SPEAKER_01
digital operations, AI operations, it all kind of sounds the same thing. Essentially, it's automation as a service. What we do is we take in people's procedures for how they handle anything from client onboarding, sales, marketing, customer support, really you name it.
Anything internal with data that could be reported through Make, Zapier, or Airtable is in our ballpark.
SPEAKER_00
Welcome to the In The Pit podcast. I'm your host, Cody Schneider. This podcast is brought to you by Swell.
AI. Swell.AI is a platform for content marketing powered by AI. Go to Swell.
AI.com to sign up for free. Let's get started with today's show.
We'll gas you up. We'll gas you up all day. What time is it? Where are you right now? Are you in Barcelona?
SPEAKER_01
No, dude. I'm in the Italian Alps. So in the Dolomites, it's snowed for the first time.
SPEAKER_00
I'm going to have to shoot that turtleneck on. You're looking too hot.
SPEAKER_01
Oh, yeah, man. It's going. But yeah, last week in Italy, so it's been good.
Do you have a visa?
SPEAKER_00
What's your deal? All right, like, are you just ex-patent and doing the whole three-month thing?
SPEAKER_01
You pressed record, so I don't know how much I can say. But no, I'm on it. We'll leave it there, and we will move on.
Perfect. No, it is a student visa that's going on a little too long, and we'll leave it at that. I love it.
SPEAKER_00
I love it. What the hell, yeah, great. Thank you for joining me today.
I don't know what you want to get into. But what I'm excited about is I want to talk about like, concept marketing workflows and like, automations around that. And then hear just about like, what you're doing and just everything you're building.
So for the audience, talk to me about, I mean, you probably have given the spiel a million times, but I want to hear from your mouth like, what you think about. Like, I call it digital operations. People call it like, automations engineers.
How do you view what you're doing?
SPEAKER_01
Yeah, digital operations, AI operations, it all kind of does the same thing. Essentially, it's automation as a service, a little bit of ass. And what we do is we take in people's procedures for how they handle anything from client onboarding, sales, marketing, customer support, really you name it.
Anything internal with data that could be ported through Make, Zapier, or AirTable is in our ballpark. So it's, I don't know, it's growing. It's a hot space right now, but it's really started from the needs that a lot of.
Yeah, it's started from a lot of the needs that companies have realized that they want cash flow, they want lean teams, they want to stay nimble. And automation has just become the go-to solution for that. And so when you can combine the things that happen in Make and Zapier with all the new AI tools that are coming out, like Draft Horse and Swell, you can do a lot as far as stacking time and efficiency and all those great words.
SPEAKER_00
No, I love it, man. No, that's sick. Well, hell yeah.
So you're still at Hampton though this week, right? Yeah. Yeah, yeah. Still at Hampton and then doing this on the side.
Sick. Okay, man. I've been just like, what I've been obsessed with right now is like content marketing workflows where it's like building these automations.
Like what are some of the things that you've seen lately that have been like you've been stoked about or just, honestly, I just didn't like, tell me, tell me what you're, tell me what the last couple of weeks in your life have looked like and like, what have you built? Like that's really what it comes down to.
SPEAKER_01
100%. So this is, this is what I was excited to talk to you about and why, why I needed to push back a few days because I made some progress. So essentially, you know, what I'm trying to do on the content side right now is just talk about all these AI tools that are coming out.
And so what I've been able to do is find a way to, if I, if I come across a newsletter that talks about AI, I've got that's, we're scraping all those email newsletters and pulling out all the AI tools and storing them in air table. And so then on a monthly basis, you can count what's being talked about the most and who's being like, who's making progress. And we can write about those in a monthly report.
But on the, on the day to day content side, there's something called the air table web clipper. And I'm going to take you on a little bit of a journey here. So bear with me.
Please, please educate me, man. There's a thing called an air table web clipper. And what you can do is you put on this Chrome extension and when you go to a website, it will pull in a page screenshot.
It'll pull in the URL, like a social media image, if the website is already kind of geared for that, or you can just pick images off that are better than screenshots. Because you're kind of, you're extracting the image file itself from the website. So what I do is I find these AI tools, I clip them and get all that, all those social media images, all those logos and everything.
And I send that to an air table base. When the air table base gets a new record, I have an AI research tool, look up and scrape everything about their website, get the founders names when they were, when they were funded, when they were founded, what the integrations are, what the pros and cons are. Other considerations that my clients would probably want to know.
That information gets brought into Zapier when the email, when the research spot responds via email, we put that into structured layers, deposit it back into air table. And then the final step is when all those fields are filled in air table, ping GPT and say, Hey, write a product review for this. And then here's an example and put it into the right format.
So the last human step is format it nicely with the right bullet points and make it all look nice and typefully. And then we send that on its way. So that's my, that's my content game right now.
I'm really excited to let that thing fly through.
SPEAKER_00
That is unbelievable. That is unbelievable. Okay. So, so you're basically just to walk through that and kind of break it down in this step. So it's like, I get raw materials.
I then have AI do some type of thing with it, like based off of a template that I provide it, and then the human, that's when they come into the loop and they're kind of putting those final touches on it. They're like, get it, you know, get it basically the client facing or customer ready. Is that like a simplified version of what you're kind of describing?
SPEAKER_01
Exactly. The human, the human step is the formatting. That's really all I want them to do.
SPEAKER_00
Interesting. Interesting. So when you're thinking about like the template outputs that you're trying to have it generate, like, so this is something that we've been experimenting with a lot.
Lili is basically like giving, giving the scaffolding in the form of like JSON or like structured data. And then, yo, like you need a, basically this is what the output that we're trying to get you to generate. Here's the raw materials to do that.
And then, you know, letting it, letting it kind of do its work. Have you, have you been experimenting with that? Or like, is that why you're basically put into this? Like, I guess you are in a way, because the air table is a structure, like that would be a structured data format is, I guess, no, I'm thinking about it. Right. So it's like, this is this consistent format. We already know it's almost like.
