If you are not an AI marketer, you will lose your job. So that's an alarming situation. Like in the next five years, if you are not using ai, somebody will replace you. Who is using AI better than you? So there's an existential crisis for any marketers right now. It's not an or. It's like you have to create the end like you exist with AI together.
Hi, this is the marketing meeting and I'm your host. Every two weeks I meet with experts and we talk about topics related to brands, marketing and businesses, and we sometimes add random lifestyle topics too. I hope you enjoy the show. Welcome to the Marketing Meeting podcast. Today my guest is Nouri. He is the Vice President of commercial marketing at Intel. He leads marketing for ai, data center edge, and commercial client segments.
He's also the author of the book, AI in Marketing, which is released almost two weeks ago. Welcome.
Thanks a lot for having me.
So in your LinkedIn profile, you mentioned that your current passion lies in commercial and ai. AI products, marketing. So we are gonna talk a lot about AI and marketing and especially there is one nuance there that I'm would like to ask a few questions about. AI products marketing, which is something that we started seeing more often and which is absolutely. Uh, which is still not there yet at an execution level, and there's like too many things to do over there.
Uh, I would like to start with the question, since you've been working with AI and marketing a lot, which functions in marketing are changing the fastest by AI at the moment?
Yeah. I mean, is changing. Everything at the companies, and I always say every company will be an AI company, but starting with the marketing group, I believe there's a huge impact in every aspect of marketing. So just giving an example from my existing team, so like we do a lot of product marketing. First of all, as you highlight that, I love all the AI products in the market and I closely work with our partners across cloud solution providers and the OEMs. But the key thing is like.
From writing a product guide to a sales guide, to a website description, to a social post, there's lots aspects of product marketing because it requires all this like messaging and positioning, framework design and in all aspects of that, like AI is really a key part. But beyond that, I think, I always think, uh, you have to put those into different categories. Like especially my team works with our partners a lot, so partner marketing is really key. Like. Intel is an ingredient brand.
So like we don't sell directly to businesses, but we sell through Microsoft AWS, Google, Lenovo, Dell, hp, ac, ACEs, like you name it. So that's the intel insight of every technology company and we try to bring that AI inside whatever we develop from a partner marketing perspective, all the collaterals, all the go-to market execution, all the social feeds. Everything reflects our brand guidance and brand narrative. Mm-hmm. So from that perspective, we use a lot of ai.
So I feel those are the areas from product marketing to social media management, and also partner marketing are the key areas. And I see many other parts of the execution, especially influencer marketing and thinking about the brand marketing. So how do you really create that brand and make sure. The awareness is there for the long term is really critical and you can drive a lot of AI solutions there.
But like all up, I think one of the key departments in an organization to adapt AI is marketing and it'll be a fueling growth for the company, for the AI era. So again, I'm super excited that my team and myself, we are using it. Not only at Intel, but before when I was at Microsoft, we were really keen to understand and implement all the AI solutions. But that's the tiering in my mind.
So starting with simple steps with like less risk high ROI, and then moving into more like the partner marketing, brand marketing, and maybe even more on campaign execution, account based marketing. So again, we can unpack that all, but I feel. That's an like a step-by-step implementation process. Mm-hmm. For all the marketing leaders,
one of the questions that I receive a lot is from the companies that I work with is where should we start first? Because there are so many areas that can be impacted with marketing and that can be evolved with marketing, but they don't know where to start. Yeah. Is there like one answer to that question or,
I mean, I show what would you commend the book? Because like I will refer to this. So again, for the listeners, I just published the AI in Marketing book and in the book I have this framework. So it's a very simple three letter acronym, aim. It's assess, implement, and Measure. So I mean, back to your question, the most critical step is the assess phase. People try to jump to the solution like, oh, like I'm using copilot, I'm using Chat GPT. It doesn't mean that you are really using ai.
In marketing. So I wanna make sure first you do the assessment on like what are you trying to achieve? Which areas, first of all, in your daily work can be better optimized with AI or leverage more with ai? And then what's the team's level of skilling on ai? So without doing that assessment, it's really hard to go and jump into the tool conversation. So that's why what I recommend to every. Marketing leader listening in is make an assessment and start with your own team.
So not everyone is at the same level. Mm-hmm. And I experienced this firsthand with my team here at Intel before, at Microsoft. Some people allow to test everything in the market, so I. There are a couple of people in my team, like they're using copilot, Claude Gemini, like they have tens of maybe tools, but some of them are really reluctant, like, okay, how do I maybe upskill the team to be at the same level? And not every tool is required. So tooling is the implementation stage.
