looked at it together and I was like, oh, yeah, here's all the data sources. It figured out on its own what it needed to pull from data box.

And I would have never I would have never known what to expose. Databox ( ) huh. Hahaha! Databox ( ) Yeah, gotcha.

it will do that. It will do that. You can say, I want to analyze my sales performance for my sales team, right? To Claude and it's hooked up to the MCP.

You might say, hey, use the Databox MCP so it avoids trying to use other things. But yeah, it'll know or Databox will know, all right, sales is related to... Rick Kranz ( ) Mm-hmm. Databox ( ) A CRM so I'm to look for CRMs and then I was like I found this one and you can set it up so it asks or sometimes it just proceed to the next step.

Rick Kranz ( ) Yeah, it's pretty clever working with data box and that. yeah. Yeah, so that's how I nerded out when you guys released the MCP. saw it right away.

This is like a big change to what we can do now. Databox ( ) Cool. Yeah, you totally did. Yeah.

Databox ( ) Yeah. Yeah. The world, don't think people realize that everybody's still thinking, how do I use AI to create content? How do I use AI to like do my sales follow up or my prospecting?

They're thinking about automating like the tasks that people don't get to or don't have time for, or it's like drudgery in a way. And it's not that hard. And I think that what's missing from the conversation out there Rick Kranz ( ) Mm-hmm. Databox (01:02:36.

12) Is that really hard stuff is possible now. It's like the work that you're automating requires like a data analyst, maybe a developer sometimes, or in the old world, right? And a management person, like somebody that understands the business well enough to say like, this is what I'm doing and what we're analyzing. are like, this is what my problem is.

Rick Kranz (01:02:41.4) Yeah. Databox ( ) those enough to ask the right questions of the data or at least interpret the data. And so when a human has to do that, like there's three really expensive resources involved, a leader, management, executive, right?

developer and an adult. like that automates some seriously expensive work. The other thing that I think most people aren't talking about is like, that takes analysis from like a once a month, once a quarter thing to an anytime thing. And the depth of the analysis can be really, really deep.

So now it's like you almost, you can have one leader, one VP, one executive, whatever, that actually is able to oversee way more because they can, we can monitor not just the actual Rick Kranz ( ) Mm-hmm. Databox (01:03:58.33) Numeric performance, but the actual work because we can observe what customers are saying or you know and have that instantly we can we can record internal meetings we have project management systems like so in my my world's changed quite a bit where I used to like delegate initiatives down to everybody and then once a quarter we'd review how those initiatives are going and they propose what they're gonna do next we'd ask a lot of tough questions and make some decisions and now it's more of like Rick Kranz ( ) Mm-hmm.

Databox ( ) I create, I have a project, I have a bunch of context files in there for what we're doing or not doing. And then I connect to a Slack MCP, Asana MCP, Dadox MCP. And I basically just started exploring what are we doing? How's it working?

What's, what's the team? here's, and then as I'm doing that, I'm like, here's the team doesn't seem to be thinking about this, this, this issue or, Hey, they should try this. And so I can get to the point where like, like, if I have few hours blocked off, I can go deep in an area of our business without talking to anyone, write a memo and with a bunch of questions, a bunch of guidance and send it off. And the following Monday, they're coming back with like, read what you did.

Here's some of the questions I have, or I what you wrote. Here's some questions I have. I'm having trouble with this. How would I do this?

And like, then they're off and running and building. And the speed at which we can do that. And the fact that we don't really have to wait for the quarter end to do it is. enables us to move so much faster.

Rick Kranz ( ) Yeah, I was just thinking about like, think about the course correction abilities. Now imagine if your, if your automobiles GPS only updated every 10 or 20 miles, right? That's how we used to run or that's how we used to run companies like we'll look at, okay, monthly reports, quarterly reports, okay, and maybe we adjust, right? You make one, you make one year, five year plans.

was the old days. Right now you can course correct as Databox (01:05:37.18) Right. Databox ( ) That's a great hobby.

Databox ( ) Right. Yeah, as soon as you make the wrong turn, can say, hey, we made a wrong turn. Let's turn around, right? Yeah.

Yeah. don't think humans are ready to, organizations I should say, are ready to move at this speed. Rick Kranz ( ) Exactly. Rick Kranz ( ) No, no.

it is cool that executives or non-executives, people who, a lot of these reports were just stuck behind firewalls, firewalls of expertise. Like certain people had the expertise and the responsibility for generating these reports. And sometimes they were, you know, I guess influenced by someone's bias. Databox ( ) Yeah, they're biased, maybe being protective of their turf or protective of resources.

Rick Kranz (01:06:33.41) Yeah, yeah, exactly. And now anybody in the organization, if you're open enough, just get these reports right away. Databox ( ) Yeah, I think if an organization operates in a systematic way, right?

