When Algorithms Rule: AI in Digital Marketing Wonderland

Alex Jonathan Brown: Ladies and gentlemen, esteemed colleagues and curious minds from around the world. Welcome to this ground. Okay, so normally when I start these, I love to do a cute little thing and we talk about coffee and it's great. The reason I did that thing, which honestly, I didn't love, thought was kind of bad, is because we have turned huge chunks of this coffee break over to AI. If you got here through our webinars or through a social media post, there's a good chance the thing you saw was written by AI. What's the future of AI? Are all of our jobs going to be taken over by the robot from the Jetsons? Probably not. But to make the case for humanity and talk about all the great, awesome, amazing things that AI is going to do for your business. We have brought in two of the smartest people I've talked to about AI in my 40 years of being alive. From Iron Horse, our CEO, Uzair Dada. And I'm so excited. I don't usually get to read like bios. But Umair, I'm going to read a chunk of your bio.

Umair Akeel: Oh, no.

Alex Jonathan Brown: Yes, I'm sorry. I have to. Umair Akeel, The CTO and co-founder of All Good HQ, is an accomplished engineer with a proven track record of delivering cutting edge products to market. Just watch him squirm as I read this, so great. As the distinguished engineer at Twilio Inc, he spearheaded the development of the Twilio AI for voice products. See, that's why we got him here. Where he worked on early customer discovery, iterating the product with pilot customers and helping close early sales. My favorite part of his bio in 2012, he won the Eloquan of the Year award as Eloquas was chief architect, including during its IPO. Uzair, I can do one for you if you want. But either way. Hi everybody. Welcome to Coffee Break.

Uzair Dada: Hello. Hello.

Umair Akeel: Hey. Hey. Hey.

Alex Jonathan Brown: So let's jump into this, I guess. In the course of planning this, we have a lot of pre-meeting chats for those of you watching. And every time it's just been like, I have some vague ideas and then these two take off and run. So there's been a lot of talk about AI and I think a lot of that talk has focused, especially in the last month or so, around ChatGPT and specifically how that's kind of opened up more people to think about AI. So I think that's where I want to start, is let's talk a little bit about how AI has gotten more accessible and what that means for, I mean really everybody, but especially our businesses.

 

Umair Akeel: I'll start because I think Uzair and I talked about this in this one word you used, accessible. I'll talk a little bit about accessibility and we also have to talk about framing. But the models behind all of this have been available to developers via API for a while. Now, in my audience here, I don't think people are using rest APIs against OpenAIs available models to try these things out. It just became more accessible. Oh, we're doing my intro too here?

Alex Jonathan Brown: Yeah. Yeah.

Umair Akeel: I should tell you what I'm doing. Yeah, let me backup. Yeah, hold on.

Alex: You got your slides ready to go.

Umair Akeel: My slides are ready. I am the CTO of All Good. We are introducing this concept of adaptive campaigns. And basically what it is All Good helps marketers grow their pipeline, revenue pipeline by focusing on an increasing middle of funnel conversions. And what we do is, we automatically adjust lead nurture campaigns based on user activity inside and outside of your marketing systems. That's who we are. I take ownership of building your traditional campaigns. I launched Eloqua campaigns a few decades ago. I am sorry for those of you that have to clickity-click everything.

Uzair Dada: Hence the name, All Good.

Umair Akeel: All Good. Yeah. Yeah. We're fixing some of those mistakes

Uzair Dada: Trying to make good from past mistakes. I like it. I like it.

Umair Akeel: We learn. We learn. And yeah, basically we're moving from traditional campaigns, which is a lot of manual setup, a lot of fixed and linear flows, to things that are automated, things that are bespoke. Most importantly, adjust automatically and nonlinear. I don't want to talk too much about it, but as you can imagine, there's a lot of AI behind the scenes, and we don't make shit up. But it became accessible. And this is the key. You could go and type something and it would give the answer. That was it. That was the entire interface. Framing is then the biggest problem. So Uzair, you and I have talked about this, why do you think framing is important? And I want to bring that up in the context of this when it makes shit up.

