Transcription
Hi, I'm Gaia. I'm Angelica. Today we're speaking to Amelia, co-founder and CEO of AI platform IV. So, please tell us what IV is.
Thank you for having me. It's great to be here. Um IV is an AI upskilling and consultancy company. So, we do two things. Well, we train teams on AI and I'm very self-aware that that's like the vaguest sentence of the century. Um but it looks really different from company to company, but essentially we help companies transition in that move to becoming an AI-native workforce and often that means training their teams who they probably hired before AI existed uh to utilize these tools um so that they can get the most out of their AI subscriptions. And then the second part of the business is we embed consultants into organizations to help them on that transformation journey. So, that can be anything from strategy, governance, policy, all that fun stuff, data, cybersecurity, and then actually building out some of these really cool AI automations if the if the company doesn't have the capability or technical capacity to do so.
Cool. Amazing. I think it'd be great if we started with you as a founder and how you got to where you are, what your career journeys look like. So, can you tell us a bit about that?
Yeah, I to be honest had no idea I wanted to be a founder throughout school, university, and even afterwards. So, I went to Cambridge. Um I did academic research there. I'm currently a fellow there now actually looking at skills prediction in the labor market. And then fell into banking. And I say fell because I studied psychology when I was at Cambridge. Uh Goldman Sachs do a dinner for the kind of elite athletes at the university.
Were you an elite athlete?
You wouldn't know I I was. I Yeah, I played a couple of sports at uni and my friend was organizing the dinner. She said, "Yeah, please come. That you know, someone's dropped out." I said, "Well, I don't do economics or banking so or banking. I don't do economics or maths. So, it's you know, a career in banking is not for me. So, I came to be a bit of a seat filler and ended up sitting next to two great people that worked at Goldman and they were like, "No, you should absolutely apply for the internship." And then
Wow.
suddenly I had a place in the internship and then had a job in foreign exchange sales and trading. So, I was there for 3 years to start my career and it was from there that I then jumped into the world of startups.
So, AI wasn't anywhere in the picture before this.
It wasn't. It really wasn't, no. And I'm the non-technical half of our co-funding duo. So, my co-founder, who's my sister, is our CTO um and I'm [clears throat] CEO. So, to be honest, you know, never had an interest in tech, which is an interesting background.
But, I think having a background in psychology is actually very interesting pairing with AI in terms of how humans approach machines.
100%.
think about that? Do you have an opinion?
I I actually will die on the hill that psychology is one of the best things you can study to go into any industry. Just understanding human decision-making, whether you're building a product or selling a product, is so important. Every everyone is a salesperson in their life, in their career.
Yeah.
And life's a negotiation and knowing how the human mind thinks, I think is so valuable.
So, your sister is the more technical side. Are you more the sales side then?
Kind of everything else. Fundraising, sales, marketing, growth, all of those all of those levers.
So, let's go back to the moment let's I assume where you were at home with your sister and you were like, "There's a gap here. We need to do this." What happens?
So, when I left banking, I worked at a startup cuz I knew I wanted to be a founder at that point. I loved the idea of entrepreneurism. I'm like highly competitive and something that I didn't feel like I was getting enough of in banking was that agency to kind of do what I wanted to do. So, I knew I wanted to become a founder, but I wanted to work at a startup first and just absorb, learn the ropes, see all the aspects of how an early stage company works. So, I was doing that at a supplement startup called Zooki in London.
Ooh, I have heard of them.
We love Zooki.
them advertised everywhere, all over the buses, everywhere. I love their recent campaign.