But before that, like what's your daily tasks look like? Let's understand what's your visual, creative design, timing? What's your, let's say text generation timing, and then. Can AI help you to drive that? Like some of the campaigns we do AB testing, we do broader testing. So can we accelerate those with ai? So these are the questions you can only ask if you have what is done today. So I think like logging everything, what you do as a group or an individual is a critical step.
The second big thing for the enterprises, of course, getting that. Secure and confidential AI built in because many of the companies jump and start to use the, uh, existing tools and share all the company data, which is wrong, especially in the world of ai. The data is the most important assets the companies have. So especially for the marketing teams, they have to know what to share, what not to share. So that assessment phase also needs to include like.
What are the tools to use and what's the data sets and customer like information that you can really use to train some of those models? So yeah, you have to handle the customer data really cautiously, and you have to protect the customer data and also you have to protect the. Enterprise assets, which are the proprietary data for the company's growth. Mm-hmm. So my short answer is exactly the first step is making the right assessment.
Yeah. I mean, there are so many questions that I'm gonna ask after this, but the first question is because, especially for smaller companies, the first reaction from marketing teams is just like to go ahead and start using the. Tools that are like chat, GPT, Gemini Cloud and so on, and I'll start using them the other day, I found myself for one of my client. I was about to upload a strategic deck to get some feedback from ai, and then I said, I mean like, so I'm doing something wrong.
I stopped myself, but because I have this experience of so many years of knowledge. But then you mentioned that in terms of security and protection of data, the companies should look into building their internal AI tool systems in place rather than utilizing. Like companies like, you know, chat g, PT or Cloud and or Google or the other partners, Microsoft as tools.
Absolutely. I think there's a definitely a hybrid solution here. So like, I wanna maybe give a quick glimpse about like how the AI systems work, but first majority of the solutions require a big training. So, yeah. And again, if you look at chat, GPT. They're training now GPT five. It'll take like roughly four to six months in an average to train all this. It's like more than like $10 billion costs.
So as a company, you cannot do that training phase, so you have to use A GPT, which is generative pre-train transformer, so you get all the knowledge. But what is critical today is like the AI agents and rack retrieval, augmented generation, just to keep it very simple. Like you have to make that assessment first. Like what's a company data and what should not be on a public cloud?
Because whatever you share with the public solutions, you might become a part of the training set for the next GPT and we really don't want that. So yeah, we have seen some examples of in the early days of copilot, like some customers used dish. Own like code base and put it on a cloud solution and ask like, Hey, can you optimize this? And then another developer can go and like, can you write the application exactly like this company, it's called Reverse Engineering.
So you technically use the system to decode what the other companies proprietary like coding is. So. From that sense, I think that will be a hybrid way. So when I say hybrid, like there will be some applications that should be run on premises. Mm-hmm. Especially when you are handling any customer data, any financial data that you really want to keep behind the firewall. And this is exactly the way that cloud computing happens.
So you don't put everything on the cloud like there's a. Privacy concern. Like sometimes there are regulations that the data should not leave the national land. So again, like you are bounded mm-hmm. Into the countries borders. So it applies to, same with ai. It's a little bit fill with today, but I wanna highlight like first, there are great open source models right now and the world has experienced a deep seek moment in January. So, which was a big moment for.
All the others to follow because I believe if you look at the implementation, the basics, like, I mean, getting a text generation, getting very basic image generation. Can be done locally. You don't need to go to a big solution and create some confidentiality issues with the company. Mm-hmm. So you have to work closely with a IT team and create a local machine to do we call that inferencing. So like.
You have an existing model, which is trained, but you can use that, download that open source model from like platforms like hugging phase and use it locally with that firewall. So your company data never goes beyond the company firewall. So this way you will be protecting the company data, especially for marketing. The key thing is creating a generative AI usage guideline for the company. And I see some examples like some companies who don't want to. Spend the time, they just block the domains.
Like you cannot go to a chatt.com domain and it's wrong. Like, I mean, you cannot block, people will bring their own devices. They will use their mobile devices or personal laptops to do this. Again, I call this a shadowy ai. Like, I mean, you, we have seen this in the era of applications when all this Android and iPhone applications are popular. If they don't get the same apps on their pc. People use their own solutions.
People are getting like $20 subscriptions and they're uploading the documents. So we want to make sure that this is done in a really hybrid way in terms of, again, the first step back to the assessment. Make the assessment, which tools to use locally and which documents to upload safely to public solutions. So I think that will be a balance. Between two, but it all starts with that assessment phase on what goes where, in which situation.