If you sit down once, if you have a long-term vision and ambition, right, that you have agreed upon, if you sit down and say, what are we going to do in the next 12 months? Do we think that will help us get there? Right? If you break that down into like quarterly projects or initiatives, and then you assign those to individuals or build teams to execute them.

And then if those teams operate transparently where they're sharing their weekly updates, their performance, what they're planning to do, what didn't work, et in some documented form. And it doesn't have to be fancy, it can be a Slack message. Like if that, if the company's operating that way, then the ability to adapt and course correct, lean into what's working quicker, it's like, it's unprecedented. But there's a lot of ifs.

Rick Kranz (01:07:34.69) Yeah, because the amount of human work and effort that it took just to create the report to figure out like, do we need to course correct? How is this working? That's eliminated.

That's gone. So now you can actually do the intellectual work of figuring out like, what is the course correct? Databox ( ) Mm-hmm. Databox ( ) Exactly, it's gone.

Databox (01:07:50.64) Yeah. Yeah. You have that data and you have the context of what was done and what's planned.

And you have customer calls if that's relevant for the project, right? You have internal meeting notes. All of that is so easily recorded and accessed now. Yeah, it's a little scary, but it's also really powerful.

Cool. All right, so what's next for the lab? What are you all planning to do next? What's your next big move?

Rick Kranz ( ) I don't know. big, well, we're staying the course right now. We are trying to, what? Databox ( ) Yeah, what's working?

Just doing more of what's working. Rick Kranz ( ) More of what's working, we want to get more people into the class. And that's where we're getting good referrals. it's mostly word of mouth.

And people coming from LinkedIn, people coming from Databox, from the mastermind group coming in. We're to be doing more work with companies, so building out for companies. So I see more of that coming. Databox ( ) Okay.

Yep. Databox ( ) more, more like a one-on-one customized, you're helping more, you're helping them more hands-on. Rick Kranz ( ) Yeah, one on one custom automation. Yeah.

Yes, more hands on actually building, actually going in, looking at what makes sense to automate, designing the automation, installing it, training, then, yeah, leaving. it's not a retainer model. It's a project model for that. And then anyone that we do that for gets access to our school so they can come in.

And that's where they can learn more about. Databox (01:09:27.76) Yeah, project model. Yeah.

Rick Kranz ( ) Customizing if they want their existing automation. They'll learn how to run it on their own if they have the time to do that. Databox ( ) Why are you planning to do more 101? Do you find that like some works just need more help and that the...

Rick Kranz ( ) Yeah, yeah. The organizations that need our help the most are the ones that have too much work. And the ones that have too much work don't have the time. Databox ( ) Right, but they maybe have some extra money to say, we know we can be more efficient, and we know we can be better at doing this more efficiently.

Rick Kranz ( ) they have the money. Yeah, we can't get to all of this like the painting contractor example. Yeah, the painting example. Yeah, so the companies out there, those are probably our best types of prospects.

Yeah, because they're too busy. They're too busy. They don't have any personnel to send. they just need a system now.

Databox (01:10:17.34) Yeah, Databox (01:10:25.98) Okay, cool. Databox ( ) Yeah.

Got it. Awesome. So I didn't say this earlier, but I was going to ask you about this. What I love is how focused you are on the getting the outcome, right?

There's a lot of, I think, AI consultants out there that are explaining how to do things, like, I'll show you how to do it, or I'll tell you how to do it, right? But but you have this like hands-on kind of approach. You had that from the beginning, I think even before, maybe it was even possible to be that hands-on, right? Like before the tools were really that good.

So like why, and I think that's the way you are. Since I've known you, you've always been like kind of a brass tacks, boots on the ground type guy, but like, why did you go that way? Rick Kranz (01:11:03.65) Mm-hmm.

Rick Kranz ( ) Because there was, well, one was to differentiate it. were in there, you know, two years ago, three years ago, there were a lot of people just launching these AI communities where they built automations and you could join and access and just grab the automation and install it. A lot of what I would call marketing. What's that?

Databox ( ) Okay. Databox (01:11:56.56) More of a teaching model. More of a teaching model.

Like they watch a video, grab it, and go. OK. So more just access to the skills. Rick Kranz ( ) No, not even videos, just grab and go.

Yeah. We're asking access to the skills. So it was a low, low cost, high volume play. and he, we started out with low cost.

We actually modeled that for a couple of months. it wasn't like, I said, okay, no, this is not where I want to be. The people that we want to work with are people like you. Databox ( ) Okay.

Yeah. Databox ( ) Okay. Rick Kranz ( ) people who are seasoned, who are in the business, are actually working and have to do stuff. What was happening is we were getting a lot of people coming in that wanted to start their own AI.