Yeah, they're killing it. They're killing it. Um so, I was at Zooki and at this point my sister was in VC at Deloitte Ventures like doing everything from, you know, the technical due diligence. And at that point AI was just starting to hit the mainstream cuz OpenAI had just launched GPT 3.5. And so, at that point we we said, "Right, we you know, we want to build something together." The original IV business actually looked quite different from what it is today. We launched specifically as a recruitment tech business and product, specifically looking at candidates coming back into the workforce after career breaks. And we've since expanded into AI training. Um but it was quite interesting watching in our first year, cuz we're about 2 years old, bit less than that, but in our first year we were very much focused on the recruitment market and we just saw firsthand what AI was was doing to it. Um so, we started training our candidates and at that point we realized, "Oh wow, the demand for just understanding AI and being able to navigate this massive overwhelm is outweighing the hiring market, which is declining and freezing. Um and so, for us it was a very natural expansion and and pivot and quite frankly, the best way to help anyone, whether you're in work, out of work, building a team, building a company, is to learn AI. It's the single best thing you can do for yourself or your company is to like really double down on learning this now because we're only at the start of the tidal wave.
In in the context of AI, why do you think the hiring market is freezing so much?
It's a really interesting question and I I wrote Substack on this and I like I like to compare it to the kind of industrial revolution or the introduction of the printing press, which is kind of a lame example, but bear with me, which is, you know, when the printing press was invented, you had this entire profession of scribes, literally people that handwrote, and they were fantastic at their job, but the market had completely changed and so all of these printing presses were opened and to hire for your printing press, if you're producing a book, you might think at the time, oh, you know, I'll hire someone that can write 100 words a minute, but actually that entire profession was was worthless and actually you needed engineers to operate your machines and that's happening on an unprecedented scale at the moment and so what we're seeing, unfortunately, is in this beginning stage of an augmenting of the the labor force, a huge swathe of people doing very manual jobs that AI can replace being laid off. Um but there are a new wave of roles that are being hired for. So, I actually don't sit in the doomsday camp of the workforce is going to massively shrink over the next few years. I think it it won't, but we're seeing a knee-jerk reaction, which is a lot of companies freezing their hiring or laying people off and so right now the workforce is shrinking, but I personally don't believe it's it will be as bad as people say.
What do you think are examples of new jobs that AI will create?
It's hard to say cuz there's always going to have to be somebody who operates it. It's never going to be a stand-alone thing. Am I wrong?
No, it might be at some point. No, in the sort of yeah, no, not in the like I, robot sense, Black Mirror sense, but no, there's yeah, fully agentic systems up and running that don't need to be managed. We have loads running at IV.
But wouldn't there be some like you're still there? Do you know what I mean?
but moving we kind of had the we have sort of I guess three stages of AI, the first being the chatbot. You had your chatbots which absolutely a human needs to be there at all points because you ask a question and it answers you. That's where most people are still at.
That's where they're at now.
At the moment which is they kind of use it as a glorified Google. Answer this question for me.
It's me.
Um or maybe like maybe summarize an email or maybe draft me a report, but it's still a complete one way the human has to do the prompt and they get a response. The next step is you can use AI for automation. So that's where you would input, you know, very simple or not simple, but very clear instructions which is, you know, if I get an email from my co-founder, then I need to be notified with a Slack message or a Teams message, whatever that might be, but they're very clear simple steps and that's known as an automation. Automations are now being AI powered so you could have a step in that automation that involves AI. Like if I get an email from Lydia, read it, classify it, and if it's of high importance, send me a Slack message. So that's kind of stage two and then stage three which is what we're seeing, you know, the highest caliber companies and individuals operating at the moment is that agentic AI and the word agentic lots of people hear but don't necessarily know what that means and that's essentially you give AI goal and it can do whatever it deems necessary to get to that goal. So in that example the goal might be I need to be notified when there's a high priority email and I haven't given it the very simple steps of what I need it to do, the triggers and the the waterfall. It has access to all my tools and it can just do whatever it needs to do to make that goal happen and those are the kind of systems that we're seeing running on their own. Of course, right now you do need those puppet masters, if you will, but I mean, we're seeing this with like huge layoffs at tech companies where that whole middle management layer
Yeah.
is going because instead of managing a team of of seven or 10, you're managing a team of one and three agents. And so, that's at the moment where we're seeing this like knee-jerk reaction in the workforce of lack of these traditional roles. But to your question, we are seeing brand new roles come out, which is fully AI-native candidates.
what's interesting to speak about here though, so sales is the kind of job where you it's a human interaction. You have to speak to the way that we know it now, it's human to human. To what extent when we when we start speaking about avatars and human interaction, to what extent do you think we're ready to start accepting agents as humans?