Uzair Dada: Yeah, so I think it's accessible a hundred percent. AI has existed forever, probably at least in the last decade, and getting better and better. And we probably all use it in some way, form, and fashion. Especially all our marketing teams and sales teams that are sort of doing stuff is largely AI enabled and in the background. We didn't see it if you were not a developer, if you were a practitioner, you didn't understand it. So it was sort of inaccessible to us. And ChatGPT as an interface made it accessible. I think the other thing that happened was it was something that you could understand. So it was transparent. I could touch it, I could talk to it, I could engage with it. And someone as a layman understands that it's a request response even though it was not totally accurate all the time, hence the term, it's make shit up, and it does make shit up.

I could understand when it's right or wrong. So I think coherence was very, very important. And I think to me the biggest reason it helped me as a marketer, was the ability to frame conversations. So if you think about it, we all work in multi-channel, multi-functional teams in an organization. And let's take some contrasting stuff. Maybe marketing and sales can speak their stuff similarly. But then when I'm talking to a content and creative person, or I'm talking to a product person, and God forbid I'm talking to an engineer, it's like you're talking completely different languages. So the ability for me to know very easily communicate and frame up what I'm saying to a front end developer or a security person that what I'm trying to get done in a way that may be there because I can personify it, putting that information into ChatGPT or whatever else, to me has been revolutionary.

I think the other thing I would say is most of us can talk, but fuel us can frame it in a way that it's coherent. So it's a bunch of gibberish because we just like to throw up or we don't say anything and we're sort of say two words. And neither one gets what you are trying to say across. And by going into something like saying this and say, how do I build a plan to do a webinar nurture? How do I plan to do X, Y, Z and getting some set of a structure that I can now augment brilliantly?

Umair Akeel: And see this is the thing. For those of you that haven't used it or starting to use it, one of the key things, and I think those of us that play with it is, when you want ChatGPT to frame things in a certain way, you can literally ask it to you. You can say write in the style of Steve Jobs, or write in the style of Lex Friedman and it understands that. And it can take your ideas and present the same bullet. I use a lot of bullet points. I'll throw random bullet points in and get answers back. But I'll put a lot of time and thought into who do I want these answers to speak to. That's the key. When you do this, and I know you've been using this at Iron Horse, how are you guys looking at hallucinations, which is the making stuff up, because that's one of the biggest challenges. That's like the pitfall. It makes stuff up.

Uzair Dada: It totally is. So I think that's where it's, I think, it's a good sort of structural thinking and maybe we move that. I don't know what the next slide is, but I think we were sort of talking about hallucinations in general. But I think the core piece is knowing you are not trusting AI to come up with something that is just automated to just go flow through. What you're doing is using gen AI to help you do what you're doing faster, and hopefully, not do the mundane stuff that a lot of us don't like to do. And so to me it's sort of an accelerant really is what it is. It gives me a better place to start, but then it allows me to actually apply in a better way and not be wasteful. I like doing research. But if someone gave me something that was curated that I could work from, that saves me time.

If someone sends me an email that's super long, I'm like, what are you trying to say? Versus if someone frames it in a way that they know that I like to see things in a bulleted way or I like to see things in a timeline manner, it's a lot more easy for me to apply and do things. But it still has to be the human part of this positioning is super important. I think we were chatting about it in an interview in 60 Minutes the other day where he said that two-thirds of the jobs are going to be different in five to 10 years. I think it's pretty awesome. But at the same time, the example he used was in 10 years a radiologist will be still doing their job, but they'll have a super radiologist bot next to them. And the super radiologist bot is just there to say, "Hey, maybe you missed this thing. Is there something here?" They're not making the decision. They are a super assistant.

Umair Akeel: But the same is true for development. Those of us that have been using GitHub Copilot, or Codex. So there's a ton of startups that are providing the same thing. I don't think you're replacing the developer, but the developers that are leaning in and saying, "Hey, I don't need to write all the code. I know what I want to write. I know the problem I want to solve." You can start and iterate.