Fill in the blank from a content output that we're trying to produce. Oh, fuck that's super interesting, actually.
SPEAKER_01
Cause I don't, I don't code, right? So I'm still on all the no code stuff and air table, air table on Zapier. It's all these just little variables. So when you're plugging something into the GPT prompt, it's like the variable for the pros of this tool, the cons of this tool, who the, like, what's the founder's name? And then if we get tricky with it, we can try to ping Twitter and find out what their handle is and tag them, right? And just reduce all those little steps.
But I really think that when you give it all those little variables and the way that you're kind of describing it as JSON, I'm just using different columns and the same row for the company. I throw that into the prompt and I say, Hey, here's an example of what an output would look like in this kind of more like a blog format. And then when the blog is written, then the editor goes in and just chops into tweets, puts it into carousels, et cetera, et cetera.
It's amazing.
SPEAKER_00
I mean, you can take that a set, like a step further though, right? Where it's like, so something I've been, I built recently was basically like, you take rough, like, I've been doing this with like Reddit posts, find viral content in some place. You can do it on Cora. You can do it on mediums, you know, any place where it's like, there's dialogue that's happening, like public forums, take that forum content.
And then you basically build it into structured data. And then we built like a Canva automation that just automatically generates like, like Instagram, like, like cards or like LinkedIn carousels. I've been doing this with LinkedIn because I don't want, I need to be on LinkedIn, but I don't want to.
No. But I'm just thinking about like what you described, like you could just be like, okay, cool. Now the blog post is written.
I think here's this next step where it's like, now I make it into like, make it into tweets, make it into a Twitter thread. I also want you to write five LinkedIn posts about this, you know, whatever that is. So what is this company under? Is it grandpa? Is that, is that what you're doing this? Or is this like in a, it's, it's own thing.
SPEAKER_01
No. So this is all grandpa. This is just for my agency.
I'm trying to get more content out there, get more, you know, just general trust about, I'm, I recommend AI tools for clients. I do all the automations. The two kinds of sides of my client game or my content game is going to be the product demos, trying to get actual time with founders and ask them more questions that I can kind of push the ball a little bit further, but then just this AI generated research totally and into the, into the formatted text.
But one thing that's really interesting to me that you talked about, which is like taking it a step further from the blog, chopping it up into these different formats, what has been your experience building those kind of a templates to automation there or right? Like getting it into it typefully. I don't think you're using type of your scheduling tweets, but like, how do you take a blog and then chop it up into those formats? Or what do you use to divide that? How do you know how does, I know you don't write very many threads, but how do you think about that kind of side of like segregating your content?
SPEAKER_00
Yeah. Yeah. I'm always looking for hooks, right? So like that's like, that's my big thing with it. What I'm doing more and more is like, I have a human basically create an output that we know is like the correct output.
And then we're using that as like the source material. And then like, we just did this for like a, like, like a mid-roll ad, right? Like within a podcast. It's like, here's our kids.
Here's mid-roll ads that we've written previously. Like I want you to write a mid-roll ad about this episode. So I can be like cross promotion on these different episodes.
But like, again, like do it based off of this format style tone that I just provided you, right? So on the tweet side, it's kind of the same thing. It's like, write me 30 tweets that are based off of this, whatever the source document is, or like, yeah, we're just, you know, I think the technical term is a corpus, so it's just like, you know, a library of text, whatever, you know, whatever that ends up being. But anyways, so it's like, here's that source.
I want you to, you know, write it in an irreverent tone. Here's examples of the ones that I've worked in the past, right? And then you can get, you can get this even further. And like, we're still building the pieces out where it's like, okay, it sounds more and more in the voice, the brand, it has that like training on top of it.
Then from there, like where we're getting excited is like what you're talking about where it's like, okay, cool. Now I have this table of output, like the content output that I just generated. Well, imagine if there's tweets, like they're in a table, right? Or they're in a CSV format.
AI is an adjacent format. AI is totally capable of doing that. And so then we can basically take that structured data and at that point, I can pass it on to like a buffer API, or we can just build our own internal tool or to a Zapier, right? And it automatically, like it goes and it schedules each of those, those tweets or those, those pieces of content.
So like, it's like taking, I more and more, how I'm thinking about this is like, how do we get raw materials that we think are valuable and then translate it into all of these different forms, like every channel that I want to be present on, like wherever my, you know, my target customer is, basically repurposing it into a native piece of content that's there. And then like every part of that workflow that we can, as much of that automation that we can, basically connecting those together. And like this is more and more, I'm starting to think about it as like content workloads, but what I do, what I'm getting fucking so excited about right now is like the idea of mixed media as the source material.
So like imagine like I, like in this is, this is where we're going with all this is like, I can go, I get an MP3 file. That's a conversation with this person. I give it the LinkedIn URL of the individual.
I give it three blog posts that have related content and like maybe other podcast episodes that they were, they were on, right? That ends up being my library. And then that library, I can then say, okay, cool. Like here's these content workflows that I want you to do, but I can chain together these different commands.
So it's like, yo, write extremely detailed blog posts that align based on this like corpus of information. Now write me extremely detailed blog posts. That's in the first person.
Now add a call to action. That's that in the first paragraph to sign up for XYZ software, right? But imagine that for every channel. And then I just like can get that into structured data that I can then put into these other forms.
So I think the thing with this that we're seeing is like, it's still so early. Like Canva doesn't really have integrations yet. So like I'm straight up just like uploading CSVs into this thing to like generate the content that I'm looking for.
Right. But the inklings, like it's still like, we're so close to all this being connected. Actually my app, like, have you helped me with this? We're just like, yeah, we're, it's what you just said, just made me think about this.