Great question. I think it really depends on It does really depend on the situation because, for example, if I call my physio to change my appointment, I don't care if it's AI because my call's getting answered straight away and I'm pressing the right key and I'm speaking and my appointment has been rescheduled. Versus if I'm calling perhaps another company and I'm disgruntled and I want, you know, human-to-human interaction or I need to be reimbursed for something really quickly, at that point you want that human. So, I think it really does vary. But we are seeing resistance. You know, if you get a sales inbound email and it's clearly written by AI, you won't answer it. You won't want to take a sales call with that person. If you get reached out to on Yeah, LinkedIn and it's it sounds like AI slop, there's a massive rejection of that in the market, absolutely. Anything that needs to be human should remain either should remain human or should just remain human-sounding. I don't get me wrong.
Yeah, exactly.
There's outbound that's written by AI and sounds incredible and it's and it works. But if it sounds really obviously AI and if it's very obviously an avatar that should not be used for an for an you know, for a human-to-human role, it can cause a massive problem and it and it can be really bad for brand reputation.
Yeah, and for it to sound human though for now, would you need a human to finesse it?
I personally would be so surprised and you know, anyone can in the comments. If you know, if you can get a very deeply human sounding sales message with one prompt to an LLM, show me your prompt because you've you've won the game. There has to be tweaking. There has to be finessing. The em dashes have to be removed.
god, those dashes.
The dashes.
Yeah, the it's not that. It's just that all of those tells like that's where you need a human and that's why prompt engineering is still the number one AI skill. A lot of companies forget that their workforce are really bad at prompting. Really bad.
Sure.
No one uses chain of thought, few shot, persona prompting, all of these incredible ways to get the most out of AI so you don't you know, it doesn't sound like that. But a lot of people just go, "Hey, please summarize this." and wonder why the output sucks.
Hm. What are these AI prompting tools that people should use?
There's Well, we built one actually. It's free. It's actually really good. A A huge problem is the prompt. And the the one of the main problems with prompting is no one has time to write a long prompt. But everyone knows that a super contextual prompt will get you the best response. What we built at Ivy um is you type your prompt in. It works on all LLMs. It's completely free. It's a Chrome extension. So you just download it from the Chrome Store. You type your prompt, highlight it, and a little button appears and you just hit improve with Ivy. We read your prompt and in like half a second we give you a a better version of it with all of the logic and the theory of the best prompting strategies.
So, do you take on clients and you analyze their business and you think you can optimize this, this, and this and these are the options we're going to give you. Is that what you do?
Yeah, so the main ways we work with clients is a client will usually come to us and say, "Hey, we're we want to become AI native as a business. We understand the importance of this. We're We're not burying our head in the sand, but we're not sure how to execute that. Do we roll out training for everyone tomorrow? Do we buy a bunch of subscriptions? Who
Also, which subscriptions do you buy? Which are the best for my
Are we team Open AI? Are we team Claude? Are we team Gemini? What are people in my industry doing? Wait a minute, I can't let my whole workforce loose on Claude because we deal with really sensitive data and we're dealing with like financial data or P&L and things like that. And so, there's so many questions. So, the way we usually work with clients is we start with a strategy day with their C-suite where we basically get everyone's heads in the room to hash all of these questions out. But essentially, what we're trying to tease out is how are the workforce currently using AI? What are their prompting capabilities? What sort of tasks are they using it for? What's their excitement level? And that's always really interesting cuz it kind of brings out whether people have
fear I was thinking
amongst them and we call them rotten apples and you don't want rotten apples in your team cuz they get in everyone's ear going, "No, don't use AI. Don't use AI. It's going to take our jobs." So, we do that and then we analyze that report, you know, turn it into something that looks great. We present it to the C-suite and we say, "This is your current state of affairs." We'll get all of that get get everyone's ducks in a row around governance, around data, around cyber. Um And and it's always such an interesting day And a lot of people walk away from it going, "Oh, wow. I didn't you know, I was a bit of a skeptic myself. I didn't want us to embrace this, but but we really should." And then from there we usually build out a bespoke training program for the workforce so that they can actually utilize the tools that they've already got. We like to use the phrase like make humans human again, you know? Like none of us should be living in manual spreadsheets, copying phone numbers across, or
Mhm.
addresses across and things like that. Spending 4 hours of our day doing that. No, we need We should be focusing on value add for the business.