Alex Jonathan Brown: One thing that comes up immediately when we start talking about using these tools for work is how does privacy factor into that equation? For those of us who are looking to use it a little more at work. But obviously there's issues with intellectual property, and PII, and all these different angles to that.

 

Umair Akeel: Yes, like I said, this is an important part when you use it as an accelerant, you have to understand privacy. This is an area I spent a lot of time in, both at Twilio, outside of Twilio as an investor advisor to a company focused on this space. If we go to the next slide, terms of service are important. And this is just one set of the terms of service. They've already announced that they're coming out with more business friendly clauses. But you have to understand ChatGPT and a lot of other providers have fairly strong rights to your data, to your interaction to train these models. So if you're using it from a vantage point of a company, you have to think about what data am I putting in here and what are they allowed to keep?

The way I think about this is two things. One, understand the use case. So if you're putting in legal documents, maybe talk to a lawyer before sharing legal documents with ChatGPT. Because if you haven't told them not to use your data, they can use your data because you've allowed them to do that. And two, just understand that you have the access to control this, whether you use Microsoft Azure's APIs or in the next slide I'm showing you, there's another product coming out. This is private AI. They're building redaction right into the ChatGPT type interface where they'll redact PII as part of your chat. These are the type of things that exist.

Uzair Dada: That I think is super cool, So I think what Umair is heading on is data and privacy has to be paramount.

Umair Akeel: 100%

Uzair Dada: Today, because these tools are still in their infancy and largely alpha, beta. So you have to be very careful what you're putting in. Because they are still learning and they're being open about it. Some are are not. So in the interim you have these interesting applications like what he's showing right here, is the ability for you to write and talk like you do, which is like, "Hey, I'm trying to send this email for the upcoming new holidays or new policies for HR that I'm rolling out. How do I write it?" Here's the information on the company, here's the type of stuff I'm doing. This tool extracts, what is it 40 or 50 different elements of PII out. Sends it to ChatGPT, gets the information, cobbles that together and gives it back to you. And when we were talking about it yesterday, you're saying, yeah, that's cool. It's not giving that to OpenAI, but these guys still have access to your information,

Umair Akeel: That's right? And so you need to read their terms of service. Now, their terms of service, they don't use any of this data for their training. And that's the thing, everything is about who has access to your data not to process the data. Processing the data means you give them an API call, they send the results back. But who has access to keep your data to train from your data? And a lot of times that's when you as a company need to decide, am I using them as a processor or am I allowing them to train on my data? And in many cases, you might want that. You might want OpenAI to train models for you, for your business. And they have contracts for it and they will custom host models for you that will make you more productive. And let's be real, if we assume how things are progressing and you are not part of this, your competition is going to do this and move faster than you. It's all about efficiency and productivity. If your competition gets more productive, you are in trouble.

Uzair Dada: I was talking to a very, very, very large senior partner at a law firm. And I asked him this question and he heads up data on privacy. And he said, "If we don't do it, we've already lost because all our competition already is." And he was referring to using this tool for legal firms from a research perspective, from a framing perspective. So I think again, being cognizant of not everything we do is PII centric. Part of it is, it's like the analogy of the Google Translate. When I'm traveling and I'm starting to speak French and I don't know French, I can say English to French. If we can just start using for the base use cases that it's really good at already, we're already so much further ahead. And there are other things that are still a little sketch. And I think those are things that we need to figure out overtime and it's only going to get better.

Umair Akeel: Yeah, I mean you said it. I've thrown pieces of legal text in. Nothing that contains PII. But just a paragraph you saw it and say, explain this to me. How does this apply to somebody that lives in California?

Uzair Dada: Yes.

Umair Akeel: And ChatGPT a week ago would just give me the answer. If you try anything that looks legal, it now throws up a warning saying I am not a lawyer. This is literally last week. Maybe Uzair's sending all of his legal paperwork-

Uzair Dada: It probably was.

Umair Akeel: Yeah, it was probably Uzair.

Uzair Dada: Probably was. Yeah, I was-

Umair Akeel: So now they say that.

