In the time it will take you to listen to this podcast episode, parts of it might already be out of date. AI is evolving at a breathtaking pace, yet most companies still struggle with adding it to their workflows in ways which will add business value. And this is the uncomfortable truth that few have addressed head-on: if AI can improve your employees’ efficiency and productivity, but you apply it to the wrong tasks, it could be doing more harm than good.
In our new podcast episode, Mammoth Growth CTO Drew Beaupre interviews Ethan Aaron, CEO and Founder of Portable. Ethan and his team at Portable are well on their way to building a platform that can manage integrations with 10,000 other systems. What’s surprising here is not the pace of their growth, but Ethan’s unconventional take on AI. Upon hearing of Portable’s success, most people’s knee-jerk reaction is to suggest AI as a method for automatically building these API calls even more quickly. But Ethan points out something very few other people are talking about: the real test of AI is whether it can add value to the business based upon each of these new integrations. Connecting one app to another just to move data only has limited value, until we can find ways to surface and activate data that can help companies hit their goals.
Ethan’s perspective on AI as it relates to Portable highlights a much larger issue facing every company right now. AI is not the product itself, but simply a tool that can be used to enhance other products by automating tasks and aiding decision-making. If we don’t clarify what types of behavior we need to change to hit our goals, AI won’t help us make better decisions. If we don’t trust the data we’re handing over to AI, or if we’re feeding it inaccurate data, then again, AI won’t help us make better decisions. On its own, AI is not a guarantor of faster, more accurate insights. First, we must understand our business and our data to the best of our ability. Only then can AI deliver on its insane potential.
Right now, AI is an answer in search of a question. But ironically, AI has revealed a universal truth that savvy marketers have known for decades. AI will allow people to build similar products faster than ever before. And that means that if everyone competes on features and price, commoditization is inevitable as we race to the bottom. The only way to avoid being squashed by this trend in the years to come is to focus on your brand. AI can help you build your website faster, produce content more quickly, and shorten development time for your product. But it can’t tell you what your brand should be, or what it could be. As Drew and Aaron point out, that’s up to you. And that’s something that will never be out of date.
Transcription
Drew Beaupre:
Stuff off. All right, here we go. My name is Drew Beaupre. I am the Chief Technology Officer at Mammoth Growth and welcome to the Mammoth Growth Podcast. Today I have Ethan Aaron, who is the CEO and Founder of Portable, which is a service that really focuses on delivering that long tail of integrations where they can create the most incredible integrations you didn't think were possible in a matter of minutes and hours of which I had a personal experience with, which was quite amazing. So Ethan, thank you very much for joining me and telling me more about you and Portable.
Ethan Aaron:
Totally, yeah, excited for the conversation. I know we caught up a week ago, talked for an hour about some of the stuff that we'll talk about today, but also just whatever was top of mind and really enjoyed the conversation. So excited for today. Thanks again for having me. So as you said, Portable focuses on all the integrations other people don't want to support, and the thing that most people think about is, oh, it's a development problem. It's like you must write a lot of code, and if you think about what we do, we've been building with the idea that we want to be able to build and maintain 10,000 integrations. Right now we're at a little over a thousand and the actual goal is the entire opposite. The goal of our company is this really interesting challenge of write as little code as possible because if you have to write code a thousand times, it doesn't matter how big your engineering team is, it's going to be painful. Same with maintenance. So it's been this really interesting operational problem more so than anything else, and it's been fun. We've been running the company for about four years so far. One week from today it'll be four years. Exactly. Focus on being super lean and doing stuff that no one else wants to do. So
Drew Beaupre:
I guess from your perspective, if you can achieve a thousand, there's very low marginal cost to go to 10,000.
