Episode 3 – Federico Pascual

Filipe Santos
00:00:00 – 00:01:23

Alright. Have you ever asked yourself how all of this AI magic works? What are LLMs? Are there any caveats I need to be aware of when integrating AI into my workflow? Or perhaps how to determine if an LLM or a solution won’t make the grade? In this episode, we’re thrilled to welcome Federico Pascual, the former cofounder and COO of MonkeyLearn with over 10 years of experience in AI and SEO. Stay tuned throughout the entire episode where we’ll address critical topics on how you can determine the best foot forward on generative AI quality, LLMs, use cases, and applications along with any limitations to be aware of. Well, with the his experience in, leveraging AI for scalable data analysis, Federico has played a pivotal role in serving top tier clients like Salesforce, Snapchat, Atlassian. He’s also spearheaded innovative strategies, including a groundbreaking SEO playbook, which propelled MonkeyLearn to dominate Google rankings in their niche. Join us as Federico shares insights into the power of AI for SMBs and enterprises, taking us along the journey behind MonkeyLearn’s acquisition by Medallia. Federico will also discuss the incredible generative AI tool that he’s been developing lately, Wordcrafter AI. With that, Federico, if you wanted to actually mention anything about what you’re involved with now, we’d love to hear it.

Federico Pascual
00:01:24 – 00:01:48

Perfect. Thanks very much, Filipe. It’s it’s great to be here. Yeah. Basically, I’m currently working on Warcraft AI, tool that tries to solve a few challenges they had in the past while I was scaling SEO at Montclair. So at Montclair, SEO worked really well for us. Thanks to SEO. We were able to get these amazing customers that you just mentioned.

Federico Pascual
00:01:48 – 00:02:26

For us, SEO was hard to scale. For scaling SEO, we needed to hire 4 content writers. We needed to hire an assistants. We needed to hire an editor and so on. Basically, it was hard hard to scale because we spend a lot of time doing mainly 2 manual processes. First of all, we spent a lot of time researching search intents That’s that is trying to understand and figure out why someone searches for a particular keyword on Google. So for example, in our case, the keyword in our space was sentiment analysis. That is understanding if a piece of text is talking good or bad about something.

Federico Pascual
00:02:27 – 00:03:26

For example, whenever someone searches for sentiment analysis on Google, they may be trying to figure out what the hell is sentiment analysis. You know, trying to understand the definition of sentiment analysis. But maybe another person is trying to do the same search, searches for sentiment analysis in Google. Maybe they are trying to figure out how it works. And maybe a third person does the same search in Google, and they’re trying to understand use cases and applications. So, basically, we spend a lot of time going through combing the different search results on Google, basically, the top ten results, reading the whole contents of these pages, and ask ourselves what is the main search intent that this page tries to solve and why Google is prioritizing this page. Also, we spend a lot of time trying to figure out what topics and subtopics we needed to cover to solve those search intents. Then with that information, we’re able to come up, with an outline, a table of contents, basically, that maximize our chances to solve those search intents.

Federico Pascual
00:03:26 – 00:04:18

And that lead us to the second part of why it was hard to scale up monitoring or SEO. That is once you know what you need to do and what you need to write about, it’s all about writing world class content. And in our case, it was super hard because it was mainly technical things, you know, talking about things like sentiment analysis, keyword extraction, machine learning, natural language processing, and scaling that type of content was super hard and and difficult. Now that I’m back to building, you know, hey, I wanted to tackle challenges and problem that I have in the past. So I remember these challenges, and it was really painful. And, you know, with the current state of AI, it’s it’s these challenges are solvable or at least, you know, help can help people work in SEO do more, you know, suffer less.

Filipe Santos
00:04:19 – 00:04:47

Yeah. No. I mean, I I love that you mentioned that because I I remember finding MonkeyLearn when I was first kind of figuring trying to figure out how to scale up clustering. Right? Like, some clustering of keywords so that in a world where you have so many keywords and you need to figure out how to, like, quickly cluster keywords and make sense of how much they’re worth to you, I remember that was one of the tools that you guys offered, and it was kind of exciting for me because it made things a lot easier than doing it through spreadsheets using a bunch of formulas that just made things very complicated.

