What Happens When Companies Make AI Mandatory?

Two years ago, Ajay Malik told his team to use AI for programming and made it company policy. We said the skill is not a differentiator anymore.

Clips and outtakes on Instagram ›

In this episode · 18 chapters

  1. The Shopify memo
  2. Ajay's first aha moment, 1989
  3. The four core capabilities
  4. What Google did, and what AI adds
  5. Don't bore us, get to the chorus
  6. Where do you start?
  7. The gap between silicon and carbon
  8. Should we give AI a margin of error?
  9. Human in the loop
  10. What the Shopify CEO probably meant
  11. Every business has a unique DNA
  12. How an AI agent actually gets built
  13. Within a week
  14. Managing the intern
  15. A seven year old with the world's information
  16. Cloud, and the security question
  17. The eighty twenty principle
  18. What should parents tell their children?
You will be surprised how many people actually want to offload their knowledge so that they are not the bottleneck.
Ajay Malik

Ajay Malik, founder and CEO of StudioX, responds to Shopify's memo telling teams to prove AI can't do a job before seeking headcount. He argues for using AI to remove the repetitive, the boring and the tedious, while keeping the human in the loop.

He calls Google search a sixth sense and AI a seventh built on top of it, explains why every business has a unique DNA, and likens AI to a seven year old child with the world's information at its fingertips. His advice: learn how to ask.

Ajay Malik, founder and CEO of StudioX, made AI use mandatory for his own team two years ago. He tells the Thinking On Paper podcast that businesses should start with AI by handing it repetitive, boring and tedious work, with enterprise knowledge search first. The episode opens on Shopify's AI mandate, which asks teams to prove AI cannot do a job before hiring.

What happens when companies make AI mandatory?

Ajay Malik made AI use mandatory for his own team two years ago. He says businesses should start by handing AI repetitive, boring and tedious work, with enterprise knowledge search first.

Why did Ajay Malik make AI use mandatory at his company?

Malik made AI use mandatory at his company two years ago. Knowing a programming language stopped counting as an advantage, he says: “We said the skill is not a differentiator anymore”, and every employee became a manager with one AI engineer reporting to them.

What are the four core capabilities of AI, according to Malik?

AI has four core capabilities, in Malik's framing: it recognises patterns, makes decisions, processes information and keeps learning. He tells businesses to map those against their own work rather than copying what neighbours do, because “Every business has a unique DNA”.

Why does Ajay Malik call AI a seventh sense?

If Google search is a sixth sense, AI is a seventh, Malik argues, because it interprets information rather than just finding it. He quotes a former boss: “I do not need the dictionary. I need the poem.”

Should firms build knowledge search before customer support bots?

Enterprise knowledge search should come before customer support bots, Malik says. Enterprise knowledge search means an AI connected to a company's own documents, email and databases that answers staff questions, which he says removes dependence on the one person who knows; his company, he says, has deployed over 5,000 models.

How do new AI tools increase a company's security risk?

Every new AI tool widens a company's attack surface, the set of points an attacker can target. Malik says adopting 20 AI tools exposes a business 20 times over, and advises keeping data on premises where possible or checking a vendor's privacy and ethics policies.

Where Mark and Malik disagree

Mark calls Malik the most pro-AI business guest the show has had and warns of collateral damage, citing David Bianchi's warning that “there will be casualties”. Malik maintains that AI is not necessarily about job loss, but concedes that if your job is only boring and repetitive, “you should be worried”.

Questions

What happens when companies make AI mandatory?

Ajay Malik made AI use mandatory for his own team two years ago. He says businesses should start by handing AI repetitive, boring and tedious work, with enterprise knowledge search first.

Who is Ajay Malik?

Ajay Malik, founder and CEO of StudioX, made AI use mandatory for his own team two years ago.

What did Shopify's AI memo ask teams to do?

The episode opens on Shopify's AI mandate, which asks teams to prove AI cannot do a job before hiring.

What advice does Ajay Malik give businesses starting with AI?

He tells the Thinking On Paper podcast that businesses should start with AI by handing it repetitive, boring and tedious work, with enterprise knowledge search first. His advice: learn how to ask.

Transcript

Jeremy Gilbertson disruptors and curious minds. Welcome to another episode of Thinking on Paper, where we dig deep and unpack technology and what it means for us as humans. We keep humans in the mix. That's what drives us. We're powered by curiosity. We hope you are too. Today we're talking about AI. I want to again, but I want to set the tone so it's not just AI because I even get bored with all of these AI headlines. And the one that I'm excited about to tease up as we move into this

The Shopify memo

Jeremy Gilbertson is Shopify, one of the largest, most successful platform style organizations out there has issued an AI mandate that basically says, Mark, if you want headcount for your department, you have to prove that AI can't already do it. So if Shopify is doing this, I guarantee there are tons of companies that look to companies like Shopify to help point their new North Star as they move forward into this new world of technology. Today, I think this question

This this headline could actually be a lead-in to our guest and what we're talking about to get to some tangible Stuff mark. We like to bring it down for folks here. So who are we talking to? What are we diving into? Let's get going

Mark Fielding Well, we couldn't get any more tangible. As I said in the green room, my goal of today's show is that I want at least one company that isn't using AI to be using AI thanks to listening to this show. I was in Paris last week. There was a very interesting talk I was listening to by the head of the Cardano Foundation. And he was painting a picture of trillion dollar companies, not too long in the future, that might have 10 employees. We're not there. We're here.

And today we're talking to Ajay Malik, founder CEO of Studio XAI. He's a technologist, he's a business futurist, he's an inventor. He is very happy and very gifted and blessed as he said, and he's going to be outlining the strategies that companies can use today to start using AI. Welcome to the show, Ajay. It's a pleasure to have you here. Thank you for thinking on paper with us.

Ajay Malik Thank you. Thank you for having me. Very excited to be here.

Mark Fielding Yeah, we're excited to be here and I will let Jeremy continue his Spotify. What do call it? A memorandum? What would you call it?