I guess there's just that knowledge gap. You just don't know where to begin.
Exactly.
Yeah.
It's such a good point because it's something that comes up with every conversation we have, which is we don't know what we don't know.
You don't know what you don't know. So, how do you know what you know? That's what I've been like, how did you get all this information?
It's essentially getting everyone to keep a timesheet. Some people already do this for a week and get everyone in the team to hour by hour write down like what exact task you're doing and what tools are you using to do it. And then with some, you know, complex prompt engineering, but again it's nothing crazy, you can input that data into ARA and it can spit out what you're currently doing that's really manual and where the low-hanging fruit is that you could build an agent to solve. We run this every 6 weeks at IV and the last time we ran it, we realized, "Oh, wow. This The email marketing software we use sucks. Why are we spending 6 hours a week on this software?" So, we immediately changed subscription to one that was a lot more AI-powered and it's incredible. We're saving so much time. Um So, that's kind of one one thing of how you can actually figure out what you don't know necessarily. But in terms of staying on top of stuff, we always have our ear to the ground. We spend just as much time like reading as we do anything else. And we're very bullish on the concept of micro upskilling. For a lot of people is the biggest point of overwhelm, cuz a lot of people just want to learn AI for a week and then
Move on.
Move on. Kind of the concept of learning, corporate learning as we know it has shifted so much. You can't just sit behind a screen and do a 50-hour course anymore.
That's the thing, I think training in corporates is so people are so disengaged, and it's just seen as a as a compliance box that you have to tick. But, what industries, based on what you've seen, struggle the most with AI adoption and why?
Great question. It's either the more like heavily regulated industries, you know, finance being one of them, law being one of them, or it's manufacturing, um construction. The Yeah, tech, FMCG are all very on it, I would say. Not That's not saying that they don't face problems. Everybody faces their own set of problems, but the regulation, um and the fear of of what data you're working with is the biggest sticking point right now for those industries I mentioned because it leads to them just freezing, cuz they go, "I'd rather do nothing than get it wrong." And that's actually extremely dangerous for the business because it means you you may not be around in 5 years' time cuz all of these AI-native competitors that have you know, been founded and built from from day one being fully and AI-enabled are chomping at your ankles ready to you know, knock you off the top spot.
Yeah.
And again, not saying that's like scaremonger, but it but it's true.
Yeah, of course you have to keep up with the times. But, what would you say what do you say to those industries? What are the loopholes that you can find to work within that environment in a safe way?
Yeah, one of the main recommended hires is that governance hire because it's so important to get your ducks in a row first, and at that point you can then roll out really effective training, but you need to know what data you're working with, what guardrails you you want to have put in place, what um you know, level of visibility these AI tools actually have over your data. And it's often really as simple as going into the settings of Copilot or Claude and toggling certain things on or off. Or it's as simple as creating a like data anonymity environment where an analyst can plug in data that has sensitive information, and it spits out the exact same data, but just with the personal information redacted. And then you can generate your report or do whatever you want with that data. But a lot of people don't necessarily know what they need to know. And so, a piece of advice is to either get help there or bring someone in because, you know, you best believe this is going to be a growing part of your business.
What are your plans for IV next few years?
Oh, great question. The number one thing we believe will never change as AI completely revolutionizes work is change. So, we're building fundamentally a company to help anyone navigate that change. AI products, AI tools can come and go, but what remains the same is this need to continuously learn. For us, our goals are are massive, you know, we we want to be the market leader in training on AI for individuals and teams. We don't want anyone to get left behind as AI completely reshapes how we live, how we work. We want people to be doing like deeply human, deeply value-add work. And we want to do it in an innovative and a fun way. And we are very, very big on our secret sauce being we we do really just like to teach AI in a fun and interesting way and help people feel really rest assured that this isn't going to take your job. It's the single best thing you can do for your career. And to get really excited about it cuz that's where you see the best come out of people in in what they can build with this technology if they feel really positive and not scared.