Federico Pascual
00:04:55 – 00:06:00

Was searching like I don’t know. It was like half a 1000000 people, something like that. You know, it was something it was a huge keyword. We also realized that it was a super relevant keyword for us because Bloomfield basically was a tool for analyzing text data at scale with machine learning. And realize that people searching for word cloud on Google were like in the first steps of the process of analyzing text. Basically, they want to they have piece of data, maybe a document or a PDF or something like that, and they want to throw it to a word cloud generator and, you know, see the word cloud and trying to find interesting insights. So we found this high volume keyword that is running in our space, and we, you know, sat down and studied the search results to try to figure out how to run for this keyword. And, basically, we found out that we needed to do something really different to what we were doing because, basically, we’re for the the rest of the keywords, we’re creating, like, doll form content explaining how machine learning works, how natural language processing work, what is sentimentized, things like that.

Federico Pascual
00:06:00 – 00:06:31

But in this case, the search intent was completely different. Basically, when someone searched for a a word cloud, base they basically want to try a tool. They don’t want to read a blog post. That was the main search intent. And, basically, what we did was but what we did was basically that. We built word cloud tool where you can upload your text data, even uploading a by uploading a file or copying your text and pasting it in the tool and get a word cloud generated with AI. And it worked wonders for us.

Filipe Santos
00:06:32 – 00:06:45

Question on on that. So, like, with the with the word cloud, do you think it was a means of getting quick insights on their data, or do you think it was, like, better used for presentations? Like, what did you hear about the tool being used? Were there any interesting cases?

Federico Pascual
00:06:46 – 00:07:25

Well, all of these, you know, all of these different use cases. You know? Definitely was getting quick insights from data. It’s like the first thing that people do when they try to figure out, you know, long form content, large pieces of data, but it was also used for presentation. So for example, we heard not only about people working in businesses and working on a presentation, but we also heard about professors and teachers using the tool in their own classes to, you know, the different, lessons and presentations that we’re doing in school. So, yeah, it’s pretty, like, pretty buried the use cases around that tool.

Filipe Santos
00:07:26 – 00:08:05

Awesome. Yeah. I’m I’m glad to hear that because I remembered using word clouds myself, and you never know. Sometimes it’s, like, really effective for educating, and other times, you just need to know something very specific about the intent or about the class of, keywords that you’re going after and the the the work that does make it make it just so much more visual. And we’re visual creatures, so I I think that’s a key point. With that point, I I guess we can talk a little bit more about, like, how modern AI is being approached. I know I know AI changes by the minute. Even from last year to this year is insane, and every single month that comes is, like, something brand new that just basically blows up your mind.

Filipe Santos
00:08:05 – 00:08:11

But I’d love for you to get a little bit into some of the basics of how modern AI works.

Federico Pascual
00:08:12 – 00:08:59

That’s great. And, yeah, it’s pretty challenging to explain really technical concepts, but I’ll give it a try. Things like ChatCPT, Gemini, all these generative AI models, basically, what they do is they predict the most likely word from a given text. For example, if you give a prompt, they predict the the next word after that text. And they do multiple times predicting the next word and the next word, and that’s how they create sentences and paragraph and the whole responses by always predicting predicting the the next word. So maybe we can play a little game. So for example, I’ll give you a fragment of a text, and the last word, I will not say to you, and you have to guess that word. Let’s see how how you will predict this, how you predict.

Federico Pascual
00:08:59 – 00:09:04

So if I tell you, I love you very blank, what’s the word?

Filipe Santos
00:09:04 – 00:09:05

You’re gonna think much.

Federico Pascual
09:11 – 00:09:12

I would hope it’s you, but

Federico Pascual
00:09:14 – 00:09:25

hopefully, yes. But, that’s how, you know, a large language model works. Basically, they predict the next word given a prompt, and they do multiple times to quell the whole answer.

Filipe Santos
00:09:25 – 00:09:35

That’s awesome. Those are really good simple examples of how it works. I’m kind of intrigued by that because I I’ve also heard of a a concept called tokens. Is that what that’s what that is?

Federico Pascual
00:09:36 – 00:09:51

Yeah. Tokens are like parts of words. You know? Imagine that tokens, like a a piece of content that has 100 words may have 1200, tokens. You know? It’s like, yeah, it’s like a smaller unit that complete words.

Filipe Santos
00:09:51 – 00:09:58

Okay. And and the speed of which, like, that happens, the there there’s also, like, a metric. Right? I think it’s, like, tokens per second part per minute.

Federico Pascual
00:09:58 – 00:10:23

Yeah. That’s basically to measure how a performance an AI model is to generate an answer. Basically, the larger the model it is, the slower or more difficult is to generate faster responses. And, basically, big model like g p d 4, or Gemini 1.5, they need, like, huge amounts of compute and GPUs basically to answer quickly.