Ethan Aaron:
Yep. So the biggest challenge for us, actually, so at this point, it used to sound insane to say that we wanted to build 10,000 integrations. We go like, no, the high watermark for ELT tools is 200, 300. When we started, one of the biggest players in our space had about 150 integrations. So that was the biggest player in the space. So the fact that we're at a thousand with a very lean team, I actually look at 10,000. I'm like, we can totally do that. From a technology and a maintenance perspective, the bigger challenge is actually going to be finding them. In the integrations world, there's different types of integration companies. There's ELT, which is extract everything and put in your warehouse. There's reverse ETL, which is take it out of your warehouse and push destinations, and there's IPaaS tools, IPaaS tools like Zapier, Tray, Workato, those tools tend to have more integrations because it's easier to build an integration with, I just want one piece of data from this endpoint into this endpoint, than a data source, 60 API endpoints. So Zapier is the high watermark right now in all of these, they've got about 6,000 applications they integrate with. So for us, it's less of a question of can we do it and more of a question of are we going to be able to efficiently find 10,000 APIs to integrate at that point isn't worth it. So yeah, it's an interesting challenge.
Drew Beaupre:
So one of the things we wanted to talk about today is more specifically around AI tools and how businesses, what AI tools are businesses actually using to deliver value versus the ones that just have a box to tick on their product sheet. But just in the general AI space, the idea of making data accessible and moving it from A to B seems to be, I don't know if people realize how important it's going to be yet, or maybe is it as important as it seems it could actually be in order to make all these different data points available for training or for indexing in a search capacity?
Ethan Aaron:
So the interesting thing about me when it comes to the way I think about AI is most people that come to me, they're like, oh, AI is going to revolutionize your part of the market. It's going to read the API documentation and build the connector, and we can agree to disagree on that. We're already really fast at that. So it's not incrementally going to add much value in my opinion. The part that's more interesting is if you ignore the fact that someone has to copy and paste data from data source from an API to a data warehouse, and you think about it more through the lens of the end goal, the business value, I think there are a bunch of different ways in which you have to think about how AI can help. So typically where this stuff starts is you're a Head of Data or you're a consultant going into a business, and the first thing you have to do is figure out what problems are really important to this business and AI could help with that.
Do we have any data anywhere in our company even if we can't access it today, that might be able to help us? In that case, things like parsing API docs or parsing application interfaces, not necessarily to build an integration yet, but more just to figure out, okay, if we were to build an integration to Productive, is it going to have the information you need to run your business? And if so, what types of things could you build on top of that data? So it's like, great, what problem do we want to answer? Of all the applications we use, what data might be available? Then it's a question of, okay, how do you get the data out, pipe it into your data warehouse, and then all the maintenance and rate limit train, all this other stuff that comes along with that that I'll skip over for now.
That's the stuff that we spend our time on. And then, okay, now your data's in your warehouse or data lake or wherever you want to put it or spreadsheet. Now the question is how do you as quickly as possible go from raw data to transformed or model data and then to data you can use for insights to, I think AI will help with some of the SQL. I think you're still going to need for a while, human beings to review it because SQLs, it has to be perfect in a lot of respects. And then how do you go from data in a SQL query to a chart or a dashboard? Which chart do you use? How do you organize it? How do you think about visual hierarchy in the dashboards you built? Is it a dashboard or is it a notification? If you don't know the answer, it might be a good place for AI to help be like, Hey, should I expose this in the dashboard or could I use an alert in Slack?
Which one's better? What are the pros and cons? And that full cycle, I think reading API documentation and taking that and turning it into code or configuration is probably the least important of all when it comes to how do you go from idea to value as quickly as possible? It's the most boring. So yes, it's useful for out of help, but the other parts to me of that entire journey are so much more interesting. What could we build? What problems do we have? How do we actually turn that data into a narrative and how do we display the narrative? Is it a dashboard, is it a report, is it an email? So a lot of the time when I talk about AI, it's not how do we streamline the costs? How do we find the high value stuff? How do we present data in a more valuable way? And I think most of the conversation right now is just entirely ignoring that side of AI implications.
Drew Beaupre:
Yeah, one of the interesting things that I heard was the idea that you would just go, you still have to go and prepare your data for it to be useful. If you want to use it to drive some kind of forecasting or something that has high value, you still need to get it into the warehouse. You still need to actually transform it in some way, whereas other players, like, let's use Zapier as an example. They're an AI company now and they're doing more point to point, right? And a lot of other companies I see, it's like, well just stream it straight into your Pinecone database. What do you feel about the ability to, do you still see a world where you have to contextualize the data before you make it available for any kind of system to leverage?