Uzair Dada: Yeah, I was looking at my lease renewal. I got this 20 page document and I was like, I have no idea what I should be looking at. I have no clue. So I literally redacted the personal information, I threw it and it's like what is important here? And it was the ability to distill. So I think we talked about the framing, which was the other way. This is distilling someone else's information to me so I can not comprehend. I think it's a two-way part that's, to me, just awesome.

 

Umair Akeel: And again, some of the things you learn when you use it more is you throw in the document, you ask the question, but then you can refine the question. I'm interested in understanding the security implications of this document I shared. Or no, I'm interested in the cost implications. And as long as it's still got that context in its window, technical nerd point here, it does have a context window. Do not lose that. It is not invaluable. Hallucinations is when it makes shit up. It's just spewing random text. It happens when you're an expert and you see it happen. It's always funny. Context window is if you put too large of a document in there, it only remembers a piece of the document. So this is why it's good that OpenAI now says, I am not a lawyer because a lawyer remembers all 30 pages that it reads. OpenAI might not remember all of it. And again, GPT 3, GPT 3.5, GPT Turbo, each of them have their own.

Uzair Dada: And so I think that's a-

Umair Akeel: He doesn't let me talking tech. He's like, "No, no, no, don't talk."

Uzair Dada: No, I love it.

Umair Akeel: It's important if you're using it to kind of like ideate. Start new smaller sessions, use smaller things and you'll find it be more aligned with what you're trying to do.

Alex Jonathan Brown: I think that's one of the things is especially if you're not as technically inclined, where even in the last couple of months we've seen the conversation on a consumer, for lack of a better term, level start to shift. And so I think you agree because we have a slide for it, that if you're at the stage where you're thinking of this as just the capabilities of ChatGPT, there are other conversations that are happening on and about other platforms. So I wondered if we could take a little bit to talk about some of the other things that are out there that are really interesting to you now, or at least worth taking a look at.

Umair Akeel: Yeah, I'm not even going to go into the technical ones because on the engineering software side, this space is moving faster than I've seen anything other than mobile a couple decades ago. I remember when the iPhone first came out, it was this crazy. But on the ChatGPT side, let's talk about a couple of things. Sorry, I've got to just... Can you guys hear me still?

Uzair Dada: Yes.

Alex Jonathan Brown: Yep. You're good.

Umair Akeel: So step back ChatGPT came out by OpenAI. OpenAI is the company that builds the model behind it. But now that their APIs are out and the cat's out of the bag in terms of how well it works, you've got a lot of smart people building on top of it. So if you go to the next slide, you are still telling the thing what to do. There's this really cool thing which is AGI, general intelligence. Now we're much further away from AGI than a lot of people will claim. But there are these products out there like AgentGPT, where instead of you chatting individually having one turn conversations, you can give a complex set of tasks to AgentGPT, and it starts to reason and it starts to create more things for itself to do. And it keeps kind of coming up.

I gave a question about, "Hey, I'm talking to Uzair at Iron Horse about this. I want you to go research ideas that we should talk about at a coffee talk to do with ChatGPT and AI." And you can see it starts with a big list of things. And then it starts to say, "Well, maybe you should talk about this. And then for this, here are some articles that you may want to talk about." And it can access the internet. It's just mind-blowing. Now, it can hallucinate, it can make shit up. It forgets its context completely. But it's insane that it reasons and starts to do multiple things.

Uzair Dada: Yeah. And compare this to today, we would Google search, you'd probably look at the first three articles and probably stop there. So we're looking at this level of insight and research. Maybe we went to a couple of websites, highly linear, probably skimming. And so we're sort of doing, even though in our head we are doing what we should be doing, we're doing it at a 10% level at best. So the ability for us to be more thorough, the ability for us to have better insights on our hand as a starting point still needs to apply the same logic that we do because when we go to certain webpages, we said this thing sucks, this is stupid, this is not right, let me go to the next one. Even though Google showed it as being the drop one, it's not right. You still have to go through the cerebralness of is this right? Is this wrong? And make that determination. But the fact that you have something so cool where it sort of completes a task and creates its own new task to get to an end outcome, whatever that outcome is...