So if you nail down your use cases, especially for marketing, it's so easy. Like for text generation, use this for image generation, use this for video generation, for example. I. It's almost impossible to have it on your local system because you have to buy almost a million dollar equipment. Yeah. Which is not gonna be worth the hassle. So you can use the generative AI tools on the cloud. Mm-hmm. But again, while generating those maybe videos or like some. Different modalities.
Voice might be another one. Make sure you don't share any private information that might go public. So again, there's a balance, uh, for the marketers there. Uh,
I'll simplify for the understanding of the people. Like with a few examples, the most common thing for a market search for a brand coordinator to do is to write social media post copy or a copy for whatever marketing campaign. And for that. They can easily utilize chat GPT or some other things, right? Correct. As long as they don't upload. Strategic documents. Yeah. And they can utilize those.
And even for those type of simple things, do we need some kind of a local solution leveraging like an open source model, or what would you suggest for those? In
my daily life, I use my PC a lot. And at Intel we have this like. I PC movement. So many of the NIV PCs, the personal computers, are now having some neural processing unit and a graphical processing unit on the device. Mm-hmm. So we will see more shift to the device itself because it is behind the company firewall and you can do a lot of things today on small language models. So, and this happened like historically multiple times, like when the. Internet happened.
You had like big data centers, big servers, and then like now your device, let's say a mobile phone had more processors than any other pc. And then the cloud computing happened. We added like a lot of data centers and we had AI. In the next five to 10 years, all these big models will be able to run on your mobile phone and on your personal device. So this is always like client server.
I mean, there will be more data, more compute on big systems, but it'll diminish and then really come to the client. So back to your question, I believe many people will start using. Devices, especially personal computers. All the devices from Microsoft, I mean, many people heard about the Microsoft copilot. There's a version called the copilot Plus. You don't have even a go to a cloud service.
It runs on the device itself, which is very secure way of using the AI features on really basic things like generating image, creating meeting summaries, and these are like simple things. I do have a busy calendar like many others. And I just asked like, Hey, what was the summary of this meeting, which I missed to attend? And copilot gives me all the summary and everything.
It's still within the boundaries of the company and it gives me a lot of productivity gain because now I can follow through all the conversation without going through 60 minutes of listening all the meetings. So these are simple things, but I believe there will be more PC related ai, like on device ai, and we call it the Edge ai, by the way. This is gonna be a big area that is growing. And maybe one last example I will give is from healthcare.
Imagine your MRI scans all the x-rays, like these are really confidential information and you can use now the devices at the edge and then this. Can run AI models and now like the breast cancer can be detected five years before because AI can detect the images very deep and using different data sets than a human doctor can. Yeah. Mm-hmm. So it's really amplifying the doctor's ability to make a statement by using ai, but.
Using the AI at the edge, which is like right at the hospital, your data never leaving, that premises directly running on the machine itself. So again, this is a good example of like how healthcare can be transformed by ai, but there will be more and more I. AI moving to everywhere in our lives.
Mm-hmm. How about, let's say things that concerns data, customer data, and then even for small companies, you would, I guess, not recommend any third party application to be used. Right.
Again, there, there should be a balance. Going back to the assessment, for example, you are. Planning to do a LinkedIn post for your company? Mm-hmm. That it, the outcome will be completely social post, which is open to world public. Yeah. Same with X or meta, you name it, like whichever social platform. So you and the context window that you'll provide, maybe again, it's a public data. You can just say, go to this website, which is your company's website. Look at this page.
And this is the key points that I wanna highlight in the social post. Write me the post. Yeah. And generate an image. Make it like more with a hook on top. So create an infographic for this. I'm telling you. 'cause I'm doing it every day. So like people are forgetful. They just like forget. Like I write an article on LinkedIn, for example, on shallow ai. It's long and people say, oh, like this is so long I cannot read it. Okay, now I created a video version of it for 90 seconds.
And then they say, oh, like this is too short. Make this like 15 page carousel file for the users to be more like easily digesting. 'cause some are like visual learners, some are like learning by in depth. And then yeah, like I think that will be a lot of tools. So I'm not saying. Don't use the, like the Geminis chat gpt or clothes off the platform, use them, but know what are you using it for? Yeah. And if it's experimenting is totally fine, by the way.
So the best way to learn these tools is just. Going to them and like asking some prompts. And you have to be getting better on the prompting because that's the new way of interacting. So you can write a program without knowing any coding skills today. Mm-hmm. Which is amazing. Same with the marketers. Like they can do amazing like messaging and positioning frameworks, like at 20 year marketing executive with the help of GPT tools. But again, you have to know.