Ethan Aaron:
What do you mean by contextualize in that sense?
Drew Beaupre:
In that sense, I would say that you're either normalizing the data, you’re centralizing it somewhere, be it the data warehouse, data lake, whatever you want it to be, normalizing it, perhaps summarizing it, be it in either aggregation or text summaries. The overall idea is that do you just load up the raw data into a Pinecone or you actually have to be intentional about how you're processing the data before you make it available. And I guess the analogy in my mind would be the data lake turning into the data swamp where you just, it is like, well, we just load it with everything. We don't have to worry about modeling. We don't have to worry about that. That's a problem of the past because we could just load it into this giant S3 bucket and the Hive or the Presto DB engine will save the day.
Ethan Aaron:
I don't think it's going to change fund- I think it'll be easier for more people to ask questions of data, but you're still going to need somewhere to store the data and run compute on top of it. That's not going away, whether it's Pinecone versus Snowflake versus Hadoop versus whatever. You need somewhere to do that. Whether or not you can abstract that away because you have self-service users just asking questions instead of writing SQL, maybe I don't yet think we're going to get to a point where you can ask big questions of your data, like true analytics around how do we allocate resources, without a human being reviewing the SQL yet. Just because if you mess up one joint, if you mess up one piece of logic in the way in which data goes from raw data to the thing in Pinecone that you're asking questions of, everything will be wrong.
So I think it's either you'll start seeing more standardized data models, they get loaded in and you ask questions in a repeatable way, but it kind of skips over all the custom stuff that a lot of people use SQL for or you have AI help write SQL get the data there, the human being reviews it, and then you ask it questions. So somewhere data will be stored and you'll be running compute against it. I think of AI as just another way of saying we want more people to be able to ask questions of our data for ad hoc analysis, it'll be things like we have a spreadsheet, we have self-service BI tools with drag and drop wizards. We have natural language processing using AI. You've got a SQL analyst, so you have all these different ways to interface with your data. AI will help, but it'll also create a lot of liability in the sense of if it is not correct, it's the same as building a dashboard that is not correct. It's a liability for everything. Or building a spreadsheet that's not correct. It's a good way to do ad hoc analysis upfront and get enough information to double click in. And then once you know you want to productionize something, you should think further along, think through the problem in a more in-depth manner.
Drew Beaupre:
Zooming out a bit, this leads to my current thoughts on how to leverage AI and trying to do it at too low of a level right now without really thinking about the larger those kind of problems and solving for those kind of problems, which is a huge lift. I'm more interested right now is what are the real tools out there that are really a force multiplier. The whole point of this is not necessarily to make my workflow 20% more efficient, what's actually going to increase the value of the work that I do by a multiplier? So the last time we spoke, you had a, I guess, what is it, even that category, the zoom bot or chat bot that transcribes and provides context called Grain.com, and we started talking about it and the thing that when I started using that product, it struck me that they actually created a product is product focused that was leveraging a huge amount of AI in the background. They have quite a few different models that they're leveraging in order to do what they're doing, but they're putting all together in an actual product that I actually want to use. And as we started using it, we realized that there is a multiplier effect.
Yes, AI is enabling this, but ultimately because it's a product and the product is driving value, not necessarily the technology. So I'm wondering what you're looking at when you think of companies like that and how you're leveraging for yourself and people around you.
Ethan Aaron:
So the two places where I see AI come into play, it's actually interesting because neither one of them is call recording and using that for both content creation and internal training effectively. And we use Grain for that when it comes to the actual call recording. The other is SEO. I write a bunch of content for SEO purposes. We went from having me write, content to copywriters, write content to people on Upwork, writing content to leveraging some AI combined with me to write some content. And in both scenarios, I do not care if they're using AI, they're not pitching me AI, they're pitching me a really, really intuitive way of recording calls and then creating clips out of that without me having a ton of work relative to a different type of software where you have to record an entire conversation and then a human being has to rewatch it over and over and over again, find the right clips, try and line them up with something valuable.