Umair Akeel: Whatever it is.

Uzair Dada: ... It is freaking awesome.

Umair Akeel: I mean, we do this right now every time you get on a customer call, what's the first thing you do? You go do a little bit of research on the customer. What do you read? You try and read the same piece of the content, and try and summarize. But if now the machine knows what I'm looking for, and can do some of that research, and just present it in a way that I can consume quickly, this starts to change how I work. There's another one, I didn't bring that up because it's all code. But you literally give it code and you tell it, "Now explain this code to me." And it knows what I understand and explains it in the context of what I understand. How crazy is that? It just saves me an hour of time, right?

Uzair Dada: Yeah.

Umair Akeel: Because now I just dropped something in and it just did the work for me.

Alex Jonathan Brown: An amazing use case for that as the copywriter on the call, if you want to have a blog post about that technical stuff and you don't want to tie up your dev for six hours trying to explain it, take that code, have it be explained, and then have your writer write about that. We've got more on our list of cool stuff though. Do you want to talk about Auto-GPT for a second?

Umair Akeel: So this one is interesting. I think we should go to the next slide because it actually gives the recipe, and then we can come back to the nerdy stuff. It can create recipes. And one student basically took this and said, "I want to create a new recipe based on this criteria." And it can go reason about what you have available, what goes with it, what preferences you give. And it'll come back and it'll write up this lemon berry crostini with goat cheese. I have not tried it, so I don't know if it's hallucinated and made it taste bad. But it looks like any recipe that I would get from a recipe book. Where this is becoming really interesting. And ChatGPT has also announced plugins recently. So in the past with ChatGPT, you just talked to the machine. With plugins, you can actually ask the machine to do stuff for you.

So if I was using ChatGPT, I could actually make it order these ingredients for me. I can make it book travel, I can make it do really hard math, because now the plug-ins can access portions of the internet. For those of you that are on the marketing side, you've seen Dharmesh at HubSpot, he's been doing some amazing demos on some of what they're doing. Now all of a sudden, imagine ChatGPT can read the internet, can connect to your systems, read and understand those systems, and then come back and build it. The last slide was a lot of the really cool technical enablement. But to me the key here is it can read from the internet long short term memory. I talked about context windows because it forgets stuff. Now people are solving that. They're like, "Well, what if we could make it remember?" And then again, you still have the access to GPT 3 and GPT 4. It's moving so fast I can't keep up.

Uzair Dada: I think there's that and I think there's a question that came up in chat saying there's all these free services, why do we pay for things like Jasper. And no, the free services are alpha and beta. These are not free services forever. So I think there's a lot of experimentation going on. So just again, these are not products yet. These are concepts

Alex Jonathan Brown: From a user standpoint, early stage

Umair Akeel: This is early stage.

Alex Jonathan Brown: From a user standpoint, that's really important to keep in mind too. If you are building your workflows, if you're committing to projects based on, "Oh, I know the rough draft takes me half as long because I'm using ChatGPT," and you're not factoring the eventual price of that into your thinking, I think a lot of people are going to run into some pretty $20 a month realities.

Uzair Dada: Not even the price. Price is just one thing. The interface completely changed. I don't know if we're going to show Alex the Mutiny video that we were talking about. Mutiny is one of my favorite new tools and for web personalization, they've got some just really awesome stuff and they've made web personalization very marketing friendly rather than developer friendly. And one of the cool things about Mutiny is that beyond sort of making the core part easy, when you go into personalization, it has incorporated sort of ChatGPT inside of it. And so when I'm looking at a text headline or when I'm looking at a paragraph on a page, I can ask it to give me a recommendation. So before it gave you six choices.