Jeremy Gilbertson It's being phrased as a mandate. I don't know if this is like leaked manifesto or all of these words that people use to describe strategy. see a company like Spotify coming out and saying, hey, we're not adding any new headcount unless you can prove that an AI can't already do that thing. That's a really interesting trend to think about. as these companies start to explore this stuff. So listeners, again, what we want you to gain from this is a comprehensive

Jeremy Gilbertson real-world understanding of how to apply AI. is challenging you after this episode to test it perhaps through some of the stuff that Ajay is going to talk about. So Ajay, before we get going, tell us about your personal aha moment with AI that you had and how long ago was it and what drove you to do what you're doing today.

Ajay's first aha moment, 1989

Ajay Malik Wow. So my first aha moment was a long time back. OK, that was 1989. That was my first year, final year project in my computer science when I did from IIT India. And it was impressive. It was like I worked on building an expert system. OK, but then, of course, as you know, we know, all know, AI almost disappeared from the world. Right. There were people call it AI winter, but AI disappeared.

and but I was really excited because it could reason it could provide me a solution based on rules and based on messages that were pre-built okay and it worked and you know those were like 10 megahertz computers those days okay we are not talking four gig or five gig speeds 10 megahertz

Jeremy Gilbertson 10 megahertz computers.

Mark Fielding The wild early days of 1989. Wow.

Ajay Malik Yes, I am old, very old. those days, know, programs fit in floppy disk. Probably nobody knows floppy disk 360 kilobyte floppy disks. OK. And it had full compiler programs and everything. And now five DVDs are not enough. Anyway, but my second aha moment that happened in Google at my time in Google, I was.

Ajay Malik responsible for engineering and architecture for like nine data centers, okay, the corporate network and

Data center cannot be built in a second. It takes time. It takes months. You know, there are negotiations, cities, fiber. There are so many pieces involved in it. And the network traffic was growing. Okay. Every six months doubling. So how do you stay on top of it? And the process was AI. It was not even called AI. It was just like, Hey, we need prediction. We need to make sure that we can be ready six months ahead to give orders for what I expect my capacity to be in six months from now. That was like, my God, that is cool. And this is how you are always staying above the curve. And that is what every business needs. I left Google. I was so excited about AI. That's why I said it was my biggest aha moment. I started my company. And the reasoning was AI can do wonders for business. Okay, we started in 2018 and of course there was the AI definition has changed. What we used to call AI now people call it classic AI and the current AI which is the generative AI is called the AI. people the terminology has changed a little bit. But AI is all about AI has two parts. One is how it helps your business.

how you can be ready for business, how you do not make wrong investments, how you spend the money right way, not waste. And on the other side, AI is about creating more content, which is what people talk about today.

Jeremy Gilbertson Don't bring up this Studio Ghibli stuff around Mark because we could get into a whole rabbit hole on that.

Mark Fielding Please bring up data centers around Jeremy. I've never heard so much data. Jeremy is from the data center world, AJ.

Jeremy Gilbertson Fair enough. enough. Ajay, there's a, I did a lot of capacity planning for, for years and years. ran a consulting company that did that and, very tricky measure trying to predict, you know, data center growth and that sort of thing. And I may or may not have been on projects in the central part of the U S that were tied to what you were doing related to their fiber plan. So that's all I can really say about that. But I think we were in the mix together on a few things. All right.

Mark Fielding Could I just add one thing to that? If you're listening to this, if you're interested in data centers, check out our episode on data centers in space.

Jeremy Gilbertson Yeah, if you want your mind blown for sure. that stuff is coming. All right. So, Ajay, let's break this down for the listeners that are either working for companies or may have their own companies that aren't really familiar. They've heard AI, they've heard all these talks and stuff. And what are some key areas that they could think about exploring to bring in AI? You hinted on a couple, but like what are some more tangible ones that we could like really break down and say, hey,

The four core capabilities

Ajay Malik All right. Let me just before saying that, say what are the core capabilities of AI so that people can think like how I can use AI, you know, like what AI can do for me. OK. And what AI can do for you. I put it in like four buckets. OK. It can recognize patterns. OK. Then you do something with it, but recognize patterns. It can make decisions. OK.

it can process information. Okay. And it can continuously learn. I think if people put those in four buckets as the capabilities, then you can start thinking in my business, is there a pattern that will help me? Is there a pattern that will help me for predicting? Like I can do my inventory management better. I can do my sales calls better, my leads generation better, or my product development testing better or whatever.

Right? Or can I do decision making? Like how do I make the decision? Right? Like when Amazon comes and says, hey, 35 % of the people who bought this product also bought this. Right? That recommendations. Right? And is there somebody else? Is the or as a business owner or manager or whoever, responsibility of making every decision is too much and being the fear of being wrong. So I having an advisor on the side.

And that's what the role AI can play. Information processing. There is lot of information. I have data in my contracts, data in my documents, data in my databases, my IoT devices, my non IoT devices, my tax returns, my passport. We have information so much. My WhatsApp messages, SMS messages, emails, right? Like it's infinite, right? How do I recall? You know what?

What Google did, and what AI adds

Ajay Malik And I will just take one second, go back to Google now. Google, right? Like we have five senses, right? And we talk about them all the time, right? But I really thought and I really believe Google search is the sixth sense. We never had the capability for finding information. And now that if you don't use it, you are like almost wasting your senses or body or what you can do. You have to use Google search. That's the information.

Ajay Malik Getting the information, AI goes one more level up. That's the seventh sense. It processes, it tells you the meaning. Rather than, searching 400 pages and looking at 5,000 results from Google, no, tell me what is important for me. That's what I want to know. I don't have time or ability or span through the sponsored pages or sponsored links or advertisement or.

fake news links in the search results. I want this. And the last point is continuous learning. I need an advisor who works with who is not static, who is not like, hey, you know, like when cloud AI comes and says, I have my information from 2024 September. Hello, that's not a gold medal. Go ahead. I need it up to date. I need the information now. AI can do that. And I think they all models are working on that.

Mark Fielding That's the powerful one.

Ajay Malik So these are the four core capabilities.

Mark Fielding Can I just ask you a question on what you just said about Google search being the sixth sense?