Right. And nothing wrong with that is and there's, yeah, and there was some great guys out there that I watched and followed in the early days that built great, great businesses doing that. And Databox ( ) OK, yeah, No recursive thing, yeah. Of course.

Rick Kranz ( ) But it's not where I wanted to go. I wanted to go with like, okay, more hands on because I think I got used to that and working with in my agency, working with companies, working with CEOs and VPs of marketing and then VPs of sales and salespeople. Yeah, it was more about did a lot of change management with people. it seems like it makes sense.

It's a higher price. Databox ( ) Yeah, right. Yours is a little more like a agency model. Databox ( ) Maybe it's necessary for orgs.

think, you know, we're what 80 people here at DataBox and like, it's a lot of effort just to give people to try stuff and then really learn it and like play around with it and experiment with it. Like, you know, I learned something new this morning when I was showing my sales team, hey, here's how you do this. And it's like, just type this and like, I didn't know you could do that. So like, I think there's, there's a lot to learn.

Sometimes it's just. Rick Kranz (01:13:33.72) Mm-hmm. Databox ( ) learn by watching other people a little bit faster, right?

And doing it with them. So you can see the tips and tricks and shortcuts and approaches that. Rick Kranz ( ) Yeah, definitely. mean, I can't keep up with it all anymore.

In the early days, was Kelly and I launched this organization, we didn't have a lot of members. So I would spend 50 hours a week watching YouTube videos, learning everything and building. yeah, it was like, at that point, I kind of did know everything about all AI platforms, including video creation and Databox ( ) no, that'll be scanned. Databox ( ) OK.

Yeah. Databox ( ) Okay. Rick Kranz ( ) and audio creation and all that stuff. Now it's just accelerated so much it's hard to keep up with.

But what I really enjoy is, we have a lab session this afternoon, invariably somebody will bring in something and say, do you know this? They released this now, we could do that. And I'm like, no, I didn't know that. Databox ( ) Exactly.

You know, it's crazy. Like we're doing a lot with video production. We have like avatars that we built that we do, you know, AI video production. We're working on like how to produce explainer videos now with AI.

So yes, we're using Hey John for the avatars. Yeah, we actually created a fictionalized like person and a YouTube channel for that person. And that way we're not, you know, pretending it's someone else, but we'll be. Rick Kranz ( ) huh.

Rick Kranz ( ) Are you using Hey, Jen? Okay. Yeah. Rick Kranz ( ) That's good.

Rick Kranz ( ) Yeah, I like that concept better than just pretending it's somebody. Are you, yeah. Databox ( ) So cool, all right, we should probably wrap. We're at hour and 20 minutes.

Yeah, yeah, I could tie your slide down. Rick Kranz ( ) Yeah, my brain hurts too, I could eat. Well me slowing down I'm slow to start with so it's gonna be pretty bad Databox (01:15:27.11) Just kidding.

All right, so what's one final word of wisdom? Let's just say someone's not doing data analysis yet. What would you tell them to go do? Like, how could they get started?

Rick Kranz ( ) I would say just go to the link that Pete's going to provide with this podcast. There's four free downloads. And you don't have to pay for anything. Just download them and pick one and install it.

It's self-installing. It's going to guide you through how to install it. And you need a data box account, but it's free. You can get the free trial or the free account, right?

So just Databox ( ) Okay. Yeah. Yep. Databox ( ) Yeah.

Rick Kranz ( ) Start there. It's the easiest way to start. Databox ( ) Got it. Yeah, and I think like people don't realize how quick it can be.

I've had it in the back of my head. I get to produce a video just to prove this to people, but you could be up and running using Claude and a free data box account doing data analysis in probably about five minutes if you're focused and don't get distracted by something else. And it would do proper analysis, not, you know, not. Rick Kranz ( ) Yeah, that's, that's, I would say that's accurate.

Yeah. Databox ( ) Cool, thanks Rick, appreciate it. How do people get in touch with you if they wanna learn more about your membership community or hiring you directly or just following you so they can learn from you? Because you do publish a lot on LinkedIn and share.

Rick Kranz ( ) I mean. Rick Kranz ( ) Yeah. Well, yeah, you could, um, you could follow me at LinkedIn. I'm at Rick.

It's my handle is Rick Kranz at LinkedIn and the AI marketing automation lab is AI dash marketing automation.com. Databox ( ) And it's Rick Krantz, R-I-C-K-K-R-A-N-Z. So you can Google it and find that.

Cool. Thanks, Rick. Appreciate you sharing everything, all your wisdom with us. Hard earned.

And I will probably talk to you tomorrow. see you then. Rick Kranz ( ) That's right. Rick Kranz ( ) feet.

Rick Kranz ( ) Yeah, for 20 a month. Take care, Pete. Thanks.