And it's like it's just a bunch of painful work. So if you think about it, I don't care if they use AI, they might not be, maybe it's just if then else statements, but so that's just real pain. I was doing human effort, I was doing human work and they got rid of it. That's amazing. And then on the SEO front, it's the same thing. It's like, Hey, I want to write a paragraph that includes these terms around this topic. The way I used to do that myself was I would just try and write the sentence and I would add some terms. If you get all the way down to one paragraph, one paragraph of me including three terms in it versus an AI generated paragraph with three terms in it, they're going to look pretty similar. It's not going to be that different as long as it's constrained enough.
But again, it saves me time. I can just be like, Hey, I want this and then try again. Give me something different. So those are both scenarios where AI is helping, but AI is not the product. I think a lot of people right now are building. They're saying our product now has AI, and people are like, I don't care. I'm not going to buy AI from you. That's not a thing. I want you to remove manual tasks. It's either help me make better decisions or help me remove manual work. If you're not doing either one, don't try and expose an AI product, but if you're helping me make better decisions or automating away manual tasks, sell me that. Sell me a solution to that.
Drew Beaupre:
And then there's both, right? If it's automating out manual solutions, it unlocks behaviors that I might have thought too tedious that actually end up having real high value because you can start experimenting with very low cost. So your training example is, okay, well now all of a sudden we have access to all of our internal conversations and in every single conversation there may be one or two or zero, but there may be one or two little clips or highlights that we would just want to save for later and start to build out a repertoire of things that we can then stitch back together for training purposes. You would never have even considered that in the past, but now all of a sudden, because you're considering that, it actually unlocks a lot of new avenues on onboarding and training and that you would never even would've done the first place. Example is your 30 minute talks, you mentioned that you'll just go pull out little clips, little anecdotes and just ask for permission. Hey, do you mind if I use this? And without being able to do that as fast and seamlessly and the ability to, like you said, it's a product. And so it's really easy to share because it's a product feature to go share clips. It unlocks behavior that makes you more creative on how you use information.
Ethan Aaron:
Totally. I actually am going one step. Initially I was using a call recording software for that, for creating these clips and being like, Hey, I'm having conversations anyway. Why not create content with that? Because my realization over the last two years is marketing and content are extremely valuable for running any sort of company or community. But I'm taking that one step further right now, and I'm actually going from, Hey, I want to have a conversation, create a clip and post it to, I had an idea about three, four weeks ago to create an online class around people analytics and HR analytics. Half the videos are going to be me recording myself talking, creating clips, and then adding them to the course. The other half are me having conversations with experts in the people analytics and HR analytics field and then using Grain and this transcript-to-clip capability set to just cut out one minute clips, get permission to use them, and then include 'em in the class.
And the idea that one, it's me plus $30 a month in Squarespace fees plus Grain and a little bit of my time because of the amount of leverage I can get from this technology, I can ship an entire probably 45 minute to an hour long class with video content in a matter of weeks. And I get to have a bunch of really fun conversations along the way. So it's like the idea that you can do that so efficiently now wasn't accessible a year ago, wasn't accessible six months ago because you'd have to go manually figure all this stuff out and now it's easy.
Drew Beaupre:
It's little things like that. That reminds me of early days of cloud computing where all of a sudden something, I remember when we had to rack our own servers and we're hosting or moving from data center to data center in order to save money because it's like, well, this data center's cheaper and you're physically shipping boxes from A to B and racking, and then all of a sudden one day, I'm sure people are still doing that, but you didn't have to anymore. So it enabled a completely different mindset around thinking about the product and the software rather than the hardware. And then a lot of other new tools came along building on top of that, which took the software, a lot of the software infrastructure out of the picture, we take for granted GitHub and all the DevOps workflows that go along in that ecosystem. Before that, you had to do it all by hand. What other tools are you looking at that they’re leveraging AI in a way that you don't- It's enabling efficiencies that were never possible before, but at the same time, it's a product and not necessarily a technology.