Like we could probably just connect through Zappi or to these other, like they, a hundred percent will have a Canva integration. So like build this template and it's like, you upload a thing into like swell and just automatically generates like all of this visual content for you. But like, I'm, how we're thinking about it is basically like, how do you create all of these pipelines for content production? Right.
If that makes sense. So anyway, just brain dumping the things I'm seeing. I don't know if any of that was interesting at all.
SPEAKER_01
No, it totally is. I mean, I think that when, when I think about what you guys are doing with swell, like the closer you get from the raw to the finished, right, is obviously what everyone wants. It's the least possible steps.
It's like, it's your go-to phrase of find something that takes someone time, automate it and ask for money. And like when I think about, I think we were going down in like the tweet, not the answer, but like in a thread, just you and I talking about asking chat to me to prompt you with questions and then just recording it and feeding it to swell. Like this is, this is what I, this is what I want to get into more is like, I can go and have someone do like the 90 minute ghost writing call and they'll put together something, or I can just ask chat to me to ask me an interesting question about automations, see what I come up with.
And then that 30 minutes I'll get probably just as much, but I know that I said those exact words and it'll remix it just as well as someone else. And it's some other part of the world. It's like, it's much faster.
It's much, it's just in the moment. And the quicker you can get that output to a scheduling software, that's what people want, right? Because then there's just no, there's no downtime.
SPEAKER_00
100%. 100%. I mean, the, the other side of this, he just made me think about this as like, imagine I prompt you at GBT, like ask me 10 questions, like mimic a ghost writer, ask me 10 questions. You answer those and then you provide the transcript back of your answers.
And then you say, ask me more questions based off these answers. So you could actually have a dialogue so it could even take it further, right? Like it could like drill down so it's less topical and even more specific based off of the expertise and the context that you just provided back to it. And like, that's really what a, you know, interview is, right? It's like, it's like, how do I, it's like that back and forth that leads to that insight, fuck, I'm going to do that later today and see what happens.
I'll send it over to you once I've happened.
SPEAKER_01
Dude, push, push that one stuff further. I mean, don't, don't copy and paste the transcript. Please go, go for like, do your recording, record somewhere, I don't know where on a tape recorder, but record you talking to the chat to be tea app because that talks back to you.
So then you have the actual conversation. You take that and be, because it's going back and forth. Like that's, that's where people, I think like, that's, that's where we can go with this beyond just the, give me the text questions and I'll speak them out loud.
Like people are using this for coaching, they're using it for emotional support. They're doing all sorts of different things or just like a, I'm driving to work and I want someone to talk to, wait about my deepest thoughts. Like you can do that with ChetnyBee Tea now, but from a content side, you don't need to have the other person, you don't need to schedule as much interaction with other people, obviously talking to you is great.
But when you want to just produce and get thought out, you now have the back, like the back to your response, you have a voice that prompts you back and it is intelligent, right? And you can say, actually, I want to talk more about automation on Canva. Like what do you, do you have any questions around that? And it'll react and generate five new ones. So if you record, the only problem is that you need to have two recording devices, right? You do one on your phone or one on your headphones and one on ChetnyBee Tea on your mobile device.
But yeah, I think that's, that's the further thing because that also produces a transcript. So you get the recording, but you can also get a transcript from that.
SPEAKER_00
Totally, 100%. Dude, that's honestly brilliant. I'm going to, I'm going to try to hack this together later today and just play with this.
So the interesting use case we just like ran into recently was basically like building, like imagine, like traditionally in marketing, like, or B2B marketing, like you're trying to serve the, the sales team, right? Like, how do I give them marketing materials that they can basically use to the close? And so what we realized recently, there's some feedback that we got. It's like people are like sales teams are looking for the ability, like, and they don't even know, they don't even know this exists, right? But these people are starting, like what we got approached with this, like, hey, I want to be able to build templates for my SDRs and my AEs, right? And then the calls that they have that go well, we just take those transcripts out of like the call recording app, because if they're serious, they're all recording, right? And then we just drop it in and it basically produces that LinkedIn post that's in a format that we know have the likelihood to go viral, like on their profile, right? And then this creates a way for them to basically like be in front of their audience, right? More and repurpose the conversations that they're having on a daily basis, right? That's like, you know, overcoming those objections. Like, it, like, it, like a, so I guess the more, the more I talk about this, like even what you just said, it's like, why don't you just record everything? Like record everything and then you're just constantly repurposing everything, right? Which is just like, it's just crazy to think about like what that means for like marketing orgs.
And like, we're just going to have so much data that exists in these silos. And it's like, okay, now, like, how do I access that and query against that? And then what's my, like, what am I, what do I allow to go in there and not to go in there? So that the data isn't garbage, right? But anyway, yeah. So talk to me, talk to me about, talk to me about like, what are the things, the common things that you're building for these, these people that you're working with, like the clients you're working with? Like, is there common trends that you're seeing? Is it a lot just like random process optimization? Or is there like, is there things that they're coming to you with? And they're like, yo, I want to build that but for my organization.
SPEAKER_01
Yeah. Typically the main thing is that people are coming to me saying, I feel like I'm in death by a thousand cuts. Like every time we do this thing, I just, I winced because I know that it's not as efficient as it should be.
And so with pretty much every client, we're not at the point yet where we can say, oh, you're an agency. So we should do X, Y and Z, like client onboarding and sales. Those are, those are wins no matter what, right? And almost no one has full automation potential there for an agency.
But when it comes to a SaaS, when it comes to an econ business, we're definitely still, we, there's a good amount of discovery that needs to happen because we personally, I don't have a ton of experience with Shopify. So I want to better understand how someone's using Shopify. And then essentially, once you know what the triggers and actions are, it's all the same, right? Like if I'm working with a CFO, if I'm working with a person selling pants, I understand how the data flows.
So you just need to tell me how your business works and we can build the thing. So that's why every time we have to start with it, almost every time we start with an audit, we say, okay, just tell me about your day to day life. Tell me about what day in the life of work looks like.