In a year's time, that sixth sense is going to be obsolete, is it not? Are we witnessing the beginning of the end of search engines as we know them today?

Ajay Malik 7th sense. You will need the Google search but the 7th sense the AI. It's like this, you know, my one of my boss, I have been lucky in my career. Okay, good bosses, very good bosses. I've learned so much. Okay, and one of my boss said long time back, Ajay, I do not need the dictionary. I need the poem. Okay.

Don't bore us, get to the chorus

Jeremy Gilbertson Here's the musical analogy of that. Don't bore us, get to the chorus.

Ajay Malik Wow, don't bore us get to the chorus. I'm gonna use this. It's a good one Yes, I will. Thank you. And that's what the AI does Google is providing you dictionary when you do Google search dictionary You are the decision maker with unassisted decision maker So I will make decision based on only my knowledge my capability my experience Limited okay, but if I use

Jeremy Gilbertson It's all yours,

Ajay Malik chat, GPT or Gemini or cloud or anything they have fed the world data in it. So they are using the world knowledge. The worldy knowledge is helping me make the decision. And think like this, no better advisor than that. That's why sixth sense is a layer below seventh sense. But the seventh sense is built on six. Without the six you don't have seventh. It's a layer.

Jeremy Gilbertson I'll you what, not to nerd out, I'm just gonna sprinkle something in real quick. So we were breaking down a book called Irreducible by Federico Faggin who came up with the first microchip years and years ago and now he's onto consciousness. We're unpacking that along with Annika Harris's Lights On audio documentary that talks about consciousness and that sort of thing. Senses, really interesting thing, right? There's a lot, can humans...

actually are they capable of generating a new sense or is it a tool that you're just learning how to use to use with your other senses? So there's a whole dialogue on that stuff that's going on in my brain now, but listeners, you can unpack it in our chapters.

Mark Fielding like plug in sensors, you actually have additional plug in sensors.

From what you just said, Arjay, and to answer Jeremy's question about where could a business begin with AI, from what you've just said, the CMO, the CFO, the CEO, HR, accounting, doesn't matter, everywhere, anywhere.

Where do you start?

Ajay Malik Anywhere, everywhere. You just have to look at the capabilities and see how you can use those capabilities for your business or your personal life, you know. And there are many applications of this, okay? But even without going into applications, you have to start looking very simple way. Okay? It's not a complex thing. AI is, you know, the beauty of the AI is AI is developed using a lot of tech. However...

AI is solving human problems. You are talking about it like as a business, human, as a person, right? So you think and you are not trying to like, I am going to and write a complex mathematical physics formula and I'm going to achieve that. No, not this. Think like this. What is that which you are doing repetitive?

What is that that you are doing a lot of time? Hey, we make these podcasts and then every time we make podcasts, we make shorts, we write things, we write a page, we write a summary. You know, those are standard things. Anything which is boring and I don't want to do. OK, whatever is that? OK, you know what? Every morning I look at it and I have to scan through all my emails and to see which is right, which is spam and go through my

200 emails that arrived last night and go through and to figure out what is important for me. Okay. It's boring. I need to do it, but there is no way easy way. Right. Or think like what is, man, it is so much work to do. I have to look at this. I have to look at this files in SharePoint. I have to look at this. There is a process. There are so many documents. I have to open those contracts. I am like already tired even thinking about it. Okay. And if you have things like this.

Ajay Malik This is like simplest, whatever is repetitive, whatever is boring, whatever is tedious. And whatever has a big process. You have a lot of moving parts. I need to see this, I need to see this, I need to see that. Whenever you have things like this in these four categories, AI can help. And...

The gap between silicon and carbon

Jeremy Gilbertson So in your buckets that you mentioned, the one that kind of jumped out is this idea of we make so many decisions, right? Some studies say that on a daily basis we make 30,000 decisions and that's one decision was me to make my hands go like this. Some are small and some are big, right? But 30,000 decisions, it's why Steve Jobs wore the same turtleneck every day. There's decision fatigue, but...

silicon and carbon, this little gap between the things that we're talking about right now, is trust. How do we trust the silicon to make our decisions? People are wary about that. People are like, okay well this all sounds great but if I put this thing in and I push this button, can I trust this AI with my career? How do we get around that?

Ajay Malik So you know what? Very good question. Okay? And I say this and you know what? I had so many discussions even in my team. Even we have sometimes, even today, very heated discussions. You know, it's I call arrogance of the human mind. Okay? I am better. I am better than machine. I am better than others. I am better at making decisions. Okay? How the AI became really picked up. Okay? So I think 2000...

14 or 2010 I'm forgetting the year right now. Okay? The CNN. CNN are the models, the AI models which do based on image. They can do image analysis and they can look at it and tell which a dog, it's a cat, it's a red light, it's a green light or whatever it is. Okay? There was a test. There is a test in Germany in which you know what the cars are going on the roads and people are following the rules. Red, green. Okay?

And first time the AI computer vision product did all the, they were driving and computer did better than humans. Humans make mistakes. We think we are better, we are superior, but we make mistakes. And we say, yeah, I made a mistake to erase human and we move on. And, but we are like, AI, AI made a mistake. How did AI make a mistake? AI cannot make a mistake.

So what I'm trying to say is we have to see it like this AI as a partner, as an assistant, and which has proven to do better than humans. OK? And that's how the AI started becoming successful. AI was able to process a lot more data, provide lot more results, and on average, they were better than humans making decisions. So will I be able to make sometimes better than AI? Of course. Will AI do better than me? Of course.

Will I be equal sometimes? Yes. So it's just a partner. Treat it as a partner. So do not...

Should we give AI a margin of error?

Jeremy Gilbertson So were you, Sue, with that, were you saying we should give AI a little bit of grace in that it's not gonna always be right all the time? Just like I would give Mark a little grace when he's not right on something or he would do for me, we're human partners on this podcast. Is that kind of what you're saying?