Filipe Santos
00:10:24 – 00:10:34

For the sake of, like, most people, like, that are not technical, that stuff doesn’t really matter. Right? But it’s just more of, like, how will the user experience be measured from a technical aspect. Is that true?

Federico Pascual
00:10:35 – 00:11:24

Yeah. That’s that metric, 2nd per second is more is important on use cases that things that for use cases where real time is really important. So for example, imagine that you want to have you have a company and you want to start using AI to automate certain parts of your customer service and you want to iterate AI into a chatbot that can interact with your customers. If you integrate that chatbot that generates answers, like, really painfully slowly, it will hurt the customer and experience overall. So in those cases where real time is really important, you need really fast inferences, you know, really fast predictions. And you can achieve that either by using smaller models that are optimized for a particular use case, or you can use, you know, infrastructure at scale, basically.

Filipe Santos
00:11:25 – 00:11:42

That’s super interesting. And GPUs. I I love that because that that kinda simplifies how I’ve seen it. It makes it a lot clearer. Because I was always wondering, like, is that just something in the background that is more technical, or does it actually have some bearing on business decisions? And clearly, it sounds it like it does if you’re doing real time applications.

Federico Pascual
00:11:43 – 00:12:33

Yeah. Yeah. And, also, it’s like it’s it’s super important to think about it’s not always convenience to use the biggest model out there or the most performant, model out there, things like GPT 4 or Gemini 1.5 or Cloud 3. There are certain use cases that a smaller model trained for a particular domain and for them particular use case can achieve, like, really similar accuracy than a big generic model. You know? And, obviously, if you have this smaller but specialized model, it’s much cheaper to train. It’s much cheaper to do predictions. I mean, it’s you know, the return of investment is much better. So it really depends on what you are trying to build, what you’re trying to do, what type of model you need to to for your business.

Filipe Santos
00:12:34 – 00:12:41

Well, perfect segue, Federico. I guess we can talk about what makes these LLMs, large language models, good or bad.

Federico Pascual
00:12:43 – 00:13:25

Well, that’s a a good question, and it’s, like, highly debatable. But in my opinion, what where these large language models really shine is basically things that are super related to what these models do at the core. Basically, predict the next token, the next word. So for example, things like, try translating texts, LLMs are really good. Writing content, writing code. Also, large language models are really good. Summizing text, they also they are also really good. LLMs are not good on things like they don’t have memory in the sense that they are not able to store new information like humans do.

Federico Pascual
00:13:25 – 00:14:10

So for example, if you interact with a role model Mhmm. Every time you interact again with a model, you renew interaction, you are basically starting this from scratch. The model will not remember you. It will not remember your preferences or your answers or who you are. Basically, what the model will do is predict the next token, the next word. That’s it. You know? And it’s super interesting how companies are currently, you know, trying to hack, or go around that limitation of LNMs. So for example, companies like OpenAI are basically feeding again the model, previous interactions via the prompt so the model can use that information from the next response.

Federico Pascual
00:14:10 – 00:14:19

But it’s not that the model has learned from that. Basically, what OpenAI or these companies are doing are feeding these, interactions previous interactions via prompting.

Filipe Santos
00:14:19 – 00:14:23

That’s fascinating. It it actually even sounds a little bit inefficient.

Federico Pascual
00:14:25 – 00:15:36

Well yeah. Because, for example, context length, basically, how much information you can add in the prompt to generate a response is quite limited. You know, CPT 3.5, I think it was like 4,000 tokens. Claw 2 was a big breakthrough in the space because it was the first model that was able to provide, like, a much bigger context window allowing up to 200,000 tokens. And a couple of weeks ago, Google announced that with Gemini 1.5 that, theoretically, that model is able to get a context windows of I think it was 10,000,000 words. So, yeah, you know, as this technology advances, the cost of feeding back the conversation is cheaper and possible, you know, and eventually, I imagine that things like Charge CPT will be able to feed every interaction, that it have historically with you. And even remember, you know, other other things about you, about your life, about what you like and dislike to generate more accurate responses.

Filipe Santos
00:15:37 – 00:16:02

That’s pretty awesome. And and I think that’s, like, kind of where we have sci fi kinda telling us what will happen with AI. Right? It’s just like it will become a little bit more sentient. It’ll become a little bit more aware. Obviously, we’re not there yet, but things are moving so quickly that I wouldn’t be surprised if certain elements are coming together and, like, sounds scary. But in reality, we just have to understand how to how to use it, how to wield it. Is that correct?