What is like gonna be really implemented and what is a learning case study for you?
Mm-hmm. And what about, for example, I have been to an AI marketing conference in Cleveland this year, and on one of the sessions the speaker was talking about how we can upload a customer. To chat GPT or like cloud anyway, and then get good segmentation, but that involves uploading some customer data into the system itself. What would you say to those type of applications? Because it's about you just.
Upload your customer data to get a better segmentation, customer segmentation from the ai, but it's public at the end of the day.
I mean, you have to be really reading all the terms and conditions of those tools. Like everything comes with a price. First of all, if something is free, I. It's always told, but you are the products, like, I mean, they, they're selling your data, so if something is free, you have to be extra cautious. I'm not going into open source because if you download the deep seek model and run it locally, your data is not going anywhere.
But if you go to deeps, seek application and upload your documents, believe me. Yeah. Though they are going to China. So like, I mean, by showing that code it is. Definitely showing it is doing that. Uh, and same applies to the other tools. So the companies are selling this. Additional layers. The good example is like chat g, PT versus copilot, so mm-hmm. Chat, GPT is a public open tool. They're coming with like more enterprise tools.
Again, they will have more and more like privacy, security concerns, but at the same time, Microsoft is using the same chat GPTs like 4.5 today. Mm-hmm. And you creating this as a co-pilot for the companies to use internally. And you know that when you use the co-pilot. It is within your domain, so again, mm-hmm. That will be more like customer stories like this, uh, because I see chat g PT as the, the ultimate platform owner with GPT, same applies to the Gemini, same applies to.
Tropic, like these big companies really provide like a generative pretrained transformer model and the model is used by other companies in different ways. The best example is Lama from Meta and as a marketeer, again, meta has like ups and downs in our lives, but Zuckerberg is doing a fantastic job on embracing the open source, making sure a ai AI is led by meta with the Lama. 90% of the. AI Solutions today runs on lama. So again, there's a huge startup ecosystem coming there.
So like I know it's the early years, it's really hard to do a due diligence when you choose a vendor, but at least in any company, like I don't really recommend like uploading your customer data for a campaign. Without knowing what's happening, try to do your due diligence, and especially don't be a part of the broader training data set, because then you're risking not only your company reputation, but your customer's data.
I think one of the things that will boom in the upcoming years is cybersecurity. People are gonna use most of this leaked information on. Passwords, address information, name, social Security, you name it. They will try to sell it to a broader audiences. And again, it might be AI is buying this data to do something with it, so the security threats are just gonna explode. So you have to be really cautious on. What's the ultimate goal that you are trying to achieve in the short term?
It might look like again, but just spend the time on the assessment, choose the right tools, and then start implementing. So I think don't like try to shortcut the first step of assessment because that's a really key of implementing AI in marketing.
Mm-hmm, mm-hmm. So in that three model framework, so one of them is aim, which is like what's our goal, I guess, right? The other one is assess, aim is all,
all up framework I come up with. So assess is the first stage. The second one is implement. Huh? Implement is really critical because then you choose the tools. Yeah, like personally, like my team, I do a lot of tools, but I try to explain on the book about like the top ones. One of them is perplexity, for example, so, mm-hmm. I like the term Ravi, the CEO calls it like the. The find engine.
So we have search engines today like Google, but Perplexity is using some real data to really give you the answers directly rather than you are searching, you are finding the answers. So it's an answer machine that gives you the answers, but of course it does it with the tooling called the retrieval Augmented Generation, where. Even you can add your company data in a secure way and it search for all the documents and everything. One of the tools, for example, I recommend is Jasper ai.
So Jasper is an interesting tool where you say, this is my brand identity. Just like the basics of these are the color tones I use, like the blue and shades of blue. This is the company branding tone. Like innovative, like all this words that we use, like energize, ex excite, blah, blah, the visual color, like design principles, like we use real people, not animations. We use this diversity in our imagery and everything. And then the company logo always goes to the right button.
And when we, and our, let's say press release, as we always end with our company statement, you just feed them as a. Like a learning process, and then you can just go like, can you create me a flyer for this event? Boom. It looks like your brand engine worked because the brand considerations are uploaded inside the model.
Yeah.
You don't have to worry anymore because all this like text generation is done with the words that you highlighted all the imagery comes up for, and you can just say like, create 10 different images to show on a social media. And if you. Work with an agency, it takes ping pong. Like, I mean, you just say like, ah, I didn't like it. Revise it with a chat tool. You can just like chat with it. Okay. Like, make it brighter, put this image behind the pc or like add some buildings to the image.