Ajay Malik Yes, exactly. will be. There will be errors. Errors are on both sides. But now you have a partner and you know you can brainstorm and you can decide and you can go from there. You know when you use Cloud, chat GPT or everything they say, Cloud can be wrong. Chat GPT can be wrong. It can make errors. And we have to just treat it like AI is AI, not super intelligence. That's the mistake.

Mark Fielding I think one of the reasons is, just on that is humans like to have somebody or something to blame, to direct the frustration, to direct the anger. And if a human makes a mistake, it's very easy to find that outlet. If the AI makes a mistake, it's not so easy to have an outlet at all. So just from what you were saying, your message to, so I'm from England and in England,

Mark Fielding Jeremy, you might not know this, but we've got a lot of old companies in England, hundreds of years old, they have a lot of data going back literally centuries of information that they keep on paper ledgers in the back office. And your message to them is that today in April, 2025, AI is precise enough, accurate enough and good enough that they can trust it now with their data.

Human in the loop

Ajay Malik Yes, and be the human in the loop. You know what? That's it. And that human in the loop, keep it. Okay? And the human in the loop will become less and less maybe, but it is still important. It will be important. But we have a partner now and we have to use it because it will help us. You know, people say this, hallucination. Very common word. Everybody talks about hallucinations. All right? What is a hallucination? Creativity.

and creativity gone wrong, more creative. If you are not creative, I can tell you, you can definitely never hallucinate. Hallucinate is just next level of creativity and sometimes it goes wrong. Right? It's like I start making a painting and I paint and sometimes painting goes wrong and really wrong. Okay? The creativity is needed. Humans, that creativity is there now. Use it. Okay? And it will make mistake.

Ajay Malik It will go further. It will be extra creative sometimes. And you have to stop it, control it. It's like, you know, you do it with anybody.

Jeremy Gilbertson This is that's that's a really cool point. I've never heard anyone talk about hallucinations as happy accidents. I think that that's a really interesting thing. Interesting thing to think about. So let's point back. So maybe if if if you know the CEO of Shopify or someone else can introduce us to to him, one question might be, OK, so imagine you're the CEO of Shopify. Let's let's do a little let's do a little play here. So

Ajay Malik I can take one day's

Jeremy Gilbertson That's right. Right. Right. So, so I presented, I was like, Hey, I need some headcount. Ajay, you say, Hey, prove to me that AI can't do it. I do some research. come back and like, Hey Ajay, actually it looks like AI can do it. I don't need headcount. I go back, start doing my work. I get AI to do some of that work. It causes a massive error. It leads to an hour and a half of network downtime at Shopify. You call me, you're like, dude.

What happened? Why did you bust the system? Like, can't believe you did this. Do you know the risk, the money, all that stuff? I'm like, yo, the AI did it. Like, you told me to let the AI do it. There it is. I'm sorry, man. Yo, yell at the AI. Like, how do we manage that piece?

What the Shopify CEO probably meant

Ajay Malik Alright, I do not know what exactly he said, the Shopify CEO, okay? But I tell you, if... But I would say, okay? It's not about that, do... Replace yourself or replace the work by AI doing it. It is use AI to remove the repetitive, remove the boring, remove the tedious, so that I can get velocity, higher velocity, higher productivity, higher quality.

There are the benefits. want what the mandate is about. Improve your productivity, efficiency, velocity, quality, reduce your costs, do all those things. Use AI in your job. Okay. And once you are using AI, I bet you I work with companies. I have worked with many companies. know, AI is always about job loss. Everybody starts thinking about job loss. AI is not necessarily about job loss.

You know what? This company started using AI for code management, for sales, and for other things. They have more work. They had to hire more people to support the business increase because of using AI. So AI is not necessarily about job loss. AI is about improving and increasing what you were doing. You see? the way... So coming back to the original point, what can businesses do with AI? Right? So we talked about capabilities.

I business should, well this is how I look at things, but everybody, you know, for a business, I look at it, business is all about focusing on the top line or the bottom line. Okay? How do I do, like, can I add new line items in my purchase order, right? In invoices, right? Or can I justify my premium pricing? Can I increase my prices? That's the key, as a top line, right? Or can I, or bottom line is all about, can I...

make my operations smarter, leaner, more predictable, right? More scalable. Can I do more? Right? With whatever I have. Right? So do better. So top line or bottom line. And I think business should look at exactly like this. Don't jump into it. What neighbors are doing it or what you are hearing. Others are doing it. It may or may not be even useful for you. Look at your business. You know your business better. You know what? Like humans have DNA.

Every business has a unique DNA

Ajay Malik Every business has a unique DNA. How it works, how it operates, what it wants to achieve, where it wants to grow. Everything is very unique about every business. Look at it like these are the capabilities we discussed and what will help me, right? Do I want personalized experience for my customers? Do I want to add a co-pilot inside my products? Do I want predictive alerts? Do I want some coach inside my product?

training. Think like this. Think like what will help your top line. Just think like this. Do not try to think there is no mathematics to think. There is no vector maths, no calculus, nothing. Just think what are the key things that will I would say, you know what I say, but I don't think it makes sense to lot of people. But you know, just close your eyes and think what will be the ultimate thing that you can offer to your customers, products or whatever. Just think and do it like that.

And you know what? Even if some other people are doing it, learn from them, copy from them or even better, I tell you this. Go today to chatgpt or cloud and ask. Write down, write down two paragraphs about your business, what you have done so far, either your job. This is what I do. We do this, this, this. is what we do. This is what I work for. Just write down what exactly it is and say this. Give me.

ideas, AI ideas that I can do. Give me tangible ideas. I do not need fluff. I do not need ideas which will take years to complete. Put a time scale in immediate three months benefit, six months benefit. Process ideas. Ask. will list them. I'm telling you, there will be some ideas. You will love them. Everybody will love them.

Jeremy Gilbertson So keeping humans in the mix, the art of asking... hold on, hold on. So keeping humans in the hunt, the art of asking a great question will still be a superpower. All right, so...

Mark Fielding In the hunt, we're going do the hunt, the hunt, Jere, the hunt.