Federico Pascual
00:16:03 – 00:16:40

Yeah. Well, AGI, it’s a hot topic right now. You know, do you have all kinds of people with all kinds of opinion there. It’s like you have this group of people that say AI is really imminent, that basically we will have AGI maybe in the next year or 2. Then there are other people, mainly in Silicon Valley, that they are also optimists, but not as optimists. And they they they say that it’s probable that we’ll have AGI in the the next decade. And then we have a group of people that are also AI experts, AI researchers that say no. AI is really far away.

Federico Pascual
00:16:40 – 00:17:27

LLMs are still basic, models in the sense that they just predict the next word, and they lack things like memory. They are also for example, another limitation of large language model is is that they are not able to retrieve things to learn new concepts. So for example, like humans do. So for example, imagine back in 2020 when coronavirus happened, neither of us knew what coronavirus was. And probably we heard someone mentioned about coronavirus, and probably we went online and start searching for information to learn about this new thing that was happening in the world. LMs cannot do that. They cannot search and learn from new information in the case that, the model doesn’t know what something is. So they cannot retrieve things.

Federico Pascual
00:17:27 – 00:18:12

Also, large language models currently, they cannot reason things or plan things. It’s like when they start replying to you, it’s not that they first think about the response. You know, reason the response and plan the response, and they then then generate the response. No. They automatically predict the next word. And, also, for achieving AGI, many AI experts say that LLMs, lack real understanding of how the physical world works. So for example, for us humans, since we are very little, we are constantly learning just by giving the physical world. You know, like having a glass of water and putting it in the table and interacting with our with different objects.

Federico Pascual
00:18:12 – 00:19:15

That’s how we build our world model and how we understand model. The lens don’t do that. They basically learn from text data. Many AI researchers say that they will always be really the their understanding of the physical law world and the physical laws will be very basic, very superficial because they just learn from text data. Obviously, there are many researchers working on that, like training models on not just data, but train models with additional types of information, things like images, video, and other type of data. Also, certain AI researchers are using Embodied AI, basically giving AI the a way to interact with the physical world via a robot or via a a drone. So the the AI model can interact with the AI world and start learning stuff. So, yeah, there are multiple people trying to improve these type of things.

Federico Pascual
00:19:15 – 00:20:27

But, again, you know, AI, at least in my opinion, is still very far away because they lack memory. They don’t search for they are not able to search new information and learn from new information. They don’t reason or or plan or at least they do it, like, very rudimentary right now, and also they don’t that doesn’t mean that AI or is not valuable nowadays. It’s like it’s completely the opposite. AI is amazing. I use it every day, and, basically, I’m able to do much more and in less time. And I can do and I kinda to make some tedious tasks that I used to do manually, I can rely on AI to do the heavy lifting for me so I can focus on more important things or things that I find more I enjoy myself more spending time on those things. So, definitely, you know, I feel that every marketer, every SEO, every content market that needs to start using AI basically to do their work better and faster, focus their time on things that they enjoy and where they have a unique expertise.

Filipe Santos
00:20:28 – 00:21:26

So what what I was getting out of that, which is for the like, thank you so much for that explanation, is that we should not be scared of the technology. There there it’s not going to necessarily replace, you know, all of the definitely a lot of roles. I think it’s more about how we understand the technology and use it to become better and maybe use them as as our own skills to improve our work and to also expedite, like, the the types of projects and initiatives that we’re following through with. And, you know, that’s that’s something that’s very important to acknowledge and put out to the audience is that it’s really here to help us, and it’s not at a point where it becomes really detrimental. It’s something that we just need to learn and understand better, which is why all of these explanations actually do paint a picture of how this technology is in reality and what what it’s gonna do for us in the near future. And but that that takes us to a perfect perfect point in time, which is what are your favorite use cases and applications of the technologies in marketing and SEO?

Federico Pascual
00:21:27 – 00:22:24

In marketing and SEO, that’s a great question. One thing that I love is that with AI, it’s much better it’s much better and faster and easier to understand our audience. Who do we do marketing for. You know? If you don’t know your audience, if you don’t know what they like or dislike, if you don’t know where they hang out, what ticks them, it’s really hard to do marketing. And, basically, AI is super useful to analyze customer feedback, for example, things like analyzing support tickets, analyzing sales conversations, analyzing responses to different type of service. AI is really good on pinpointing these insights that in the past, at least as marketers, we need to go through this data manually to try and find these patterns, what people like or dislike. And now AI, we can point that out, like, really fast and really quickly. Yeah.