And then it's still keeping the, the framework in the mind of like, Hey, this is your company profile and this is the imagery, this is the font, this is the logo appearance. And then everything really applies fast. So. There are marketing specific tools like this, and maybe the third one, and I will stop there, is HubSpot. So like we use lot of the account based marketing. Mm-hmm. We wanna go deeper into the customer journey.
It's not just like creating the top of the funnel with like demand generation, but really closing the deals, following till the end. Mm-hmm. And you require that. Like nurture with emails, like social integration, all the webcasts like, I mean, there's multiple joints that you collect the data from. And HubSpot has this solution with AI that you can really interact with it. Like, Hey, where is this customer X with the journey? What's next best, uh, reach model?
Can you generate a customized model? And imagine you are creating an a BM account based marketing journey. Customized to each individual. Mm-hmm. Today, it's almost impossible to do it in a manual basis. Like if you have a thousand customers with a five people in that each account you have like 5,000. Reach points just for a single customer, single product. And if you are a company like Intel, we have billions of people to reach out. So yeah, it becomes impossible. But for AI it's not.
So we can send individualized content on different channels and we follow through. Like, I mean, you have seen this on Netflix now we catch you on LinkedIn. So like it really connects like wow, I mean. Intel knows how to work with me, kind of a impression. And then you really go through the funnel approach. So again, like there's lots of tools that we can highlight, but making that implementation stage as a next stage is critical.
And you have to learn from experiences and you have to upskill your team on the Jasper usage, copilot plus usage, HubSpot usage. So you really upskill everyone. So everybody has access to the. Greatness of AI tools. Mm-hmm.
Mm-hmm. What, how do you see around in the marketing teams? Because already marketing is a lot of work, like so many things to create every day, and then it's sometimes feeling like we are saving the world. This is the same thing I'm keep on saying to people, but now we are. Entering the big era, which is like you have to assess your needs. You have to find the right tools, you have to implement using those tools within the team, and it's just like it comes up as a big other function.
Living in the marketing team, what would be the solution to that? I mean, is like getting a outsider consultancy or like a team or like building the team inside the, of course it depends based on the company, but is there any, uh, best practices that you see around working with so many other companies?
Absolutely. I mean, it's all about creating the urgency. Mm-hmm. I believe we have to be really bold on this one. We. Like, if you are not an AI marketer, you will lose your job. So that's an alarming situation. Like in the next five years, if you are not using ai, you, somebody will replace you who is using AI better than you. Mm-hmm. So there's an existential crisis for any marketers right now, so it's not an. Or it's like you have to create the end, like you exist with AI together.
So yeah, it is, there's lots of concerns in the job market today, and I believe it happened across all the technological innovation, like when the internet happened like two thousands. Mm-hmm. So people said like, oh, like I mean internet. Like I will continue to sell my books on the bookstore. I will continue to be a brick and mortar and look at what happened with Amazon and look at all the industries. That has been impacted by that.
So similar and like I was responding to a Gartner analyst and they were asking me like, what's your traditional media spend $0, like I don't wanna be on. And we spend a lot of money on marketing through our partners. So like, we wanna make sure we customize our journey on this digital channels and we go as deep as possible. On the account level, engagement by individual engagement, which means that without the help of ai, the future is not possible to add the value.
And there will be people who will use AI on their daily lives every day today. But if you are not doing it, like spending the time to upskill yourself, it'll be really hard. So the best practice that I recommend is. Create that virtual team, tiger team, whatever you wanna call it, but like get the first influencers on the spotlight. Mm-hmm. And again, making sure that this is an inclusive environment.
Like if someone, let's say Jane in the team is like using the tools, get her a, maybe a spotlight like AI in marketing workshop within your group and show that. Mm-hmm. The others that like, hey, this person is using. What are you doing with AI today? And then next round, get the others, and then it's like a turning table. Everybody gets a spot, so it creates an urgency first. People will see that, okay, the rest of the team is really spending some time and they're using AI agents.
Oh my God. Like I need to upskill myself. Yeah, because I believe everybody has time. It's all about prioritization. So, and these things only will work for the companies if you spend the right amount of time on experiencing them. So just use it for your personal life and then start to see the changes. I. Then like follow the key people. I do share like every day, literally for the last almost six months.
Mm-hmm. One thing about AI in marketing, I try to share like how to create an AI agent and there are like many people like me who are really willing to share this information. So just follow the right people on the area and then again, pin there maybe. Or bookmark their post. So I think the best way of implementing is using the tools and really showcasing to the others because that's the next stage.