Do you think AI is making experts obsolete or more valuable?
I believe AI is making experts much more valuable. AI is based on the knowledge on the internet which is there's a huge amount of it, but it is finite knowledge and so anyone that has any element of novelty in their opinion, creativity in their opinion is hugely valuable. What we're seeing AI take is very manual, usually very junior level roles and so experience is now hugely worthwhile in any industry and the best combination that we're seeing lead to the most value is experience plus AI upskilling. So someone that has, let's say, 10 years in sales, but they are completely AI enabled. That is kind of the perfect storm across every single function, experience and AI fluency.
So you mentioned creativity. Where do you think creatives sit with within AI?
Creatives that are burying their heads in the sand and saying AI isn't helpful to their industry are lying to themselves, are killing themselves. The creatives that will win and are winning are ones that are using AI to aid and assist what they are already doing and to help them do it in a more human way. Great one is, let's say there's an ad agency or a, you know, branding agency, instead of wasting their time and money doing photo shoots for a canned drink, again, maybe controversial here, Nanny Banana can do that just as well as a human photographer these days. And so that creative agency is saving so much time, money, not just for themselves, for their client and then they can do what they do best which is creative strategy. And I think a lot of people forget how that's where the value essentially really deeply lies.
Do you think AI is making us smarter or lazier?
I think AI is definitely making us lazier. A great study was done by MIT a couple of years ago and it was called your brain on ChatGPT, like likening it to a drug.
And it was incredible because well, it was not incredible, sorry. It was terrifying because it showed that when you rely on these tools so much, you lose brain cells kind of fundamentally is what the study was saying. And you see that. So many people, especially young people, grads, not only are they living on TikTok and they, you know, have the attention span of a goldfish, but then they're also they've learned to rely on AI before needing to find something organically on Google and actually read stuff to find the answer. And I think that is incredibly scary for like this new wave of talent entering the workforce.
Which is where education then comes in.
But then I think it's also like kind of savage. It's also kind of natural selection in that way because then it's like, right, well, who's going to be the ones that can appreciate that and see that and have that difference.
Something I find Something I find really scary about AI is you're seeing kind of people not learning and reading as much, but then you also see because people are utilizing AI, there's like so much less new knowledge entering the world and entering the internet. And that's what AI reads. AI
as well, yeah.
AI reads the internet. It reads labeled data. And the less people that produce that, the dumber we all get as a collective.
thoughts.
Exactly, that original thought. A great example would be book writing. More people are using AI to write books or films or scripts, and that means the data that AI then learns from and is modeled on is like its own slop. What's that going to mean in 10 years if you want to write a book with AI and the last 10 years of books have all been written by AI, it's just going to be and all that kind of stuff.
a standard. You don't have a standard anymore.
[snorts] Yeah. Do you think AI has peaked or do you think it's just started?
I think we haven't even seen the break of the tidal wave with AI. Um I think there is so many capabilities out there that have not been unlocked. I think if you look at what's happened in the last year, let alone 5 years, the pace is incredibly scary. I think AGI is something we can achieve very realistically in our lifetimes. Um
What does AGI stand for?
Artificial general intelligence, which
yeah?
Yeah, it's it's kind of an interesting one, but it's essentially that AI feels, acts, and has sentience.
Yeah.
Like a human.
Autonomy.
And has that autonomy, and with that you could have a whole host of benefits and problems as well. Um it's a really interesting debate about whether that is even possible to be reached, but yeah, there's some there's some incredible new companies being launched that are like building cuz right now there's three main companies that build the foundational models, OpenAI, Google's DeepMind, um and Anthropic. Anything else that you use on a day-to-day is a wrapper, so like Copilot is a wrapper, it's not an actual model. Um but there's a lot of startups that are getting funded, like we're talking billions of rounds in their billions of dollars in their first funding round that are building brand new models with capabilities we can't even dream of.
Well, thank you so much, Amy.