Mark Fielding Just on your point there, Jeremy, one of the things that personally I've been looking into is what RJ has just explained, probably for a lot of businesses requires a third party to come in and help them to ask the right questions, to use the right, to prompt engineer, to build the picture, to use the AI, to understand the back and forth and the question structures to ask them.

Ajay Malik Yes, it is. Yes, and you know what? It is we do also, as a business also we do that. You know, we call them discovery calls or we help them discover, right? And have meetings with the people, have the business owners, department owners and see what they can do. And we have experience, we have seen what other businesses have done. Like my company, we have deployed over 5,000 models. Okay, so we have a lot of applications out there. So we have ideas like what people do, similar business, similar space. AI can help you also, but everybody needs it.

Ajay Malik And I'm saying it, you can use a human advisor, you can use a company like me, or there's so many companies out there. Everybody's doing AI. Okay? Or you can go to AI itself and ask, what can I do? And talk, have a general discussion. You know what? It is fun. It is fun to engage, you know? I sometimes worry that, we start talking to AI so much? I am emotionally attached to my AI when I'm working with them. And I am like...

Sometimes saying the bad words, sometimes saying the good words, sometimes saying thank you. And people are like, wow, Ajay, you are very polite today to the AI. And it is like different.

Jeremy Gilbertson So if you're listening to this and you haven't listened to our episode with Kevin Kelly, it's the one right before this one, the emotional aspect of AI and communicating with AI. We dive deep into that if you want to check that out. let's, Ajay, let's, so you've stated a pretty good case for businesses should start to play around with this stuff, right? So can we start from a StudioX perspective, this idea of no code,

AI agent, can we start there as a jump off point and like, is that? What could it actually do step by step from from creation to using it on a daily basis?

How an AI agent actually gets built

Ajay Malik Alright, I will go over this. It's not a sales pitch. It's okay. You know what? There are many people offering products and services. We are one of them. We think we are special. But you know, everybody has to decide and make your own choices. But I'll tell you how StudioX helps and what you should be looking for in any other product when you selecting. In the StudioX world, our whole promise is you can use a cloud-based product.

or you can install it on your premise. You know what, I'm worried. I don't want to put all my books, all my everything, all my personal contracts in cloud. I'm kind of worried about that concept. Okay, so you can run everything on premise, so you can decide first thing. Hey, I'm using a cloud version or I'm using a non-cloud. Even though we are encrypting data at rest, data at motion, we are doing all the right security things, state of the art stuff, but peace of mind and.

Sometimes business requirements. If you are a government or defense contractor, you do not want your data in cloud. You are not allowed to. Okay? So things like that. Then second step, start with a core of your business. Core of a business is knowledge. All my data, all my documents, all my contracts, all my system. Where is all the information? Like for this podcast, like for your team, right? All the thousands of podcasts so far, right? And all the information about them.

Just imagine if you put that, hey, yes, thinking on paper, very nice, right? And you have this. And just imagine you say StudioX, I install it, whether in cloud, I connect it to my data source. That's all you do, connect. Within five minutes, you have connected your data source. You have put all this is my confluence, this is my SharePoint, this is my uploaded documents, this is my w colon drive, c colon drive, whatever. You put everything. Knowledge is there.

Ajay Malik Once the knowledge is there, your AI bot is ready. You can talk to it and it will answer anything. You can say, hey, which episode we had a discussion about emotions with AI, or any other topic you can ask or give me somebody of those topics or give me 10 podcasts which were similar or anything. You can find all the information. That is the step one. Able connecting your knowledge and ability to find information. Step two, more AI agents around it, rules. What can I do?

You know what? let me connect my email. OK. And then what will happen is every time I send an email, every time an email comes to Jeremy at Think on paper, thinking on paper dot com, what happens? It will check in the I will send an email to Jeremy myself and the email I will say, hey, can you every morning look at all the emails arrived since yesterday and just make a summary of the top three things I need to do or five emails I need to respond and just text me?

at this number. Done. Suddenly, so basically start using StudioX, you add the knowledge. Knowledge can be in your emails also. Email is your running knowledge all the time or tickets, support tickets. You can create rules and you can create simple rules by using human language, specifying it and suddenly AI becomes part of your workflow. Now you are offloading your task and start offloading your task, offloading your task, whether it's emails, whether it's sales or hey,

I got a bill of material and I need to look at all the products and see what is the price today. Check it all the web on this. So you know using, so that's what StudioX enables. You do natural language conversations using the channels you already know, email, WhatsApp, SMS, on your knowledge and connect it to your live knowledge, whether it's email or it's whether your ERP systems, the customer database systems, for everything you connected.

suddenly AI becomes part of your workflow, you start creating workflows. That's what you will do within a week. I think you will have a lot of AI elements in your business.

Within a week

Ajay Malik Within a week. Yes, that's the thing. Yes. The key is, you know what? AI is... A lot of work has been done by Sam Altman and all the people, smart people in the world. We don't have to invent the AI. We will use it. And you know what? you, it's like this. Microsoft Excel. You know, like people move their books, right? Everything was on paper. Thinking on paper. Everybody was writing everything on paper. Right? And they move to Excel. AI is just new software.

use it as a tool. You don't have to create Microsoft Excel. We don't have to create visual basic programming language. We don't have to do those. We just use the Excel and we can do wonders, right? Same thing with the AI now. AI is the next level. AI is the new software.

Managing the intern

Jeremy Gilbertson So that's a great explanation using us as an example and you've actually sparked some ideas for me already so thank you for that. Alright, so if we're using, if we're spinning up these agents to offload some of the remedial tasks, the repetitive tasks, the crunching tasks kind of stuff and we have them set to do that and send us outputs that we can now as Mark, as Paki McCormack in Most Human Wins talks about us moving to a higher level of abstraction using these new tools and new information to build new things. How do you manage an AI agent? do you, is it, how do you, yeah, how do you manage? Like you manage a person. Like if we had an intern, if we brought an intern into Thinking On Paper to do all of that stuff for us.

We have to manage that intern, set expectations, we have to check in, we have to make sure it's doing the right thing. What are some keys to managing AI agents?