So, and if you have that urgency, like hey, there's a virtual team or a tiger team developed, and when you show this to your management, I think that will create the big change. So the last letter is measure, because everything you do, you need to measure it. So, and marketing success is always like. If you are not putting the right return on investment measures on any activity, you will be judged by the sales team because sales team never loves marketing. They say like, you are a cost engine.
Like they should give you the money to us and we will hire more sales representatives. Uh, majority of them. Again, I don't, there are great people who knows the value of marketing, but you have to show, like, you have to say like, marketing is helping you. To drive the sales. So we are like yin yang, like making sure we are working closely to drive that customer experience and really helping the company to grow on more profitable way.
And this is the only way that if you set the right measurement mm-hmm. Like I will double the reach. Of the customers. I will increase the sales pipeline five times with this NIV AI project. And then if you set the post and then like you shouldn't change the goalpost, like you should keep it there and you have to overachieve that. So yeah, I'm a big fan of this OKRs objectives and key results. But with AI, you can really quadruple the outcomes. And when you show it to the sales leaders, like.
Last year I was driving a billion dollar pipeline with the help of this AI campaign. I made it $4 billion. I mean, they will just stop. Mm-hmm. And they will say. How can I use AI more for other products? So I think it's just that recursive thing. So, and then you have to go back again, like do the assessment for the next project, implement it, measure, and then measure it. If you don't measure and if you don't communicate the results, you will not get successful implementation.
But if marketing leaders does this properly, they will get. More funding from the leadership team on doing more marketing stuff because they will see that the impact is like quadrupled with the AI solutions implemented into the marketing journey.
Mm-hmm. I wanna ask one question because like when a company does a marketing campaign or like the regular social posts, like day-to-day things, it's. Easier, comparably, easier to implement it. But when it's a joint marketing campaign with some other companies, then it's another thing. And I know that you are running joint marketing campaigns with big companies like A-W-S-H-P-A-A. So how does it work there? I mean, like how do you, what makes a good story?
Technology partnership deliver results on your, on our side. Yeah. FF
first of all, again, like we are trying our best, like there is no immediate success. We are all in this. Learning journey with all the other companies we are working with, like Lenovo, AWS, and many others, like because they're also implementing their solutions. The key thing is making sure our goals are aligned on what we are trying to achieve. And I always start, I. Any outbound campaign activity with that goal.
Like, Hey, we are launching this new product and we will put money behind it as a co-marketing. So what we want from the customers to see, the key thing is like the customer outcomes, and I'm a big believer of the outcome-based marketing. So we have to put like, is this creating more innovation faster? Or are we saving more dollars for the customer so they can do more things? So, mm-hmm.
A good example with AWS is definitely like when customers use the instances that we create as an example, they save almost 30% instantly. Mm-hmm. And we communicate that value. We say like. Use this instances on AWS and then save this amount with, for example, with Google, like we do a lot of inferencing with them, and then we say like rather than using a GPU, like you can use this CPU with an AI acceleration. So. We help them to innovate faster. Mm-hmm.
So it's all about saving money or making money. Like there are only two outcomes that the customers really care if you communicate it well with the partners. And if you align on the goals, on like, Hey, we will communicate this, and this is the exact thing, like, I mean, with this product. 28% performance is the key, and these are the customers that we will go and market that. This is the changing word. So you have to be really account specific.
You have to be really data-driven, and you have to be really outcome-based. So if you manage those three, AI will help you to get those tailored messages. And then you will not see a billboard campaign saying like, Hey, this is the big thing happened, but the related people in that specific account will see their outcomes and it's a bigger value with AI today. So again, I'm really a big believer of that outcome based marketing approach in that call marketing partnerships.
Mm-hmm. I would like to ask two more questions about. AI in general. One of them is agentic ai. Could you please explain it a bit more in the context of marketing and how it applies to marketing in this new world?
Yeah, absolutely. So agentic AI is the topic for 2025, by the way. Mm-hmm. So like we are living in this era of reasoning and the big change of like this reasoning models are, it's called the chain of thought. So like mm-hmm. What the models can do today. Is like you gave them a task and then they think like a human, and then they break it down into multiple steps. Mm-hmm. I will just made up like, how can you cure?
Cancer, for example, like, and then historically it'll just give you like a bunch of texts. Like curing cancer is like, and then like you get some text prompts, but I don't want it, so I want you to go and reason about it. So what it does is it's called the multiple inferencing. So it runs the model like, hey. So Nouri asked me about this. So what's the core reasons? No, like, go deeper and then find the core reasons.