Ethan Aaron:
On the AI, on the data front, the only thing I've seen so far that's been useful, and maybe I'm just not exploring enough yet, is just the ability to debug my SQL statements. If there are issues and I have that natively in Retool and it works good enough to point me to what's going on, I honestly wouldn't say it's material to my workflow. So that's one in the data world in the, I spent a lot of time in marketing. The other one is using AI to help me figure out what content to write, and then also then leveraging AI to write the content. So for me, I use a combination of a tool called Clearscope, which helps you figure out what terms matter for an article going after a specific keyword on Google. And then right now I use Gemini because I find it has more up-to-date information than ChatGPT to actually help me craft paragraphs in my articles. So I'm not going in being like, Hey, write me an article on this. I'm going in being like, Hey, I want a paragraph and I want the paragraph to be about this and I want these terms in it, and it works really well because I'm using my domain knowledge to define, “Hey, here's a very concrete problem I need you to solve for me.” And it spits out content that's
Drew Beaupre:
Quite good. Be it Bard (now known as Gemini) or ChatGPT, you roll back a year when ChatGPT was in its early days, it was never meant to be a product. It was meant to be a technology demo, and OpenAI just assumed that someone was going to come along and build a better product and no one did. And so ultimately they made the choice to turn like, okay, we're going to start to actually turn this into a real full fledged product and invest in it. And you can really see how that's really shaped their trajectory in the last year in terms of moving towards more like a product company than a technology company. At least from the outside. Their dev days was almost like reminiscence of an old Apple keynote where they're really talking about the product and how it's going to accelerate new ideas. And so it feels like some companies understand that everything needs to be product focused, but other tools that they're just slapping on the technology and it really doesn't- it’s harder to use than it actually solves any problems.
Ethan Aaron:
I think right now we're seeing a macro trend of, I was talking to one of my friends who runs a company in the customer experience world, and all of his clients are coming to him saying, “Hey, what AI features do you have?” And he's like, “I haven't found the exact use case for AI to give you those features.” But my hypothesis as to what is going on is major technical innovation taking place in the AI world. It's not products yet. It's a lot of technology. Boards and leadership teams of companies across the world, whether it's venture capital as private equity, public companies are coming in being like, Hey, this is going to change your business. You need to think about it.” That goes from the leadership team to, in his case, the head of customer experience and they say, “Hey, you're doing customer support for our clients.
How are you leveraging AI?” And I think right now what's happening is no one knows yet. That part has not been productized yet. Yet. So these customer experience leaders are saying, I need AI because I'm accountable to my own leadership team and my own board to show that we're going to leverage technology to innovate. And the vendors to a certain extent are being forced to catch up because that's where the budget is going. Because these boards are saying to their customer support and customer experience teams, you need to figure out how to streamline by 30% or how to take on 50% more tickets with the same headcount, figure it out. And these heads of customer experience are saying, we're not going to be the people that figure out AI and they're going to the vendors and the market to startups and saying, it's on you.
And I think 90% of these AI features are going to fail maybe more, but someone in that world is going to figure out how to productize AI. The more interesting part is the competitive dynamics are also fundamentally changing. Let's say this hot new startup in the customer experience world finds the perfect AI feature for customer experience, and it's just built on top of APIs from Microsoft, Google, Amazon, whoever, ChatGPT, OpenAI. What stops every other customer experience vendor from selling the exact same feature? Not much. So it's this interesting thing of I think the barrier to entry in this world, it's going to be easier to do more with less, but it's going to be very difficult to prove how what you're doing is really incrementally more valuable than everyone else in the market because everyone else in the market is going to be doing the same thing.
I see it now with emails. I'm assuming you get a billion emails from garbage autogenerated stuff, it's like email is useless if you try and send me a cold email, I'm going to block it and report it as spam because I get so many of these. And I think we're going to see that happen everywhere. And that's the part that I think is going to be really difficult here is how do you navigate the noise? You might create a cool thing, but if you do what stops 20 other people from all doing the exact same thing? I don't know. I think that's going to be a challenge.