Ajay Malik I love this question. I will add one more thing before this. And with you said this, like hiring an intern, two years back I told my team, we made a company policy. Like you say, mandate. We said the skill is not a differentiator anymore. That I know C, I know Java, I know programming. No, use AI for that. That is not.

Okay, so don't come to me and say that, I do not know how to do that. No, you know, you know how to do that. You know this question that I do not know how to do that is not a valid question. It's not a statement. You have to just do it. You have to just go to AI and figure out five minutes and you will know the answers. Okay. And the second part, I said everybody in the company from today.

is a manager. You have one engineer reporting to you and that's AI. Everybody has it. So you are not alone. Okay, so you are not just working by yourself. You have an AI engineer, you have an AI working for you and you can get a lot of things done. The beauty is this person doesn't talk back. The person works 24 by seven, doesn't have attitude, does not have get stressed and does not it always helping you. And you know what? Anytime you are like not satisfied, you just say new chat and move on and it starts fresh. Okay. So that's the thing. And then the third thing which you are asking and that comes naturally. Then how do I know? How do I manage this? Is my employee, the AI working well for me? Is it like doing what I want to do and how do? So the good part is you do not have to do when you are managing AI, you do not have to follow the standard processes or like California policies for performance improvement plans and HR reviews and anything, you are the reviewer, okay? And use it and if you are not satisfied, learn how to talk to it. That's the difference and that's the biggest thing in the world, okay? And like I think you hinted it earlier, there is a concept of prompting, how to talk to it.

Jeremy Gilbertson HR reviews, yeah.

A seven year old with the world's information

Ajay Malik Okay, AI is like a seven year old child with the world information on its fingertips and ready to please and very creative. Okay, so it will leave no stone unturned to give you an answer at high speed with the knowledge it has. But if your question is not right, you will get an answer. So.

You know, a lot of religions say one thing, sorry to a lot of religions say this. OK, do not look for bad in others. Look inside you. You know, do not look for God outside. Look inside you. And, you know, concepts like this. Right. Here is how I see AI.

Jeremy Gilbertson But what do you do when that seven year old starts running down the hallway screaming with scissors in their hand?

Mark Fielding Well, it doesn't does it because you program that all in. fact, it the the HR yearly review is done by itself. so because it's continually self learning, you don't actually have to worry with the seven year old running down the corridor with the scissors because it's all been programmed to not do that, to to iterate, to compound its knowledge, to compound his experience, to compound what it's doing correct. And in fact, yeah, it's it's it's its own HR.

Ajay Malik and you use it, this is the point and you change it and start a new chat if you don't like the screaming. but. No, no, no. Sorry, let me just say one thing. Okay, so you are responsible for asking the right question, giving the right directive. That's what I was trying to say. So you know what? It's not like I see bad in the other people. It's because I am the bad.

Jeremy Gilbertson If you don't... Well, what if... All right, so what if you... Yeah, go ahead.

Ajay Malik Okay, don't blame the AI. Don't start looking at it. Learn how to use it, to instruct it. Well, it will learn with you. It will grow with you. But you have to learn how to ask. And you know what I'm telling you, the better you ask, the better answer. Sorry, go ahead.

Jeremy Gilbertson No, that's a really good point. keeping humans, what Mark? Keeping humans in the hunt. you could actually prompt an AI in a nefarious direction, obviously, right? And if you do that, the responsibility is on you. But the coupling and tracking, right? I'm just thinking about like, all right, we're society, we're a microcosm. It's me.

Jeremy Gilbertson It's Mark, it's you and know, you and I, Ajay, we use our AI and it's great. You know, we do AI performance reviews and they, it's awesome. We do it for the right thing is in Mark's like, hey, as you grow up AI, I want you to rebel against me a little bit at 5 % a year. And the AI would in turn do that, right? Or find ways to do that. So Mark could be down the road cultivating this AI.

that is aiming to eventually kind of supersede you in a way. So it is all about.

Mark Fielding Why would you want to do that though? That's one of those things with like, it sounds really nice in hindsight. It's really romantic. It's like, yeah, we should all grow up and rebel. Wouldn't it be great? And if that's your kids, okay. But if it's an algorithm, you don't actually want it to rebel. You wouldn't program that self-destructive rebellious streak into it.

Ajay Malik And but it depends as Jeremy is saying, it depends what you want to get out of it, right? If you really wanted it to rebel, it's okay. You know, but my top engineer last year, he said one thing and that was so, you know, he's so good. Okay. He's like 10,000 times better than me. Right. So I am like sometimes I'm, I am also very passionate and I'm arrogant and all the humans, everything was the normal human. Okay. And, but

I question, right? Like, hey, and I want something, want, and I go after it. And then he said, Ajay, it's OK to challenge me. You don't have to stop. It's OK. Go for it. Right. And that is what it is. But when I see when I hear when Jeremy saying it's OK, let me I challenge you. Go for it. Because you know what? Maybe something else will come out which we were not thinking. And if nothing else, you know what? New chat and move on. OK, so you will be.

But we're talking more like AI like this, but just coming back to the business, the applications. So we talked about you can do applications for top line or bottom line. Top line applications are more around like, hey, how do I do personalized experience, voice control, predictive alerts, coach built-in product, co-pilot built-in product. Or the bottom line applications could be real-time monitoring in your manufacturing.

You know what? Everybody, well I should never say the word everybody. Lot of people, companies have their cameras, know, monitoring their factory. But just imagine those cameras could monitor more. That images can be fed to a software, that software can do something more without any having investment or something and it can look if the machine is down, if a person is hurt, if a person is in an area they should not be, or if a person is lifting a big glass wrongly or something like that. So you know what? AI can help in those operational things.

start thinking like how AI or computer vision will help you with quality control, how it will help with the safety in the business, how it will help with the or maybe it will help with your customer support agent. You know people don't want to talk to bots. I understand people do not want to like and you know what and everybody is very hot on talking like voice bots. Young people talk with fingers they don't talk with mouth most of the time right so you want to

Jeremy Gilbertson It drives me crazy.