Maybe the response will take six minutes to me, but it, it's following a chain of thought. Mm-hmm. To get the result. Like us, it's becoming more like, it's called a mixture of experts. So you imagine you have an expert on math, but literature and sociology and marketing, all these experts working together, they will craft a better response to you. So with AI agents, what's happening is you automate all these tasks. That you wanna do using AI in a flow format.
So that makes it really like, it's almost automating the marketing process end to end by using a chain of thought like we do with reasoning. You say, I wanna do a. Marketing campaign for this product, and I want this to be automated in a way that when customer X responds to this, find the blah, think about it, put the right wording, and then create a response. Send it through.
Generate an image, like you can't just like create a chain of processes to be followed by a single action from the customer. So it becomes like a, your work process as an agent. So you technically have an employee working for you 24 7 without moving a finger. Everything is automated. The key thing is like how do you define that agent to do that work? And if that goes like hallucination, where do you control it? So that's why we need the Perplexities of the world.
We need the Geminis of the world, and we need HubSpot, like all this tools to make sure it's connected within. But in my mind, this is the area where we will see. More innovation and more tooling. Where do I start? If you don't have any opinion, start from the AI agents. Like try a couple of predefined agents. Imagine these are like when you open PowerPoint, you can just start from a blank page or it gives you like nice and good templates. Mm-hmm. So AI agents are almost like template solutions.
You can just use some of them. Test them and then customize them for your workflow. Mm-hmm. And there will be companies, more consultancy companies who are gonna be providing AI agents to you. And again, end of this year, I think we will be shocked because there will be, again, I'm really a big believer of AI agents becoming into the workforce. You will say, I have 20 people in my team, 10 humans and 10 agents. So. The existing 10 will be maybe 10 plus agents together.
So because I believe everyone will be a part of an AI merge journey for us, again, it's a little bit scary, but again, mm-hmm. 2025 is the year of adjunct ai and that will be more. Execution happening on the marketing and some business operations front with that automation work that's possible with reasoning models.
Mm-hmm. And now, which would mean probably then we are gonna see a lot of marketing around agent AI companies, I guess. Right? Exactly.
Probably. And this will happen at the. Top level. So again, I can clearly see looking at Microsoft, I mean, they already publicly announced, like they will come up with agents. You can rent agents from companies like Salesforce today. Mm-hmm. So imagine Oracle is doing the same like these, these big ERP customers, SAP, one of them, like they do finance, HR, marketing.
Like a lot of the modules on this ERP solutions, you can have agents and then rather than just selling this agent one time, you can just create that workflow. And make it a chargeable thing for the business. And then they will do a amazing job by the learning and reasoning. So it'll be a good compliment to your team. Mm-hmm. And for especially small customers, this is a game changer. Like you can hire the best HR consultant for your work.
Best social media stretches for your work virtually because. Let's say these are gonna be like $20 per month. So I don't think you can even find anyone to work for two hours on a marketing space, but this is incredible. So that's the low of scale. Mm-hmm. I think we will see more agents coming from more vendors, but some of them will immediately reach that scale phase, which will go. On the DNA of the marketing departments.
Mm-hmm. A friend of mine recently asked me a question about agent ai. For example, I am like a marketer. I would like to hire an agent. And this means like, it's not B2B marketing, then it's a B2C marketing. How do you make sure an agent AI campaign is better than the other agent AI campaigns? Uh, and I asked him, what's the trust that you're trying to build? Because marketing is about trust. Is it like security that you're giving a message your agent AI is more secure, or your agent?
AI has more years of experience, marketing experience because it's developed by better marketers, something like that. The last one is about, uh, a GI, which is artificial general intelligence. Could you tell a bit about that too, uh, as we wrap up?
Yeah. Yeah, absolutely. And again, like I will refer one last time to the book, so I'm gonna read chapter 13 on the book again, like I try to explain it in a really simple word like. A GI. For the listeners, like today, we are living in this narrow ai, so AI is doing some specific tasks, great, sometimes very close to human, but with a GI, artificial general intelligence, AI will be capable of every skill that human can do. In every discipline.
So from physics to math, from social to creativity, like every aspect of human parity will be achieved. That's what we call it, a GI. Mm-hmm. So it's not over by the way, second, like it's like superhuman level. Mm-hmm. So you can really, uh, achieve that. So, which means that a GI will. Drastically change the world we live in. So it's not about marketing anymore. It's like, let's go up a level. It's the nation's level.
That's why today there's a big race on being the first one to finding the A GI, because imagine. There will be a lot of benefits of being the first one reaching out to a GI and it requires a lot of investment because you need to build that big hardware, which also requires lots of energy to run and also lots of energy to cool down because AI is consuming lot of energy and there is like solutions like Lucid Cooling nuclear.