Yesterday when I was going to... Alex said, "Hey, can you record this video for me?" As I went in, the interface I had the day before it was gone. So I was like, okay, this is cool. This is actually cooler. So this is an example of our page. I went in on the page looking at our site, it's go in. And I picked up our who is Iron Horse description? It's there. And then I said, "Hey, cool. Take this and make this targeted for the developer marketing audience. And so what would that look like?" So in line, I didn't have to go to Alex, my contents route just to go do this. It gave me a recommendation. I'm like, oh, this is pretty good. And in line I could sort of click this and go apply it. Same thing is happening in Canva. Same thing is happening in Notion. The AI integration in your normal workflow is that thing where you'll see some new interesting use cases that are going to be easy to adapt.

A lot of the tools we're seeing today are sort of sit on the outside. And they're cool from an experimentation perspective, but it's like a notepad. It's not my core application. It's kind of a cool thing on the side. I keep my to-do lists there. I experiment there. I keep my notability on my iPad, but it's not the core. I think it's the application integration that's going to be awesome. So the other one that I think I love, love, love, love, love, and I'm so excited about is Adobe Firefly. So this is a cool video. Let's play it through. So the ability for us from a creative perspective to now do stuff. So we talked about all these interesting content use cases and we've been using that. But the creative use cases that are coming out, and this as an example of me being able to change stuff up on existing things, be able to contextualize it.

I think there's other examples of being able to create different versions of it, is just absolutely mind-blowing. And this is again an example of historically to do certain things. I may have gone to someone and not being able to explain, but perhaps I can now explain better and then give this to someone to evolve. But the fact that even for someone like someone on your creative team to be able to do their job better, to be able to search things better, to be able to do production tests that, trust me, most designers don't like to do. Do it easy, do it faster, and do it in context because now the back and forth has gone away. So just think about just the efficiency of the workflow and the reduction of random tasks. That to me is the best and awesomeness example that I've seen on the creative side. This is not all yet. There's some beta stuff that Firefly has. But some of this new stuff that's coming in, man, it's radically cool.

Umair Akeel: Yeah. I mean OpenAI, DALL-E, Midjourney, Stable Diffusion. These are all on the image space. And again, it's easy to iterate. With a designer, you know somebody's going to spend six hours doing something. So you want to have a really good thought about what I want. I want to see a dragon flying on top of a pigeon. Two words, Brooklyn Bridge, because I want to see what that might look like. I probably wouldn't ask a designer for that. But I asked DALL-E. Yeah, I asked DALL-E and they made it. And I'm like, no, dragons don't write pigeons. They don't look funny.

Uzair Dada: No, the designer would've said, "No, that's a stupid idea."

Umair Akeel: That's true. That's true.

Uzair Dada: So I'm saying that the fact that it's allowing you to communicate in a way because... And designers' probably right, they've got the experience. But the fact, the experimentation that this allows.

Umair Akeel: Exactly, and that's the key.

Uzair Dada: That perhaps we are resistant to do because it takes cycles today. I think that's key. Yes.

Umair Akeel: It's the iteration. And then I think Olivia asked this question just to make sure I address that. These are not free services. So I'm paying a lot of money for a lot of these services. Microsoft Copilot, GitHub Copilot charges money, ChatGPT for the pro version charges money. A lot of these other tools charge money. So it's not exactly free. But again, coming back, Midjourney, Stable Diffusion, DALL-E too, out editing on DALL-E to kind of what Adobe Firefly. It allows you to iterate so fast and it allows you to very quickly explore the idea space. You might still want to go to a graphics designer when you have the idea you want to stick with. But it's a few clicks to try random crazy things. It's insane.

Uzair Dada: So as I tell my 18-year-old daughter at San Diego State who is in a sorority and she's like, "Oh, dad, I went to my sorority and they had free dinner today." And I said, "Honey, it's not free. It's included." So I think it's when we're talking about why do you want to pay, no, it's included. You're paying for these services. These services are just evolving. So I think it's like we need to think about it. But I think that's a really good question to think about.

Alex Jonathan Brown: So we talked a lot in the last little bit specifically about the creative possibilities of AI. And I think for a lot of people who have job titles like senior copywriter at Iron Horse market, that's me. I think there's been some panic at various stages. And so the question that I think is really interesting is for people who are running businesses, who are making strategic decisions for businesses, how do you see AI shaping the way we work? And then specifically who's doing work for companies?