Ajay Malik So I don't think they are looking for a bot to talk to. They just want chat and text and things like that. Right. So if you are looking at that, so agent assist, like I talked to a customer support agent, but the customer support agent now uses the bot to help. So not the end user. So you can help your customers in so many ways, unique ways, right. Enabling your customer support people to be better, enabling your troubleshooting, your service team to do better enabling your sales team or inventory team for demand forecasting to do better things like this so there are many applications start thinking in those and you know what I will add the sales plugin always feel free to schedule a meeting with me and I will help you okay or go to anybody okay there are so many people like me out there just go to them ask them hey how do I do this is my problem and then find the best fit for you

Jeremy Gilbertson Okay, so they get excited. They're like, all right, I believe in testing some of this stuff. And so they go back to their boss and like, yo, I found this thing. I think it's great. The boss is like, yeah, that sounds kind of cool. You what do we have to do? Well, you all we got to do is you put this little agent on our, on our computer and we got to feed it with some day. the IT department goes, yo, hold on. We're not putting anything on our computer and we're not sending our data anywhere. So how, how do you navigate like internal IT departments?

And because that's probably the initial pushback, right? Is some of that.

Cloud, and the security question

Ajay Malik 100 % and this is why I said if the company is not

Even if they are cloud friendly, AI is a bigger challenge. Okay? Because AI is using your knowledge and that you know what? And it's not about the company. It's not about your vendor. Okay? Like when I am the vendor to my customers, I have all the best intentions. Why will I try to mess it up? Right? And so there are all the small companies, big companies, even the big companies, know, OpenAI, they have a cloud and everything. They have the best intentions in the world. Right? But there is a $400 billion industry.

That's called the cyber hacking industry. These are the individuals or teams sitting in some room somewhere around the world and their whole job is to hack the hell out of the system. Find that information. Find my phone number, your phone number, last purchase or your credit card or your social security. So many things they want to find out. Those people will go after the cloud and if the AI...

You are putting information and that cloud is hacked. Your information is hacked, very private information. So IT is not wrong when they are like, oh, I don't want in that direction. OK? That's why I said earlier also, and we do it in StudioX also, think on premise. Try to put like 20 AI tools. 20, oh, we are using AI. We are using AI for recording meeting. We are using AI for video. We are using AI. Now you have 20 tools you are using. You have exposed your attack surface 20 times.

There are 20 places you are connected. You have to be careful. And that is the thing. And then there are, you have to also focus on the ethics of the AI company. Are they using your data to build models and learn and become better and sell those models? Because at the end, that is a very risky proposition also. So you have to think those two things. So I always say, try to find a solution in which your data doesn't leave your building. Try to see that. If it's not possible,

Ajay Malik Make sure the company you are working has strong privacy and ethics policies.

You need that. will not tell IT, no, it's okay to let it go. It's AI.

Jeremy Gilbertson These are some of the most succinct and tangible takeaways I think that I've heard on implementing AI into business. think Mark, your point in having you, the listener, test some of this stuff out, I think there's some great nuggets that you could pull out as an action roadmap if you're interested in this. I think you've been spot on, man.

The eighty twenty principle

Mark Fielding Well, let's use the 80-20 principle. Okay. For listeners, for businesses, which are listening to this, they haven't started yet with AI or they've they've used chat GPT. What is going to be the biggest bang for their book across verticals? So across marketing, across finance, it doesn't matter manufacturing, whatever it is, what would be kind of the one or two AI tools that they should introduce first just to get the.

that maximum impact as quickly as they can.

Jeremy Gilbertson have a guess, I have a guess on this one. What do you think about customer support?

Ajay Malik Yeah, support is amazing. Although before that, I will add knowledge here. Enterprise knowledge search, okay? Business knowledge search. Knowledge is so crucial, fundamental to your business. It is needed when you are sick for somebody else to take action. It is needed for you to take better actions. It is needed for decision making. It is needed for your better inventory, better production, better demand, better sales. Your knowledge, you have knowledge in two types. The...

text knowledge or image knowledge, you know, and the other is the number knowledge, which is like the numbers, the sales results, the inventory and product, all those things. To those things, I think I would say first go for knowledge, enable knowledge, enterprise knowledge search for your business so that you do, you have the answers. You are no more looking for, hey, only Bob knows how to get me a chart on this or only Bob knows this data. Get that information. Okay. That is the first part. Then.

all these things come immediately. A customer support, agent assist for your customer support, support for your service team for troubleshooting, support for your operations to do better inventory management, prediction, all those things will come after that. I think first step would be knowledge. I think that's the most important application because that makes your company, it's like this.

All your company is inside with one AI agent which is there for future. Even 100 years from now it will have that knowledge. So you have just suddenly transformed all your company knowledge and removed the dependency on one or two individuals or 20 or 200 or some computers into a thing which you can rely on for future forever.

Jeremy Gilbertson Here's really interesting, and that's great, knowledge and implementing training programs, right, for new employees, onboarding programs for new employees, things that new employees can go to and be like, hey, has anyone else ever dealt with this situation? And that's really cool, but here's an interesting thing, know, here's the human side of this, you know, so what happens to Jim when he gets hit by the bus? know, Jim knows all of this stuff, he's got 30 years worth of experience, and now you're like, yo, Jim, can you spit out everything?

into this repository so we can, you know, we can use that experience. And then Jim starts to go, wait a minute, I'm actually getting pretty old right now. My stuff is valuable. My information is valuable. Should I just freely put it into this model so they could extend Jim and when Jim retires or when, hey, we don't need Jim anymore because Jim put all his knowledge in here. How is that going to, I don't, how are people going to navigate that as owners and also as knowledge holders?