Plants today it's using natural gas to be more environmental fluently, but like there will be a debate about that, but this requires a huge, huge like investment. So, and there will be few nations achieving this and there'll be few companies achieving this and the rest will, again, that will be agis across. So with that, this is gonna change the marketing world forever because you have. The superhuman. That's not just great at marketing, but that's great in every single aspect.
And again, this will not happen overnight, so it'll, in my opinion, again, just looking at the numbers and everything, we will achieve a GI in 18 months, so which is 2026. 2027. It's not far. So it might be either OpenAI, Google Meta, or we don't know. I mean, atropic like there's like six, seven Frontier Labs in the world who are working on a GI and one of them will. Achieve this, and then the others will follow.
So it's all about feeding more data, getting the fine tuning, and it's called the reinforcement learning. So you add that human touch and then like you add more and then the system starts learning by itself. So again, we are, we know how to get there. So now we know everybody's racing to get it done. So the critical thing is like 18 months is so near. Yeah, but it'll not. Instantly come to like, Hey, like I'm getting your marketing job. It'll take five to 10 years to be commercialized.
So as usual, it'll be used for many other things before it comes to business and a GI in marketing. But we have to be prepared. And then maybe the last thing I will highlight. I said I will not show the book again, but like I have a visual on the book. So even the feature is gonna be an interesting one because after a GI, we are entering to a NIV area of a SI. So again, a SI stands for artificial super Intelligence.
Again, like I will just show the page on the book, but like this is a three layer approach. So we are in the ai, we will go to a GI. But the next level is artificial super intelligence. And this one, we don't know when it'll happen. It might be 10 years, 15 years, maybe more, maybe early. Mm-hmm. But the artificial super intelligence is the area where AI is aware of itself. Some people call it singularity, and then it is overachieving any human by far. I will give you an example.
If you have a dog or a cat, they hear. Different frequencies than the human. So you cannot feel what's going on on the street. But if there's another dog, all the doors are closed, they hear it, feel it, and they start to bark or meow. So that's the way that they are even better here, like hearing capabilities than humans. Same with like Falcon, the Falcons, like. A rabbit from like 2000 miles above because they have this like, uh, special eyesight. We don't have that. We only see like limited.
So imagine you are feeding the AI with all this invisible line of sight on the hearing, smell all the vision. All the information that we cannot interpret as human beings and AI can. So we are creating that artificial super intelligence. So again, it might happen in 10, 15 years. Again, my bet is like 20, 35 or so. Mm-hmm. Like at least 10 years until we achieve that.
But the feature is going there and we are like creating a. Intelligence far beyond the humans for the first time in the history of life as we know it. So, mm-hmm. The future is super exciting. I think from being in this in the early days of marketing is again super exciting. My maybe final thoughts is like, make sure you upskill yourself, make sure you read books, you listen podcasts, you really try to be curious and.
Try experimenting all these things in your enterprise environment as well, and create that virtual teams, because if you don't, you might lose your job. Your company might be out of business because only the companies and the people will survive who are embracing the change with ai.
Mm-hmm. Other than your book, is there any book that you highly suggest to. It doesn't need to be related to AI at all, but what's one book that you would recommend to listeners? There
are a couple of books I can pick from my background, but like this is my, one of my favorite books, winning the Week. Oh, okay. By Deir and Carrie. So this really helped me for personal productivity. Again, like AI can help you with something, but you have to control your time. So that's one of the key books. I need that book. Need the second book I. Uh, Eaton Mo's co intelligence. So this is like a December 20, 24 book, so like a couple of months back. But he's a professor.
And then this book is really written to explain like, how do we coexist with ai? So this is like mm-hmm. In the age of a GI, so
Nice, nice. Especially winning the week. I need to read that as I'm having difficulty managing my time. Absolutely. And since you're in Seattle, what's your favorite coffee place in Seattle?
I mean, I live in Kirkland, so again, we have a special coffee spot called Zaka Coffee. They have this like blend of Mediterranean and also the American Oh, okay. Culture mixed together. So yeah, like I like that place. It's in downtown Kirkland. It's a nice spot that I meet my couple of colleagues from. I, I work remotely. Mm-hmm. My team is in Santa Clara, Oregon. And like, again, it's a remote world, so it's always great to go.
For a coffee and yeah, like I totally recommend the chi latte at Zika. So they have a great, uh, brewing there.
Mm-hmm. I hope I can meet you there one day too. Thanks so much. Awesome. Yeah, absolutely. Uh, for joining me at this podcast.
Yeah, thanks a lot for having me.