What's your pain points? What's a magic wand look like? And then we kind of build out those solutions. And sometimes I, with my content base, I do show these to people on, show it to people on sales calls. They're like, Oh, when can you build that for me? And I'm like, Hey, give it, give me a few weeks to make sure that it's running the speed that I wanted to.
And then we can talk about cloning it. But that's one of the beauties of doing things in air table is that you can just clone those bases and the tech, the interface, the automation they all stay. So that's a big part of my business right now is like making a really solid template for content, which is like, it involves the hooks.
It involves the formatting. Once that thing runs, that could be my SaaS entry point, right? Is like just the drop a quote from your favorite book or write, write a book that you're reading right now or not reading. I don't care.
And it will pull a quote out every day and repurpose it and give you a tweet. Right. That could be something that very easily gets done using air table and just templates. That's amazing.
SPEAKER_00
That's amazing, man. So, so on the outside of the content side, like, so talk to me about like the financials, like how are you seeing it being used there? And I, I mean, I, I'm always just looking for like the very tactical pieces. And I know this audience is that as well.
I'd love to hear that on the financial side. And then also like on the Ecom side, like how are you seeing these workflows being built? Question.
SPEAKER_01
I think since so much of what I do is an air table and so much, so much of what a lot of my clients are building is based on stripe, a really easy win is to just pour it over all of your, you know, essentially accounts receivable from stripe, all that data into air table and build a base on top of it, like an interface. So you can see how much cashflow is coming in. And I think that I really like to do is combine that with a company called FinTable.
Let me double check. That's what it's called. This is a high, yeah.
FinTable. Highly recommended for anyone, both personal finance and business finance. If you're using air table, it ports over all of your transactions.
So you can essentially see how all of the money is going into and out of your accounts. And when you pair that with your stripe data, you can then start to link it to your, if you have an SCRM, you can say, all right, well, this, this payment is coming from this client. So I know that their receivables are on time or they're going to renew soon.
So on the finance side, like FinTable plus stripe data, and then put an air table interface on top of that is really nice to understand how much money is going to vendors, how much money is in my credit checking, saving, etc. For a business, like that's a no code thing that it's also a great, like pet project for someone to get started. If you're trying to learn air table, there's no better way to make something tangible on air table, like your finances.
So hook up FinTable, start playing around with it, start building, like they have a nice template, the data comes over automatically. It's here, all that stuff's good. But then you can start to really have a motivation to learn a air table a lot better and the interface builder and the automations because it's your money.
Right. I think it's one of the best starter cases I've come across.
SPEAKER_00
Super interesting. All right. So I'm just like, I'm also just realizing that in your role, it's basically like you're getting a business education and a tone, right? Like these, these things that you're doing, like you get to see all facets of the business, which allows for you.
Like, I mean, you're running this agency, but like, I know at some point founders, they go and build more things, right? Yeah. And so it's like, this evolves into that, right? Where it's like, okay, cool. Like you just got to see all these different ways to automate every aspect of these companies.
And like, I think that's the thing that I'm, I'm super cute. Like, I'm a, yeah. Anyways, like it's not just a, like another way that this can be used is like in the, like the customer experience space.
And like intercom just came out with their version of this. I think it's called thin or whatever. It's, it's, it's not super new, but like it's basically like a chat bot that's trained off of your corpus of information.
And it like, you know, communicates with people through your, through your, your chat application. But I'm just thinking about that, like for like early stage companies, like you can provide like a pretty unreal experience, like just by having that there and like give this accessibility to the data, but like it can even go further where it's like, okay, cool. Like we found your email, we went and scraped who you are online.
And then the first message that you get within the application is like, Hey, blah, blah, blah. Like super excited to have you here. Here's some ways that we think the tool could be useful to you based off of what we found, like what you do on LinkedIn or like, you know, your social profiles that we scraped from, or even like your personal website.
Are you doing any like sales automation stuff? I think you're doing stuff similarly like this in Hampton, right? Where it's like onboard really things, but.
SPEAKER_01
Yeah. So I think that there's one company that I'm really interested in. It's called a Synapse AI and they are building a tool that they claim can check someone's LinkedIn profile, scrape all the information there, recent posts, recent jobs, what they may be hiring for, and then turn that into a cold email that you could then send a cold email message you can send to your CRM, your email service provider, et cetera.
So this is something that I think is really interesting because they're offering up to 800 credits profile scrapes per month. Potentially if you pay for the highest tier, but like LinkedIn starts really cracking down on you around like 400, 500. So I'm curious to see if they can deliver on that thing.
That's a little bit concerning to me, but on the sales side, there, there are some really great companies for this. Like you can use, there's one like Tripify, Meet Alfred, Sales Flow, all these things that do automated like outbound, but the real key is like, can I find a, find a contact in Sales Flow using, or in Apollo or clay, finding the information based on a search criteria, or can I, can I export a list from LinkedIn using phantom buster, enrich that data, run it through chat to be T, create it, create a 160 character message, and then flow, flow that information into these outbound messaging techniques, because that's really the difference maker between I'm sending you a, a automated message that has a merge field for your name and your company name that can be spotted a mile away. But when you get a GPT generated content message that has the company you worked for sounds more human sounds unique and also weaves in some of this other data that maybe comes from your LinkedIn profile.
That's the type of content that you might actually open and respond to. So it's, that's, I think where a lot of the puck is headed in, in the cold outbound sales automation, it's just a matter of how you don't get limited on the LinkedIn scraping or, you know, is the account executive or the middle manager, do they have a personal blog? Probably not. So like, what am I going to use to throw into that chat to be T prompt?
SPEAKER_01
I'm going to create something that's interesting.
SPEAKER_00
No, that's super. I saw this recently. It was basically a Chrome extension that would just like write an email and then add it to your drafts.