Uzair Dada: Yeah. So for me, I see it, as I said, making us all less frustrated because we each have frustrations with each other. Sometimes we don't talk to each other the way you want to talk to me and I want to talk to you. So that the translation error, the telephone issue that we have, I think largely goes away in a good way if you start using it. Second is it starts doing some of the things we don't like doing. So it makes us start at a better point, hence we are more efficient. Third, it actually allows us to do things that we wish we were doing but we didn't do because we thought that was wasteful.

The example we used about sort of the pigeon flying over Brooklyn Bridge, blah, blah, blah, yada, yada. It's like, okay, that's super cool. Yeah, I want to think about what that is because it sort of curtailed my creativity or curtailed my ambition because in the name of efficiency, you just made decisions. And perhaps there are some interesting nuggets that I ignored. So to me, that's what I'm fascinated about. To me, I don't think it's eliminating the jobs. I think it's making jobs really more fulfilling to what we all wanted it to be.

It's like I don't like cleaning my house, I don't like doing laundry. I wish I had a bot that did it for me. It's sort of that same thing. So I could kind of hang out and do more things that I wanted to do, but I still have to do that. It's the same thing. If I could make things that I don't enjoy as much more efficient or made sure that I had the super radiologist bot so I could do my job better, I think we would all be fulfilled. Will it impact certain jobs that are highly mundane and repetitive? Yes. But I think those people will probably get reskilled to do things that they probably prefer doing tomorrow.

Umair Akeel: I think I'm going to take a slightly different angle here. And I think this is what your question is, what will they be doing? I have no business thinking about pigeons and dragons and Brooklyn Bridge pictures. I'm an engineer. Similarly, a graphics designer may feel like in the current world they have no business calling APIs or writing programs. But I think that blurs. As a graphics designer, I may now be able to have the confidence of trying to write some code to automate something that I would love to have done. But I just didn't have the skills. I knew the problem I had to solve. I just didn't have the skills. It's kind of like I'm a huge fan of Zapier, Zapier on steroids. And by the way, Zapier has also integrated ChatGPT into this. The number of tools that are integrating this capability into analytical systems.

Everybody on this call at some point in time has thought about, I wish I could run this SQL query without going to my data engineering team. The whole reason I started the startup I'm doing right now was because that was the problem. Marketers cannot get data out of all the systems they need. But that's going to change in the next few years. Yes, you're going to be able to do that. So it's not just about becoming more productive in domain. But it's saying, Hey, can we solve the problem better if we could go out of domain as well?

Uzair Dada: I think you were talking about it takes away the linearity of our org structures today.

Umair Akeel: That's right.

Uzair Dada: Today I feel like I can do this and then I have to go to you to do this, Alex, or I have to go to someone else. I have to go to Ezra to do something else from an engineering perspective. And frankly, when I go to Ezra to do some engineering stuff to say, "Hey, can you fix this Marketo template," or whatever else, it's like, "I don't want to do this. Why can't you just do this?" So I think there's all of these edge cases that are, I'd call it noisy. I do it because I wish you did it. And so it, I think, makes us sort of more complete. And when we talk about sort of collaboration, I think it makes it more collaborative. Because again, I'm speaking to your language. Maybe you are the decision maker on it. But I'm giving you something you're like, yeah, this looks like it could go.

Umair Akeel: Right. And remember, this is what went five months from the launch of this tech.

Uzair Dada: Yeah. Yeah.

Umair Akeel: It is important to again, go back and be, this is month five. Yeah, GPT 3 with the Da Vinci models came out last year. But only a few people had access to it or only a few people started exploring it at that point. The explosion is happening right now as we speak, And Google just came out with Board. So ChatGPT is not alone. I've been playing with Board recently to figure out, hey, what are you better at? Or Anthropic has come out with Cloud. So we are at the edge looking at a lot of ways where we can move faster. I think that's going to change how we work and what we do. That's still kind of up to the folks here. How are you going to organize yourself to try these things out within your own teams safely with privacy and security, being conscious of all of that, how are you going to change this? I asked you this. How are you at Iron Horse planning to change how Iron Horse works?