Federico Pascual
00:22:24 – 00:23:34

These are the top things that people like from these survey responses, or these are the top things or the top five pains customers or leads have mentioned in sales conversations with our team, or these are the top things mentioned in support tickets. So it’s a really great way to better understand our audience and also to figure out what we need to do with our different marketing actions. So, yeah, AI is really good at analyzing vast amount of data super fast and super quickly, so we don’t have to do that ourselves. Also, yeah, I’m really excited about those manual tasks or things that we need to pay attention that currently with the current state of AI is not necessarily needed. So for example, in related to content and content marketing, you know, for example, in my case, as you may notice, English is not my my native language. I speak Spanish. So whenever I write something in English, I need to pay attention, to grammar. I need to pay attention to clarity, to sentence structure, you know, in order to try to figure out if it makes sense.

Federico Pascual
00:23:34 – 00:24:15

You know? But I consider myself a really good writer in the sense that I always know what I need to mention in a particular section or a particular power up. I know this is the idea I need to transmit and how I need to transmit it, and what angle I should use, and what piece of data I should mention. And so with AI, it’s it’s amazing because I can just tour whatever whatever I have on my mind really fast and don’t have to worry about grammar. I don’t have to worry about clarity. You know, I can ask just ask AI to rewrite whatever I have on my mind, and I just dump it there. I can ask AI to make something coherent about that, and that looks good. So, yeah, that’s use case that I really, really love.

Filipe Santos
00:24:15 – 00:25:07

That’s awesome. Yeah. I I imagine, like, the translation aspect is pretty important, especially, like, on the last episode, we were talking a lot about small and medium sized businesses. The fact that they usually don’t have access to lots of resources, being able to translate articles, translate concepts, even have chatbots that can communicate in multiple languages. This helps a great deal. And and as you mentioned, like, just even putting your own thoughts into perspective, copywriting is so key to the way that we communicate, especially in business, that it it’s, you know, any little help that we can get that takes us over, you know, like, walls when we’re we’re stopped and we just don’t understand how how to be creative or to get over a specific impediment, AI can help us there. So it’s really great to kinda hear those use cases. I do think that with, you know, knowledge and with that power, there’s obviously a lot that we can continue to learn.

Filipe Santos
00:25:07 – 00:25:20

And I think everything in life is about learning and always about change. So are there any, like, educational resources or tools that you think would be great to mention to the audience, like, you know, things that they could take advantage of right now?

Federico Pascual
00:25:21 – 00:26:12

Yes. No. It’s it’s incredibly excited that we live in a world where learning is basically free. You can learn any difficult concept or difficult area, basically, for free, thanks to Internet. There are many things that are great resources to start learning about AI, start honing your abilities with AIs and, you know, to take advantage of these new type of products and this new technology to help us do better, basically. So, yeah, one one thing that I can, suggest is Brittney Mueller’s guide, introduction to large language model. This is a great guide targeted for marketers. Winnie does a great job explaining how large language model works, but she keep it keeps it, like, nontechnical enough so that anyone can understand.

Federico Pascual
00:26:13 – 00:26:59

And it’s a it’s a great resource for learning how alarms works. Also, I can we commend our website is called deeplearning.ai. It’s confounded by Andrew Engel. Andrew Engel is legendary AI researcher. Basically, he was cofounder and and and and the chief science officer at Google Brain, basically the AI division in Google. He also is cofounder of Coursera. He recently cofounded deeplearning.ai to share all kinds of courses and resources to learn things related to AI. So for example, they have this course given by Andrew Engen himself that is called generative AI for everyone.

Federico Pascual
00:27:00 – 00:27:44

It’s great because you have this legendary AI researcher that is also like the OG of AI educator teaching these concepts related to generative AI himself that is, again, targeted to nontechnical people. So you don’t need to know about how to code. You don’t need to know about technical things on how AI works to do this course. So that’s also a great resource for these, really great podcast that I like. That is basically the Lex Financial. Lex is AI researcher. He worked for a long time in name in MIT, researching AI, in particular, Computer Vision. He has this podcast, I think, for more than 5 years now.

Federico Pascual
00:27:44 – 00:28:34

He and he started the podcast as AI podcast, interviewing all kinds of interesting people in the space. So over there, they’re really interested in interviews with people like Ilya Suskever, one of the cofounders of OpenAI. He’s currently a chief science office officer in OpenAI. Also, Andrej Capati, he was former head of AI at Tesla and also is a legendary AI researcher and AI educator. And, also, a further interview that I recommend from the Alex Freeman podcast is Jan LeCun, also a legendary AI researcher, and he’s the chief AI scientist at at Meta. And those 3 interviews are particularly great, you know, to start understanding how these world class AI researchers think.