So like you could be prospecting on LinkedIn, find somebody. I can't remember the name of it, but basically like it would generate, you connected your Gmail account. You could probably build this honestly with just like a Zapier integration or like a make.
com integration, but it basically ingested what was on the page. And then it wrote a cold outbound email, saved it to your drafts. And then like, you could just like, stay on LinkedIn, write like, you know, write a hundred drafts.
And then at that point it's like, okay, cool, I can go send those like, I can send all those from like, from my Gmail or whatever I'm working out of. But I think things like that are just going to man, you could probably disrupt the cold outbound, like cold email. I mean, I'm wondering how you get around this.
So you'd have to use something like a scraping bee or like one of these proxy providers to basically like get that data, but LinkedIn has it so locked down anymore. It'd be so tough. Anyway, so actually super, super interesting though.
SPEAKER_01
I just came across, let me write up, let me get the full name, obstacle character recognition. Okay. So don't know if you've heard of this idea of like taking a screenshot. Computer processes, essentially it's coming out for open AI anytime soon.
I'm sure. But you've got a image to text.io and they've got an API.
So bear with me here, right? Take, imagine taking the Chrome extension. You scroll down to the part of the person's LinkedIn page where they talk about their experiences, maybe you take two screenshots headline and then their experiences last, last three jobs, two screenshots, set it on its way, do the obstacle recognition and that extracts the data. Now you're getting around the ability for a scraping tool to get rate limited because you're just sending screenshots.
And then that creates the content into the Google drive, into the drafts. So I think that's actually got a lot of potential. If you can get the Chrome extension to take a screenshot that, and then that triggers the zap.
I mean, I know a buddy that who put together a Chrome extension, not knowing how to code using chat to be tea. So if you get a screenshot, you send it to the API for image recognition, created draft, that's all in Zapier for sure.
SPEAKER_00
Totally. That's pretty crazy. I think Amazon provides access to this as well.
What is it? Amazon image recognition, but it's basically like, we use this for another tool, but it was like, it's the same idea. It's just like image to text, right? It like extracts the text out of it. Let me find the name of it though real fast.
Amazon and text extraction from image. Of course, it's just Amazon text extract. Perfect.
Very, very, very original name. Man, I feel like there's so many of these that like can be built. Like so you're building this agency.
How do you see this evolving? Like, do you want, like, is this like, you know, one man show until it turns into something else? Or you're like, are you trying to build like a repeatable workflows or like, how are you thinking about the business structure? I'm really curious about that.
SPEAKER_01
Yeah. Two sides of that question, I think. So one is repeatability of the product.
And then the other is the, is the business structure. So repeatability of the product. We're kind of in our dating phase of every industry.
No one's a stranger. Come on in and we'll try to help you. Right. There's the Econ businesses. There's the agencies.
There's the SaaS. I want, I want to experience everything because I know it's better to eventually niche down and become an expert. But ultimately, I think there's room to, to disrupt the strategy consulting management consulting industry here.
Like those guys are going to be continuing to get business from the Fortune 500.
SPEAKER_00
And that's great.
SPEAKER_01
Exact Gardner is something that is definitely high on my, my competitor list. But I think that, you know, you pay attention to the AI tools. You, you are aware of what's going on, but you have the infrastructure of, we can actually build the automations.
And so the way that the business is structured is that you have automation engineers who are, who are just leveraging the commoditized skill of automation. And it's commoditized because make and Zapier, they've got certification programs. So I can go to Annie Jo-Shmo and say, Hey, do you have intermediate two? On make certification.
And they say, yes, they show me a certificate. Then I trust them, you know, 80%. Right. If the name's there and doesn't look forward. Yeah. Great. So commoditized skill there. And then you want to have local, general time zone, good English speaking process consultants that are there and do the management consulting role of, I'm going to get into your business.
I'm going to attend some of your meetings. I'm going to do essentially the job that I do now, but you scale that out and you say, I have one process consultant and one automation engineer. Each of those guys have five clients.
You get a project manager for each little hub. And then you can really scale this thing out. And I'm really interested in just continually hiring young people out of college that are looking to be part of this AI wave and say, all right, show me what you've built in Zapier.
You have to have, you have to know what's possible to be automated or else you're not going to be good at process consulting. Show me something you've built. What AI tool is interesting for you? Are you fun to hang out with? All right.
Let's give this thing a shot. Like I think that there's a real way to scale this out, which is people who are hungry and interested in learning, because that's ultimately like, that's all you and I are doing. We're just learning.
We're just figuring this shit out. You know, like no one, it's so early.
SPEAKER_00
There's no, there's yeah, a hundred percent.
SPEAKER_01
There's just no foundation that is solid yet because everything is coming out every week, new tool, new, whatever, new funding round. I just want to be a sponge. While those things, while that foundation gets set, it's like the cement's being poured and it might take six to 18 months to dry.
So if we're just playing around close, as soon as that foundation's strong, it's so that's kind of the way I look at it.
SPEAKER_00
Yeah, a hundred percent. Man, I think you're thinking about it. So, so like, yeah, honestly, just how you verbalize it is like so spot on.
Really just like put words to the things I've been thinking. So thank you for that. This is why I do these, right? I have people who are smarter than me tell me how to think.
But yeah, like, gee. I think you're right, though, with like, yeah. I mean, you've heard of people say this so many times, but every time there's a technology change or like a base level technology change, like the game like gets rewritten, like the rules, right? Totally.
I think that's what we're seeing right now. Like I'm just looking at content production. Like the idea of a social media manager is like irrelevant to me now, you know? Like that's that job role is probably just going to be an AI that runs.
And then that that's going to be how every company basically functions. Yeah. And like, what are these other things that are going to happen because of that? Like also like lead validation. I just talked to this, a friend that's working at this company, and they're basically using that same scraping idea where it's like, cool, like we've got like contact information, we go find everything we can about them, validate them, like rank them out of eight, you know, or rank them out of one to 10 of like how value or how good of a lead they are for our sales team.