Drew Beaupre:
Yeah, that's a really, really good point. I don't know, because maybe the problem, it's all low hanging fruit right now and everyone is grasping for the obvious, like semantic search sales bots and things like that. And it's going to take someone to really find something out of left field that is a whole new category and no one realized they needed it until it was there. And that will be really the killer thing. And I think that maybe that is the typical pattern all over again. It's just so compressed. All this is happening. The cycle is going to happen over the course of two, three years, where in the past it might've been 15 years. That'll be interesting for sure.
Ethan Aaron:
If you take out the need for a human being, a lot of workflows and companies, and if you truly believe that you can use a combination of AI plus logic to create products instead of technology and someone does it in your space, what stops everyone else from doing it? If all it is is a technology and AI problem headcount, the cost for engineers that know how to do this is going to skyrocket. It already is. But it's a question of like, okay, cool. Does everything just become truly a commodity in the tech world, which I kind of do believe a lot of things will commoditize, and if they do, I think the two things that matter, one is content and marketing. Do you have a brand? Because if you sell the same thing as 20 other people and you don't have a brand you will lose.
So I think marketing is going to become more important and truly creating a brand. And then the other part that I think about a lot is let's imagine a future where you can use AI to build ELT connectors, what we do, and everyone could just go click, run this, spin up my Python script, and then let me run it in the background. I think the cost, the ongoing maintenance and compute costs and are going to become material enough where, sure, we might compete on costs, but if I've architected a company and the way in which we use humans and AI more efficiently than you have, even if we both use AI, I can still cut costs 75% more than you can, and if that's the case, I can compete on cost. So in a commoditized world in most of these markets, whether it's content creation, video recording, et cetera, I think it's going to be not just how do you build the AI based product, but how do you do so? How do you architect it in a way that actually is more cost effective and how do you create a brand? I think those are going to become more important in every market.
Drew Beaupre:
How does proprietary data change the equation if the access to the ability to do various generations through LLMs or whatever the generative models you're using are, how important does your proprietary data become in order to solve the niche problem that you're really trying to, that you see value in there? There's a niche. I see value in it. No one else can do it. Even though this technology, underlying technology is available to everyone, no one else can do it because they don't have the underlying data to support it.
Ethan Aaron:
I think there are two ways of getting proprietary data. I think they will become extremely valuable. One is your own proprietary data. A lot of companies don't have anything that's really material, but if you do, again, if you do, it could be really valuable. A lot of that, but a lot of it, I would even say more so is going to come from your ability to create marketplaces and get access to data from other people and use that in a secure, compliant, ethical way to create benchmarks. I've been spending more time in the HR analytics world, and something that I've come across that's really valuable is not, oh, I just want insights into my data, but I want insights into my data, and I want to know how that compares to hundreds of other companies in the HR space. That's not really a technical problem.
It's a technology problem, but let's say you could abstract it away with AI or humans, whatever, it's, it's really a marketplace and contractual problem. I've worked at LiveRamp, which was effectively a marketplace network effect business, and then another company Arbor before that, that was a pure marketplace. We had a bunch of data providers and a bunch of data buyers. Without your marketplace, you add no value. So I think as it becomes easier to build the technology, your brand will matter. But to your point, that's a really good point of you also need to have access to data, whether it's your data or permission to access to your clients, and you can't do this without explicit permission from them, but to create things like anonymous benchmarks, to create things that, not that they have to train your models, but just something that can't just be created overnight with Gemini or ChatGPT and I, we're going to see, I think it's going to take a while.
I'd give it five more years before that becomes the only thing people have to compete on. But I think for the next five years, it's going to be, oh yes, let's use this. Some companies will figure it out, build a product, no one will catch 'em because they don't realize that they could. And then in five years, we'll see a new layer of platforms emerge that actually allow you to build this stuff pretty fast, and then it'll be, okay, great. What does it take to build a company? I think it'll change marketing cost. And to your point, I wouldn't even say data. I would say network effects are going to become absolutely critical and everything else will commoditize.
