340. How A Serial Founder Is Rebuilding The Future Of AI w/ Husayn Kassai | ollo

0:00I'm Husayn Kassai. I'm the founder and CEO at Ollo. Ollo is the simplest agent builder for non-technical people to build and run agents that work.

0:07Amardeep Parmar:So we had you on a few years ago. Yes. And you've made a bunch of privets since then, right? Correct. So can you tell us like that journey? So you must have been on it here, I would say, late 2023. Correct. If I rewind right back to 2010, started on Fido, the identity company, built it for 10 years, along with two co-founders. And that essentially was the precursor to Gen.AI. It was computer vision machine learning. and over that decade I saw how as AI strengthens it lowers the complexity especially for non-technical people to use it and that in my mind has always been the unlock of what AI is able to deliver.

0:42So we started with enterprise search three years ago the power of AI to at the time with the models unlock the ability to retrieve things and since then being pushed by what clients have asked is to build agentic capabilities on top so that ultimately companies are able to compound what they're doing and be able to do it and share it across different teams.

1:02Amardeep Parmar:And like as you've been going through this journey right so obviously Onfido is massive right and then you decided to go back into your arena again start again from scratch and you've also taken so many lessons from that to do what you're doing today. What are you doing differently now? There is startups have a pre-product market fit, product market fit and scale component. So I came from a scale mindset and I had to rewire myself to what is it and what should you be doing other pre-product market fit. So the contrasts are essentially at a scale stage, you're constantly delegating and you're constantly going for market share and you're hiring.

1:40It's really, really the opposite when you're back at the pre-product market fit stage. You shouldn't be delegating. You should be doing as much of it as possible as yourself, especially customer calls, invoice processing, paying payroll. You certainly shouldn't also be hiring. You should have a lean team, 6, 8, 10, 12, depending on what stage you're at, but no more. Small agile lean team, because that is the best critical number to have impact and yet have strong communication, be able to pivot, be able to be agile. So there's one thing around what are the different things needed at an early stage again, and to be mindful of that.

2:17And every 18 months, it changes as you go through different phases. The second thing probably is I'm just enjoying it a lot more this time. So the first time, it was just one fire after the next, and I'm expecting in every startup is one fire after the next. But because I was 20 when I started on Fido, and I was maybe 23 when we had a team of about 25 people, and many of them had mortgages, families. So it wasn't like a fun student society that wants to get on with doing cool stuff anymore. which is my mindset because prior to that my only job was being present until Oxford Entrepreneurs as a Student Society.

2:52So that responsibility is something to be mindful of and therefore there's always so much pressure that one is not able to take a step back and like enjoy it as much. But this time I'm genuinely enjoying it a lot more.

3:09Amardeep Parmar:How do you make sure that you enjoy it now? What are you doing differently in terms of like is it better balance is it like just not taking on tasks you don't enjoy so it's certainly better balance but the difficult thing without experience and judgment and so on is you get pressure from different sides and if you're a founder especially first time founder trying to do something that hasn't been done before you're going against the grain for the most part and you have a gut feel and you want to push in a certain direction but those around you external internal media trends are often against you so you have to be pretty resilient and able essentially to power through and you're going to doubt yourself right so not just at the early stages but even if you kind of make it and you're on sort of tens of millions of revenue you kind of hire sometimes senior executives as we did from the most prestigious companies in the world, Amazon, Google, Meta and others.

4:12And your gut then says, okay, this many worked out, but some certainly didn't. And your gut says, this is not working out. But it's again, am I going to be able to go effectively and tell my board that I want to ask someone who's just come over from, I'm just going to make up a name Google to leave. And then they'd be like, you're 25. What are you, what are you, look at their credentials. So those are the kind of things that make it difficult. Whereas this time around, because I've been through it, I'm just much more comfortable in decisions that are more steady and I don't need to worry about persuading others.

4:51A big factor is I'm semi-bootstrapped this time, so I'm not dependent on other things. Most of the 10 years at Amfido was, in my memories, is like fundraising, one way or another. I'd get like nine month breaks between rounds, but at the time when he wants to raise the next round, you kind of start six to nine months before you need the money. And every 18 months you're kind of in a cycle. But this time, I've got my own funds, I've got external funds, so it's kind of the right balance of pressure and yet the control of the decision is mine, which really helps.

5:25Amardeep Parmar:Do you think as well, sometimes people want to have balance from the very beginning. Do Do you think it's only possible today because you've got that experience or could have found and listened today a shortcut in any way? Because I think it's kind of difficult for somebody first time around to really be able to do what you do. Balance is work-life balance or what kind of balance? I guess being able to enjoy it in the same way. Because you can't necessarily push back as hard when you don't have the experience you have. So there's balance in many dimensions. there is the balance on building something big and yet optimizing for let's go as big as possible whereas if it's your first startup you're still on credit card debt it's like hey there's this like this is weird because in paper wealth it's hundreds of millions but I still have a credit card bill at the end of the month that I'm trying to figure out how to pay.