Filipe Santos
00:28:34 – 00:29:24

I really appreciate that because I was actually going to ask you about your favorite podcast. So it’s it’s great that you brought that up as a resource and in particular those episodes because, as you mentioned, if folks understand a little bit more about the people who are researching the the the way that this has come about and just their their methodology or their thinking, goes a long way. Now maybe people don’t have all that time. Believe it or not, there are these interesting applications of AI where I I forgot the tool, but you can actually and I might actually link it down under the video at some point. There’s tools that will let you summarize the podcast so you get the kind of the gist of what’s happened in that podcast if you don’t have time to listen to it all. But I’m I’m sure that hearing from these big minds in AI will just open up worlds for you. And especially as a marketer, it’s critical that we understand what’s happening and what’s changing. So I really appreciate that, Federico.

Filipe Santos
00:29:24 – 00:29:50

I I guess switching gears here, let’s talk a little bit about understanding the role of AI, specifically in SEO and content writing, because I think that’s one of the best applications right now for us all to take advantage of. So as we kind of, like, start talking about that, maybe you can tell us, a little bit about about how you’re working with it. And obviously, you had quite a bit of experience both on the SEO side and the i’s AI side, so I wanna just, like, open it up for you.

Federico Pascual
00:29:50 – 00:30:51

Well, yeah, with AI and SEO, it’s all about, you know, going from CU to 1 faster. It’s it’s not that AI is here to replace us or to allow us to just sit back and do nothing. We can’t could automate our job completely. No. That’s not the case. AI is just a tool that can help us do more and do do it faster and and in particularly going from 0 to 1 much faster. So for example, Microsoft recently released, the result of a research that they did on how Microsoft Copilot in Windows 3 6 6 365 was used and how it impacted the the work of people using Copilot in Windows. One of the findings that they found on this research was that more than 80% of the people claimed that they were able to create the 1st draft of a piece of content lot faster and significantly faster.

Federico Pascual
00:30:52 – 00:31:32

And that’s amazing. You know? He’s like, when you first start working on a new piece for SEO, starting is always a pain. You know? It’s always, really hard. And with AI, can we can do it at much easier and faster. Also, as I mentioned before here, it’s like, with AI, it’s all about spending less time on things that we don’t like or things that we got a lot of manual work. So for example, instead of spending a lot of time researching search intents, you can have the AI do that for you and share its findings. Also, again, you don’t have to worry about spelling, about grammar. You can have AI do that for you.

Federico Pascual
00:31:32 – 00:31:52

And, yeah, and the way I see this, like, as people working in SEO start using more AI tools as part of their toolkit, I think they will have more time to do things that not only they like more, but where they can add more value and where they can, you know, use their expertise to shine.

Filipe Santos
00:31:53 – 00:32:59

That that sounds amazing. And I, you know, I wanna I know you you don’t really wanna talk right about WordCrafter, but I can I can say from this point of, like, having benefit from the AI experience of, like, creating outlines and understanding how to be competitive with a piece of content because, as you mentioned, there’s a number of different ways you can go? You have to meet search intent, but on top of that, you also have to offer enough breadth of a topic to to make the the content valuable, or at least you have to deliver what the promise is of, of that particular search. So my experience with WordCrafter, and and thank you for that, is, is that it’s been extremely good at looking at how competitors fare in the space for a specific keyword or phrase. And then it kind of gives me a sense of where I need to go with my thinking and how I need to kind of build out a piece of content that’s relevant and particular, let’s say, you know, it’s particularly individually built around my audience. So I I guess with that, I I can kinda let you take it off and talk about, like, why WordCrafter has gone the way it’s gone and and why it’s so good at coming coming up with quality content for

Federico Pascual
00:33:00 – 00:34:02

SEOs? Yeah. Because, you know, a lot of things about SEO is writing the content that people need when they are searching for a particular keyword. And as I said, you know, AI is really good at analyzing vast amount of data really quickly. Data can do that seconds, you know, things that can take humans, meaning so even hours or days, you know, where I can process that information, analyze it, and get insights right away. So, yeah, that’s how I think WordCrafter can, you know, help people working in SEO, you know, to understand really quickly what are the search main search intent for a target keyword, understand, identify what topics and subtopics you need to cover in your content. And also, with ForeCrafter, we are putting a lot of focus on creating a user interface where the focus is collaboration with AI for creating content. Because in many cases, for ranking in SEO, you still need creating world class content. In particular for really competitive keywords.