Ajay Malik It's a balance. It's a balance. You will be surprised how many people actually want to offload their knowledge so that they are not the bottleneck. I want to like tell everything so that nobody is always bothering me and I can do what I want to do. Okay. Even the senior people or sometimes people really genuinely care and hear man I will be retiring in one year or two year. I need to make sure the knowledge is there somewhere so the business can proceed without me. Okay. And sometimes it's the mandate from the CEO.

thou shall do it. Okay? And you do it. And you know what? Fear of job loss is natural. Okay? But job loss will happen for other reasons. Okay? Job loss will happen for economy going tanking. Job loss will happen for so many things. I think by adding the knowledge, enabling other businesses to use your knowledge, other people to use your knowledge and move fast, know, 27 % of sales deals.

do not happen because there was a slowdown in the information. Either the answer did not make it at the right time or something like that. Just imagine that each business can be now 27 % faster. And that will help you ultimately. Tell me one thing. If you are enabling the business and business is growing, will you be fired? No. Company needs you more and company needs more people and it's not going to be easy.

One point which we keep missing and we should not, human in the loop. You are that 15 % that uses the AI and knows if the AI is doing it right and is always working better. So, and you are improving, you have something else. I don't think people should be afraid of it, right? But if you're doing really, if your job is really, really boring, repetitive, and that's all you do, you should be worried.

Mark Fielding That's a lot of people. That's my thing in this, like I refer to it now as almost collateral damage that there will be. We had David Bianchi on a few weeks ago, Hollywood actor. He was saying AI is not fair. The distribution of AI will not be fair. There will be casualties. You are incredibly optimistic, RJ. You're incredibly hopeful. I don't think we've had anyone on the show who is as pro AI in the business world for you. So.

What should parents tell their children?

Mark Fielding Just to end this, I wanna ask a question about the future workforce. I wanna ask a question about my kids, listeners kids, and my daughter is nine, my son is six. Obviously, in 15, 20 years time when they are in the workforce, things are gonna be different. What would you suggest parents try to do to orientate their children to?

Mark Fielding them at this time.

Mark Fielding focus their children's attention to where should they be thinking about education? Where should they be thinking about things which are obsolete in the current education system? How should we be thinking as parents who are, who are, I know you said we shouldn't be worried, but we are, so what should we be thinking?

Ajay Malik Alright, so I will start like this. Okay, like you know, as a parent, if you think your child can see better, you get the eyeglasses. Okay? And that's how your child performs better, because the child needs eyeglasses. Okay? Or sometimes you think, no, you know what? There are these, what are those, you know, those blue light glasses, they will make you do better with computers and things. You don't really need them, but if you have them, they are better. Okay? Or things like that. Parents do that.

AI is like that. Treat AI like a tool to help your children and enable your children to use AI. Do not be the one like, don't use AI. No, no, no, no, no. OK, let them use AI. Use AI and become proficient and efficient in using AI. OK, because you will be the master of that AI. You have to become the master of that assistant. How to use it, how to get more done from that. OK, that is one thing. Expose your children with AI so that they can use it. That is first thing I would say.

Second thing is that human in the loop, that 15 % is about problem solving. That human in the loop is about the human connection and problem solving in context of human relationships, human context. So you have to really make sure your children are more focused on that problem solving. That the next layer, which is will be always different and as a human, we will always keep doing it.

Keep doing it. Even if AI does it, you know, I do it all the time. Okay, my personal productivity with using AI 25 times. Okay. And I believe when you said Jeremy that there will be trillion dollar companies with very less people. It will happen. Okay. But the key thing is there is still, you know what, sometimes I'm sitting with multiple AI agents and going into different questions and same questions and that problem solving, that is the skill. Please focus your children to learn that. That is the most important thing. Like just them learning the Pythagoras theorem. Okay? It's okay. It's not, it's all like, you know, the sine, cosine, the tangent, the theta, the trigonometric formulas and all those things. They're okay. And I think education, education is important. Whether you do traditional systems, college systems, school systems, work from home systems, study from home system, whatever, doesn't matter. But education is important. And I will say education of AI and how to use AI important, use it.

Ajay Malik That will be important skill and that monitoring that if the AI doing right, there will be tons of jobs in that, trust me. That monitoring AI jobs are going to increase tremendously.

Jeremy Gilbertson Ajay, I couldn't agree more on a lot of this. I think about this a lot and you know, the next iteration of the human condition in a way. And I think tenacity and grit are going to be important because like you said, you challenge your employees, your organization, you have the ability to understand and know everything. So go figure it out and go find it in all of that. So moving through the beginner's mind. Mark, I know there was a lot of chatter on LinkedIn this week about us being big proponents of the beginner's mind. It's not comfortable to be in that spot. It's not comfortable to think on paper sometimes, but I think that is a tool that that is a capability that we can cultivate today that seems counterproductive, right? It seems like, technology is just going to do all these things for us, but we need to cultivate these very human capabilities to make sure we can use technology to our biggest advantage, think. Picking up what you're putting down, man, that was great.

Ajay Malik Alright, very nice. No, amazing talking to you guys, really. It was fun. I enjoy talking about this and I enjoy talking in general.

Mark Fielding Thank you. Well, we enjoyed talking to you. We have to have you back to speak about your, some of your, your inventions and some of your exploits, which sound fascinating, but we really wanted to keep it kind of on track today, but I'm sure. What's your favorite invention? I'm just going to ask RJ, what have you made?

Ajay Malik It's so many, okay. Let's not go there. But I just enjoy what other people make, okay. And I enjoy what the products. I am a geek. I love geekiness in everything. And I like what other people create also. My favorite word in life is create. And I live with that and I enjoy when I see a lot of creativity around.

Mark Fielding Let's end there. I think all of our favorite words is creation of one type or another. brilliant way to end.

Jeremy Gilbertson We'll disrupt.

Jeremy Gilbertson Yeah, and disruptors and curious minds, if you're getting to the end of this episode and you've listened to it, I need you to forward it to somebody that really wants to understand AI because the nuggets that were dropped by this gentleman here today, the tangible action plan to try and test and learn and get better is in the middle of this episode. So please send it to someone who needs it. You can find us always thinking on paper.xyz.

We'll share a bunch of information about Ajay and StudioX and some of the things he's doing. We have a book club we're reading, Irreducible by Federico Faggin, about consciousness, AI, the difference between carbon and silicon. We're in the middle of all these different worlds, but like, subscribe, hang out with us. Be curious, stay disruptive.