So they can basically like prioritize the reach out based off of like, you know, here's this grading structure and criteria. And then I saw that same idea recently applied to like employee hiring where it's like, cool, like we're looking like here's the job description. Here is their, here's their application.
Sorry, they're CV and then rate, rank them like from one out of 10, like on how, how well they fit this JD, like are they actually like a good result? And then you could rank those people against each other. And like, what does that mean for all of these industries? Right? I'm sure the consulting set. You know, I wish I, sometimes there's some moments where I wish I was 22 again and just like, I would just go make an agency that just like made clips, right? Like just makes clips, like unlimited clips for podcasts or something like that.
I keep saying this to like, and everybody that I'm talking to lately, but like, pick something like what you're doing and you could just make so much money doing this, like, especially if you're value pricing or just like automating a bunch of the workflow. So anyway, Yeah, no, I totally agree. Tell me businesses that you've seen recently.
Oh, go ahead. He's cutting off. Good.
SPEAKER_01
No, you're good. I totally agree. It's really, it's like, get in, get in front of the right people, but use the tools that are existing on the, on the playing field.
Right. It's like, we have ways to speed up these processes so much that have never existed before. So if you put in a little bit of elbow grease to figure out which tool you like the most, run with it on clips, run with it on tweets.
You know, I think this, what you said with social media managers being a thing of the past, like when you get the right inputs for like, all right, I'm going back to the very beginning of the conversation with the chat bot and the, and like, go talking about producing content. So I'm wondering for you as well, we'll try to come back to this if I can remember, but with swell, what if you did a chat bot from open AI, nothing crazy with the API, but you hook it up to 11 labs and you had a voice that read what the chat bot produced and in real time, you have that kind of instantaneous podcast host. Right. That would be the person that interview yourself. Exactly.
And then you just throw that in and both sides of the conversation sound human. You can use it for clips, right? You can do it. You can take the transcripts, you can write content, but like, I've been thinking about this since the beginning.
And I think that that's, that could be a really interesting conversation or like marketing tool for someone.
SPEAKER_00
I mean, that's a product. Instantaneous podcast, right? Is like, exactly. Yeah. If like interviews use you make content like thought leadership content over a week or whatever.
SPEAKER_01
Have you seen those?
SPEAKER_00
Exactly. Sorry, my friend does this and it just, it makes me laugh so hard, but he basically like sets up where looks like he's on a podcast. Like he has like a mic, you know, that's like hanging down and he's like looking off in a different direction, but like talking to the camera.
But he's like, he's like LinkedIn famous, right? And it goes so viral and then people are like, where are the, where's the source material?
SPEAKER_01
Where's the, you know, they're like asking for the whole episode.
SPEAKER_00
But I'm just imagining like, if that's same idea, but you could basically have the person that's interviewing you, right? Me to say, I, so like they, they throw these softball questions to you. But yeah, I mean, there are people that try to get you.
SPEAKER_01
There are people that try to get you booked on podcasts. It's like, instead of trying to get someone booked on a podcast, produce this tool and be like, Hey, I'll get you on the podcast and we'll create the contact. What do you, what do you want out of this? Do you want recognition or do you want the content? Like pick your poison, but I can do, I can do the content.
Totally. I think that's a really interesting proposition. Totally.
I've, so I've seen podcast host.
SPEAKER_00
I don't know how I feel about this. What's up? Now I was going to say, what were we talking about? But yeah, I've seen, I've seen podcast hosts recently start to charge to get the raw materials, like get the pod, like get clips, get, you know, get even a recording. So they're basically like holding the recording hostage.
And then I just had a friend message me and he's like, yo, this company's trying to charge me five grand to get this episode so that I can use it on my other social channels. And I'm like, holy sh, you know, and he's in the different spaces in like a, like the science world, but I was just like, that is the first thing I've ever heard. But like, I mean, like that.
Oh, it's so messed up. It's wild. It's wild.
It's not even talking about distribution to their own audience. It's like, yo, this is just like, just to get the files so that I can then go and repurpose it. I think the ownership of those things are going to become more and more valuable, but also there's so much out there publicly.
It's like, I don't know. Talk to me. Have you guys, I mean, it sounds like you're doing this already, but it's like basically using mixed media sources.
Like we're starting to experiment this experiment, sorry, experiment with this more and more where basically you like, imagine I can provide MP threes, LinkedIn profiles. I can give it like, you know, three blog posts or whatever. And then it basically create new content that from that like positive information.
Like, have you started to do that kind of with, with air table and that type of stuff already as well? Or, or, or, and I think I've asked you kind of a similar question already in this conversation, but I guess I'm just like, yeah, I think I'm trying to beat that idea up more and more.
SPEAKER_01
So I'm trying to think of which multimodal system can take that in an API. Right. Like, I don't know. I haven't, I haven't gotten my topic.
I mean, if you, if you're using, if you're, it's like the multimodal side of it, right? With one call, how are you going to take in an MP three potentially an image, a blog post via URL? That's the, that's the problem is like, I'm going to give it these different sources, which API, which models API allows you to take in multiple sources? Because that's the thing that I struggle with there. Like air table, you can store all of those things up to a certain limit on attachment size, like an MP three would probably run into some issues at a certain point. But it's the sending it to a model that I think poses the issue.
How do you get all of those things to transfer via API? Unless you're putting it all into like, you're in the chat interface for quad or you're in the chat interface for chat to be tea. And you put in the different files and you say, Hey, crawl this website. Now give me the information at that point.
It's not really, it's, it's kind of outside my wheelhouse because you're manually doing the task.
SPEAKER_00
Yeah, totally, totally. I wonder if you could do this in air table though, where it like, it does all of the calls separately and then saves them into like a new cell. And then those cells get combined.
And then that's the like, you know, composite document that you're ended an up sending. I don't know. There's probably a way to hack it together, but it's funny.