Drew Beaupre:
The network effect is. That's really interesting angle on it all. So to wrap this up, do you have a particular method for- you see something new tool come along? What is your thought process in terms of, is this just another spin on the litany of new AI tools that come out every day, or is there actually something new novel here that I believe can drive, really drive a compounding value effect?
Ethan Aaron:
I tend to look at tools through the lens of can I use them at Portable? And when I think about our own product development, how can we serve our existing customer base and community better? So I haven't spent a ton of time recently just looking at tools for the sake of it. I'm looking at things being like, Hey, I'm trying to write content. What tools can I use to write content? My go-to way of learning about new technologies and tools is pretty straightforward. I pick an ecosystem. Let's say it's content marketing. I go online and I search for give me the landscape of MarTech tools or of content creation tools. Typically someone creating an image that is a landscape or wrote an article that is a bunch of tools, it might not be perfect, but you get three or four of them, they give you a pretty holistic summary of all the tools.
And then I go to their websites and I don't look at their websites because websites rarely actually represent what companies do. I'll either click get a demo and spend 20 or 30 minutes just having someone pitch me and show me a demo of their product, or I get a free trial and I look at it. So it's like it's a lot easier than you would think to get up to speed on any market. If anyone trying to get up to speed on the data world and be able to say, oh, this is what ELT tools do, behavioral analytics tools, transformation tools, just get 30 demos. You can do it all this week. It's pretty easy to just go to people's website and click schedule time with me and give me a demo. No one will, no, no one's going to block you and say that you can't have a demo of their product unless it's a competitor, then they probably will. So that's my go-to way of learning. So in the AI world, if I have a specific use case, whether it's how do I think about call recording software and what are the right solutions, demos and free trials or in my opinion, the fastest and easiest way to learn. Same thing on the content creation side. Same thing as I think about new types of technologies,
Drew Beaupre:
And to your point earlier, does it matter whether it's AI or not or does it make something better or improve something faster? And it doesn't really matter whether it's AI, it's like this is just a good tool, full stop.
Ethan Aaron:
Yeah. To me, when I think about content creation, I don't think about, oh, this is AI. I think in the back of my head I'm like, it might just be some logic. It might just be logic. Someone wrote code. But to me, the alternative that I know very well is you have to hire a contractor or go on Upwork, find the people, hope you find the right person. Then you have to go communicate to them what you want them to do. Then they're going to write something probably not going to be super great. Then you have to go back and give 'em more feedback. They're every hour that goes by, they're charging you for it, or it's a project based work, and every hour that goes by, they get less and less motivated to help you, and then they do some updates, and then you have the content that you want and then you ship it.
That was the alternative to me saying, Hey, I need a paragraph that includes these three terms. Gemini give me that, and it gives me the paragraph I want. So it's not the fact that it's ai, I don't care. I just want a very easy, free way to say, give me a paragraph that includes these terms and AI in a chat form is able to do, if you give me a human, if you give me software powered by humans, that is just free, put in a ask a question, and a human being gives me the answer for free or for 20 bucks a month, it's the same thing. I would probably prefer the human over AI because it'll be more correct. It's a cost question at that point. You can't, it's going to be very difficult for someone to staff up an army of people on Upwork to write those paragraphs, but I don't care if someone wanted to do that.
It might be cheaper than training a model, but I don't think a lot of people think that way. I don't think they think about it as get rid of all the pain. Give someone a click to sign agreement that says 20 bucks a month. I don't like Grain. Maybe they have a human being that watches videos really fast and writes transcripts. I do not care. I just want the transcript and I want someone to create a clip when I highlight text, and they just happen to do it really fast. And my guess is they're using AI, but if they're not, I don't care. And I think that's really important. It's not the AI, it's just the ability to remove pain that the opportunity is right now.
Drew Beaupre:
Yeah. Amazing. Well, Ethan, thank you so much for your time. This was great.
Ethan Aaron:
Totally, absolutely pleasure. Love to do this again soon.
Drew Beaupre:
See you. Bye.
Ethan Aaron:
See you.