Federico Pascual
00:34:03 – 00:34:45

Creating world class content, it doesn’t matter if it’s generated by AI or a human. It’s all about the value and expertise setting that it has. We believe that we should not rely on humans or AI, but we need to take advantage of the strengths of the 2. Because AI is really good at analyzing data really fast. AI is really decent at creating, like, first versions or first draft of things and writing these things down. But on the other hand, humans, we have the unique expertise. You know? We have the experience, the knowledge, the know how. Also, humans, we have the unique perspective and the unique voice on things.

Federico Pascual
00:34:46 – 00:35:26

So we put a lot of effort into creating a user interface where humans can collaborate with AI to take advantage of of of these things. So maybe you want to have the AI take the first stab at something and write the first graph, and then you can take a look and iterate it. You can do that with World Crafter. Maybe it’s the other way around. Maybe you want to just dump your thoughts on a particular subject that you are a domain expert and don’t worry about grammar, you can do that and then have the AI rewrite that into something that looks really good. So we put a lot of time into a product that maximized collaboration between humans and AI, again, to make world class content faster.

Filipe Santos
00:35:27 – 00:35:46

Taking a step back because I I was fascinated by the tool itself. Right? There’s there’s a 1,000 other content writers, but the there’s a different approach to to WordCrafter that I kind of really appreciated. But take me back. What was, like, the how did you come up for the vision, around this, and and how did you kind of think about solving problems with this tool?

Federico Pascual
00:35:46 – 00:36:38

It all started about thinking about what problems I have in the past while I was working, for example, in the SEO. But what are the challenges in there? Figure out if there is an opportunity for AI to help in those steps. You know? Once was I I was digging into these notes and these spots. You know, I realized that many of these challenges that I have personally in the past and that I really suffered in the past, By using AI, it can be less painful, even more enjoyable. So that it’s it’s all it’s how it started. Long term here, our vision is I honestly believe it’s all about collaboration. It’s not just relying on AI on everything and you just sit back. It’s all about the symbiosis of humans and and AI, and I want to build a product that takes that to the core.

Federico Pascual
00:36:38 – 00:36:55

Right now, we are focusing on SEO and creating long form content. But as we mature the product and we grew and have more resources, our idea is to expand to adjacent spaces with that core concept of enabling collaboration between humans and AI.

Filipe Santos
00:36:56 – 00:37:20

Alright. And and and with that, I know that there’s so been so much change, and you’re continuing to evolve the product with that change, especially in the industry of AI. It’s the age of AI. But, basically, as it continues to change day by day, week by week, and month by month, what do you think are the skills that you feel are most valuable for SEOs to pick up and to really get good at in order to be relevant and to to upscale them themselves?

Federico Pascual
00:37:21 – 00:37:57

That’s the golden question. This answer may change in 1 mile or in 1 year. You know? Obviously, it will change depending on who you ask. But, personally, I think that there are, like, 4 key things that we can do to take advantage of this basic revolution that is happening. I think first of all, I think it’s really important specialize in one niche. Like, become leading experts in a particular area. You know? That is more important than ever because AI is really good at understanding topics, like, in a superficial level. And we’d feel keep keep better at that.

Federico Pascual
00:37:57 – 00:39:16

But AI still struggles and will struggle for a long time to understand really niche concepts that are either are new or they are not as much information available or that are really complex, and you need, like, deep thinking and deep listening to fill things out. So I think it’s a way to differentiate ourselves and take advantage of of of the current situation is yet to be really good and specialize in a really deep, niche. A second thing that I think it’s super important right now is, like, right now, everyone can create consent create content that looks decent. It’s like, for example, if you go to LinkedIn, you know, you have a lot of people talking about the same thing in a similar way. So right now, I think it’s super important to differentiate yourself is to have, like, unique voice and a unique point of view because now the value is not just creating concept creating content around a particular topic. Anyone can do that. It’s all about having a unique voice and unique perspective on things, a unique angle, a unique opinion. I also think it’s super important to level up our editorial skills, like a movie director that you’re what scenes you need to include, what scenes you need to cut, what silences you add, or where you add the music.