So that's like what we're building as well is like, just we want you to be able to just upload anything as like raw materials, raw oil is what somebody called it recently. And I thought that was really good. Love it.
And then we just refine that into like whatever output that you're looking for. Like you give us examples of the outputs that you want. And then it's just like, that can be saved as a template.
And then in the future, we can connect it to your hub spot scheduling or your, or your CMS for the blogs get published. Like we just built WordPress and ghost integration. So you can send the blogs like that we write, like, like directly out of swell to your site.
But anyways, yeah, just like going further and further into that pipeline of just like, I think you made a really interesting point. I just want to reiterate it. Like the goal of this is to get content on, on platforms, right? Like all of the stuff in between that doesn't matter or like leading up to that doesn't matter.
Like as long as it gets to the platform and it's what we want on that platform, like however it gets produced, people don't care. And I think that's the thing that people get hung up on right now, especially like content creators or like, I would call them the content marketers. That's probably how they would think about themselves.
They'd work at these companies currently. Like they, they are still like in the mindset of like, Oh, the process matters to the end result, but it's actually that's not what makes business impact, like what makes business impact is the end result. There's just the content existing on the platform that is like getting you whatever that is, free impressions, traffic, et cetera.
Like that's actually what's making the business impact. So I think that's just going to be a mind ship that happens. But I mean, that's, it makes total sense though, right? It's like the, the, the tools to do that have changed.
Like it's not, I think people are thinking, I don't know, personally, how I think about AI more and more is like, how do I get it 80% of the way there? You know, like, how do I take again, those raw materials and get it like as close to that finished product as possible and then refine so that it gets closer and closer, but not like, not obsessing about, like it doesn't have to be exactly perfect. Like I was just talking to somebody today and they're like, I don't care if it's like, you know, I just don't want to listen, like, I don't want to listen to a 45 minute episode and then write a, a, a three paragraph summary. Like that just sounds like a massive time suck.
So if you can just get me, you know, 80% of the way there, my team can basically touch it up like the rest of the way. And so anyway, just that idea, but applied to everything is where it feels like this is all going. So.
SPEAKER_01
No, I completely agree. I think that the human, the human in the loop will still be there. But when you eliminate 80% of the time, it's like one person can now manage, either multiple accounts, multiple, who knows what, but you, you cut out so much of that initial brainpower of drafting and the first edit and like, maybe I sent it to another friend and then I do another round of edits, but like getting that first thought out is a thing of the past, right? And I think that that's where, if you extrapolate that out to as much of the processes, you feel comfortable.
And that's the other part of it is like, where do you want to draw the line? It's not so much, you know, what your manager wants or what the end, like, yes, we need to get on the platform, but where do you want to draw the line is also going to be an important part of this. And some people are just going to say, give it to me on the platform. And I speak and it's on Twitter 24 hours later.
Some people might want that. Some people might want to be the, the artist that does the 20% editing or 40% editing or everything comes from the dome and it's, it's human 100%. Who knows? But I don't know.
It's that side of it of it's, it's not the AI that's going to take your job. It's a person with AI that says 80% of this doesn't need to happen. I'm my human intellect can cover the last 20%.
SPEAKER_00
Are you getting a lot of people that are like fearful? Like when you come into these companies, they're like, shit, am I going to lose my job? Like say, I see a CEO hires you or something like that. Like, is that a concern? That's like, I mean, you're probably insulated from them, but I imagine that there's like, they're concerned that it's basically just going to automate away, automate away their work.
SPEAKER_01
Honestly, it's not my clients. It's the platforms themselves. That's where I see the concern.
It's, it's the, it's the new AI tools I look at in their FAQs. It is an FAQ that I see a lot as will this replace human jobs? That's the thing that's a little bit like, ooh, Twilight Zony, you know, it's like, they're anticipating those questions because they're getting them. Obviously, it's like, it could replace a job, but more effectively, it's going to have one person with this tool or two people with this tool that do the work of 10, that's going to be the thing that, that ends up solving the problem, right? And the problem being inefficiency or getting things done faster.
So that's where I see the concern is the tools are saying themselves, no, we're not going to replace humans. We're going to augment humans. I think at the end of the day, the business owners just want lower OPEX.
SPEAKER_00
Yeah, exactly. Like it's, it's just, it's a cuss cutting funk or cuss cutting function. That's, yeah, I just talked to this company and they, we just, we just got the AI to do this, which is still crazy to me, but it basically like, like a raw conversation that happened with an individual.
And then it's like, build a narrative arc based off this. And then that gets handed off to an editor. And then the editor basically like has these points of relativity to look to like clip the things to like, create that narrative.
And then within these like segments, so imagine it's like minute one to minute three, a minute, five to minute eight, et cetera. Like they want a narrator to be able to basically transition you between those two, like clips or thoughts. And so then we have the AI then write transition scripts so that the narrator can be basically like connect these ideas and create again, to align that narrative arc of the whole episode.
So imagine it's like, I'm, you know, it's basically like a, they called a narrative podcast, but this is like something that we're seeing be possible. And, you know, they're, you know, and when I initially started the conversation with this company, it's like they, they're telling me that this is, you know, there's no way that they, this could be used. Like AI could be used because it's just so human heavy, right? And then you come in and their workflows and you're like, Oh yeah, like here's all these things that we can abstract.
And it's like, this is exactly what you're doing. Like a grandpa, right? It's like, show me what you're doing right now. And I bet you I can automate 80% of it.
And it's that same, it's that same idea, I think. But yeah, again, I mean, that's on the media production side. Like now it's, now it's like, look at, look at all the, like, look, I think the, the top of the toughest part that's going to happen from all of this is going to be like, how do you make it flexible enough to fit into everybody's unique workflow styles? I think Zapier and like, you know, what you're talking about, the stack that you use has made it very like flexible to do that.