Uzair Dada: Yeah. I mean, we are literally in every single discipline thinking about what can we do better? We're already starting to use it for... There's some really interesting experimentations that are happening about, hey, in a certain workflow where I try to go create something today, it's very iterative. I'll use the webinar example just so we all get it. I have to create emails, I have to create ads, I have to create social posts, I have to create AB testing for certain headlines. I got to do blah, blah, blah, blah, blah. Such a linear process to get stuff. So can I start initially with a structure that gives me all that stuff? Say, "Yeah, this looks largely good. "Hey, Alex, can you take a look at it to see how we improve it for this audience? Do you have any other thoughts?" Great. Versus it goes to Alex and he's sort of going through each thing one at a time.

So to me, I think this is the accelerate part, the efficiency part, taking away the noise that something someone doesn't do, I think to me, I think that's kind of where we are thinking of some interesting use cases. And the translation, most of all the translation. The fact that I can now do something and come up with an idea that may be crazy cuckoo like the dragon and the pigeon example. But it's cool because I'd never thought about it before. I think it's just brilliant. Because for us it's not as much about what we can do better, but it's also just learning about possible and not possible. And this is allowing us to do that a lot better.

Alex Jonathan Brown: We've got only a couple minutes left. But I do want to go out on the question that I met through in the chat because I think it is really interesting. What are two to three, so maybe we each do one, non-obvious things marketers can do differently with these technologies? So over the course of this chat, we've thrown out a bunch. But if you had to just pick one thing that you can use AI to change the way, especially let's say marketers, work tomorrow, what would that be? And then we'll plug some stuff and we'll get out of here.

Umair Akeel: I'm just really biased. Rules are a bad thing. Rules are hard to work with. Clickity-clickity campaigns have got to go. So for me, that's the thing that we're working to get rid of. And I think a lot of other companies will benefit from not having to do rules for everything. So go from linear to non-linear.

Uzair Dada: Yeah, and I'd say even within the rules, we give lip service to everything. We don't really take that job well. So even if there are rules, we don't do the rules well. So an example there is we all are obsessed about talking to personas, and building right on conversation paths, and conversation arts, and engagement paths. Do it now. You can do it. There's no freaking excuse not to do it. Talk like Steve Jobs, talk like a developer. Have that information even. It may not be perfect, but it's better than what you're doing today. So I think there's absolutely no excuse to not follow and do good practices that we all have been talking about because I think we've been doing a crappy job at it.

Alex Jonathan Brown: And then mine, again, from a content creative standpoint, I love to use AI as the worst writer on the team. And you just acknowledge it's going to put out like C+ work. And you have someone else, a human being like, do that edit. You just save so much time from the, I'm staring at a blank document and I don't know where to start standpoint. It's so much easier to edit mediocre copy into really good copy, and start from scratch. So that's mine. Gentlemen, this has gone really fast for a 45-minute chat. Thank you both so much for joining us. It's easy to do the plugs for me and who's there because we work at the same place. We're at IronHorse.io and you can find us online at ironhorse.io. No dot in that one. On your social medias. But Umair, where can people learn more about you?

Umair Akeel: Email me. We are an early stage startup working with marketers. And I am at umair@allgoodhq.com.

Alex Jonathan Brown: And when he says early stage, we have had a bunch of meetings. And every meeting has started with a new incredible update about the startup.

Umair Akeel: We're moving fast, that's all it is. We're moving fast.

Uzair Dada: It literally changed before this meeting. So yes, it's changing dynamically. Just like ChatGPT.

Umair Akeel: There we go. Exactly.

Alex Jonathan Brown: Thank you all so much for watching us. If this is your first Coffee Break, we do this about once a month. Again, learn more at IronHorse.io. But until next week, we kept you a little late, but hope you had a nice coffee and it's time to get back to work. Bye everybody.

Uzair Dada: Cheers. Bye.