Federico Pascual
00:39:16 – 00:40:37

With content and marketing in general, the use of AI will push us for us, instead of doing the modern work of writing every word in a piece of content, for example, we will take a take a step back. We’ll play a role that is more related to, like, a supervisor or an editor, you know, that, you know, that can guide the AI to towards a particular outcome, towards the AI to join a particular content that mention this point of view or covers this topic or mention this data. So I think more and more, it will be super important. Not the ability to write everything ourselves from scratch, but to have, like, that editorial mindset or what makes things a piece of content great. And lastly, I think it’s key to take advantage, of these AI tools and use them. You know, it’s it’s like we can save a lot of time by using AI to automate all kinds of things that we currently do day by day so we can better use our time, you know, either in focusing more on the things that we enjoy, focusing more on the things that we are good at, or that where we have a unique expertise or unique point of view. So definitely, you know, it’s super important to adopt these technologies and and surf the wave.

Filipe Santos
00:40:38 – 00:41:00

Now you brought me to a point in time where I’m I’m like, okay. And not to put you on the spot, but I’m very curious to get your on a your your take on, like, what’s happening in the future in terms of, like and in the by future, I probably mean 6 months in terms of SEO and and AI, particularly as it goes with LLMs. I would love to get your take on what you see.

Federico Pascual
00:41:01 – 00:42:20

You know, SEO will change dramatically because, search narrative experience from Google, s a SGAE is here to stay, and the same thing from chat GPT is also here to stay. The way I see this, like, there is no doubt that traffic generated through SEO will decrease over time. In my mind, it’s not that generative AI will lead our launch completely, but we need to understand that the future is not a a place where people just search for stuff to answer their questions, to find information. We do need to understand that the future is where in some situations and in certain cases, generative AI will do a better job at answering a query and providing the information that the user is looking for. But, also, there are many other, situations, where, you know, the traditional search, you know, the tailings will still do a better job, but and it all depends on the on the search query. So, definitely, we will see a a decrease in traffic generated from AI. But there is, like, a an optimist in me in the sense that maybe we we will see I don’t know. Nobody knows what will happen, but maybe we will see a decrease of 50 50% of the traffic that we get through SEO.

Federico Pascual
00:42:21 – 00:42:59

Another way to see this that maybe thanks to s through maybe thanks to AI, doing good SEO will be, like, 50% more productive. You know? Instead of doing everything from scratch and spending, I don’t know how many hours researching, I don’t know how many hours writing, editing content, and producing content and launching it and distribute it. Maybe we’ll do it in, like, 50% faster. So the return on investment will be quite similar. You know, we will see definitely a decrease in in in search volume, but also we we will spend less time producing good SEO and producing results, thanks to SEO.

Filipe Santos
00:43:00 – 00:43:45

That’s a great perspective, and I’m actually it it makes me think that maybe the type of work we’re gonna be doing is gonna be closer to revenue generating than it is to informational, which is, again, much easier for AI to handle than than the conversional part part of things, right, the transactional part of things. But, you know, who knows? We’re we’re kind of watching that, and I think we’ll have to watch you to see also what what’s changing because you’re at the forefront of that. But I wanted to thank you, Federico, for so much for, like, chatting today and kind of more importantly, uncovering a lot of these key aspects and information and methodologies and ways that we could apply AI today for this audience. And, you know, before we sign off, I would love to kinda get an idea of maybe how folks can get in touch with you or can start using Wordcraft or AI or learn more about you or follow you.

Federico Pascual
00:43:46 – 00:43:54

Yeah. Yeah. Thanks. It was a great conversation. Thanks for having me, yeah, for reaching me out. I’m on Twitter. I’m on LinkedIn. Just search for Federico Pascual.

Federico Pascual
00:43:55 – 00:44:09

You can find me there. I’m happy to chat and have a conversation. For word crafter, yeah, it will be great if your audience can go to word crafter dot ai and request an invite there, and I’m happy to give access to the platform so they can give it a try.

Filipe Santos
00:44:09 – 00:44:34

And I have to recommend it because I’ve loved it so far. I think it’s, it’s one of the best quality tools around building content for SEO that I’ve I’ve used so far. And, Federico’s amazingly, you know, like, on top of things. And if there’s anything weird, he comes and he addresses it, and and the platform has just been getting better and better every day. So appreciate it, Federico. Thank you for your time, and I’m looking forward to seeing what happens with WorkCrafter in the in upcoming months.

Federico Pascual
00:44:35 – 00:44:36

Thanks, Filipe.

Filipe Santos
00:44:37 – 00:44:38

Thank you.

Federico Pascual
00:44:38 – 00:44:39

Thank you. Bye bye.