And that's something we're really trying to think about, like, as well as, like, how do we make it where it's like, because everybody does everything different is just what we found, like with any content or media production, they both like, every, there's no, there's no standardized format. Right. Like you don't just like, this isn't how you, you know, it's not double entry bookkeeping. Like it's not like there's this thing that you do and you do it this way.
It's like, Oh, there's a, you know, a million ways to get to the same, to the same output, or, you know, to get to the same, like whatever, whatever the thing is that you're trying to create. And so we're basically just trying to figure out, like, how do we, and that's, I think that's going to be our biggest challenge is like, I'm thinking through this, but I'm pretty excited though. So it's another random thing, but like, we're starting to work with, or experiment with like prompt chaining is what I'm calling it.
So like, this and then do this and then do this and then do this, but within the content production workflow, are you like, I imagine with what you, you're building, that's like a lot of what you're doing is like, it's kind of almost these like, if thens, and then basically using the AI, like, are you, are you chaining the AI's actions, like together, like, and what's. Like, have you done anything like that recently?
SPEAKER_01
Yeah, I think that's a really good, really good conversation because prompt chaining is really going to be what prompt engineering becomes like the future of, I guess, you know, you can, anyone, anyone can ask for a prompt. You can ask JetDBT like, Hey, write me a prompt to write a tweet and it'll give you something, it'll ask you questions and it'll give you something. But like the ability to take an input and, or take an output and use it as an input into the next thing, like the real, the real important thing that I found with prompt engineering is that you cannot have multiple outputs for the same prompts.
Like, JetDBT is much better and much smarter when you give it a specific set of instructions with an example and it's a single output. So for instance, take a podcast, write it, write me a summary. Okay, great.
That's the first one. Take the summary in the second chain, second prompt, take the summary and write me a tweet thread. Okay, great.
Or write me a LinkedIn post easier and it needs to be snappy. It needs to be in this tone, whatever. Great. Then the third one is take this LinkedIn post and write me a hook that has this type, write me 10 hooks that have these different formats and then deposit those hook types into this cell or this whatever where I'm going to read over them and see which one I like best. And then once you have all those hook types, you can combine all of them and like you can have the human in the loop say, I like option two.
And then the last step is, okay, put the hook, put option two hook on the LinkedIn post, schedule it on typefully. Like those are the types of chains that I think are going to be really important. And there are two companies that I'm following very intently on this.
Let me check my little database here, products, prompts, management. All right. So the two ones that are interesting here are prompt folio and, oh geez, where'd it go? Prompt folio is the one that I know of.
I'm in talks with the founder. I can't, I can't define the other one right now. But this field of like SAS for prompt management is going to be big for sure.
Because these guys need to think about prompt, prompt folio is taking the integrations approach of saying like, here's the variable. I'm building a prompt in the prompts folio environment. And then I am able to test and refine the prompt just like you would in chat, BT, and then I put in variables.
And the variables are where when I plug this into Zapier, I plug this in the make, I'll take in the information from air table. I'll take in the information from Apollo. And so you get to build it in a nice stable environment, but then you get to plug that in and use that prompts for Zap one or Zap three, but you can maintain the prompts in prompt folio.
Does that make sense? Yep. Totally, totally. So that's really neat for the integration side of things.
And the other player in the market is going after the team side. So they have their environment for teams to collaborate and show revision history and, you know, Cody made a change here. Grant deleted this almost like Google doc editing tracking.
They're doing that for prompts before the prompts. Yeah. Exactly. That's going to be huge, right? Because all these, there are going to be so many different use cases.
I see AI is like right now the simplest thing is, you know, in Hampton, is this person a parent or are they not a parent? If yes, automatically responded Slack saying, check out our parenting channel. But that's a human decision. Open AI can do that in a heartbeat, right? So totally.
You take, you take the binary outcomes, so you scale them up to prompt chaining, writing content, writing a good hook. Those things need to be chained because it's too many outputs for one really good piece of content. So chaining it, getting better.
That's, that's going to be a big piece. And there are players there that are going to make really great software. I like prompt folio a lot.
SPEAKER_00
Nice. How do you spell that, Grant? I was just trying to look them up and it was like another company was like prompt folio.hu came up, but the website wasn't working.
I don't know if that's them.
SPEAKER_01
Honestly, they're still in beta. So I don't know if their website is live, but it's prompt the word and then folio, f-o-l-i-o.com or don't even know if they have a website.
SPEAKER_00
That's fucking amazing.
SPEAKER_01
I'm in their community. Oh yeah.
SPEAKER_00
Well, that's perfect. I think that, I think that honestly is the perfect segue out of this. Like all this stuff is early.
Nobody knows what they're doing. And like, that's the opportunity here. So thanks again, man.
For coming on. I'd love talking to you. You always get me so hyped on this.
So, but yeah.
SPEAKER_01
Nah, same man. One other thing. Nah, just that I think the marketing stuff, like there's, there's going to be so much potential in industry disruption, role disruption, job description.
Like it's all changing. The whole field is moving from consulting to marketing to sales. Like there's so many different moving pieces right now.
And the only advice is just to keep an ear out. You know, it's all you can do. Everyone's just trying to figure out what's, what's working and what's going to, what's going to get adoption.
And as you say, distribution is more important than product.
SPEAKER_00
I love it, man. I love it. Well, sweet.
Where can people find you?
SPEAKER_01
Yeah. Twitter is at Grand Hushik. LinkedIn, I think is my name.
Yeah. Grand Hushik. And then a grandpa.
co. That's the website. So trying to build a little agency here.
Awesome.
SPEAKER_00
I love it, man. I'll include you in the show notes. And if you need automation, go talk to Grant.
This man literally builds the most gangster shit I've seen. He's inspiration to me. So anyway.
SPEAKER_01
Doing the best we got. We appreciate it. Thanks for having me, man.
Talk soon. We'll talk soon.