6:23So there's a balance around, you know, all the eggs are in one basket. And as such, how can I get some sort of evenly distributed? And that's why secondaries are good. Small secondaries, not too much to get you distracted and like spoiled, right? I'm not into like tens of millions. In my mind, it has to be half a million, one million, whatever. So there's that balance. Second time around, you can go all in in a more material way. And this also is one of my reasons why I think, statistically, founders that are from high-income households and privately educated, essentially, they're five times more likely to raise seed rounds.

7:04And that in part, in my mind, they're able to take risks and start companies because they're also not worried about paying the bills, right? So we're missing out on many founders, Asian or otherwise, that frankly are shuttled with student debts or a mortgage or what or even rent that how can you possibly think about innovating right um as such is balance on that that this this sort of helps work-life balance yes if i say i talked a bit about the advantages of second time there are disadvantages when you're 20 20 all the way to 30 right i can't remember going under london underground any point and actually waiting on the stairs on the escalator if i'm being very honest with you this is like therapy for me i now don't always go up the stairs i just wait for the escalator to get me there uh as an example as like a contrast um it's partly because i'm able to take some more time uh now to put specific numbers the first i you know there's this whole thing around 996 founders which i think is utterly stupid and completely false in terms of it not being effective but And I'm not saying startups are easy.

8:17I'm not saying you mustn't work hard. You shouldn't come into this thinking it's easy. And you shouldn't have founders encouraging others to come in and pretend it's easy. You certainly will work harder than most other roles. Most people out there are working hard jobs. Many are doing two or three jobs. I worked at McDonald's in my first job. I worked at Marks and Spencer's. It's not many less hard than this. But crucially, so it's hard work. But 996 is a bit excessive. and you're not working effectively. It's a bit about working smarter. It's not just about working harder. And as such, there's a balance to be had in my 996.

8:56That gets it wrong. Doesn't mean if you don't have a sprint coming up or you have a deadline or a customer, yeah, you should work all weekend sometimes, but as an exception, I was the norm. So back to balance, I am able to now do, if I used to do six day or five and a half day weeks, that would have been put my numbers like 9 a.m to 6 p.m but then another couple of hours at night previously now I can do nine to six and that's fine five days and I normally do four hours on a weekend like very time bound so I'm probably doing call it 20 25 percent less in terms of actual hours than my previous setup but I would like to think I'm twice as insightful with context and judgment.

9:42So I'm able to make fewer wrong mistakes. I'm able to be more decisive. I kind of know what I want. I'm spoiled by having incredible team members. And I mean, like incredible team members, that all of that helps. So balance in that less energy, and less hours than I would a bit more on the sort of wisdom, perhaps. And I think that kind of that definitely helps.

10:08Amardeep Parmar:Yeah, I think it's like you said as well. I look at process as seasonal, right? So this is meant to be a quiet season, but it's a little bit busier. Summer months. Yeah, and then say September to November, I know it's going to be busy, but then only December and January I can kind of take it a lot easier. So for me, it kind of has that kind of cadence there. You said about the big goals, right? What are the goals for Ollo? On seasonality, on basically being able to, well depends what size and scale you are at a company pre-program or kithorium scale i can so at later stages um you kind of you you prioritize where you have what you might call like quieter periods for other things yeah so summer and december are very much planning periods uh anytime it's the end of the quarter be that uh june or december it's sales crunch time because they need to get their numbers in.

11:01So it's actually one of the most busy periods at a scale, at a late stage. If you're in a startup period, and it's sort of quite, again, back to planning, we do our company offsite at the end of June, last week of June, every year. And that's essentially pretty important. So back to all those goals. I believe special companies are built based on the team and the people, crucially, working towards the right targets and the right goal right so we are we went through a period of like token maxing and now agent maxing we're not we want organizations are crying stuff few agents and agents that work so a key focus for us is how many reliable agents do you have inside your organization running and the reliability really is a key key word here on that's one key metric that we want run towards we're going to maximize reliable agents inside organizations and we want what we want to reduce is a number of corrections a non-technical person needs to make to have an agent be reliable.

11:59And essentially, large language models are useful. Essentially, it's the internet's best average answer. When you apply it to a specific task in any sort of work, it often won't do a good job. So therefore, you have to give it feedback loops. Those feedback loops, the orchestration, essentially a harness. And our job is to make that feedback loops as straightforward and simple for non-technical people to work on, refine, and essentially get the outcomes that they want. Those feedback loops, we want to minimize as much as possible for the end users. So our goals primarily are essentially reliable agents that are out running.

12:33And secondly, the number of corrections needed for clients to get to those reliable agents.

12:38Amardeep Parmar:And as well, right? Like what people are selling to, they're going to have so many different people talking to them and trying, they're trying to figure out like what's the best path for us, right? When you're talking to customers or when you're trying to get new clients, how are you doing that? How are you framing that to them? Our approach that works well is workshops. So you have a lot of misconceptions on AI and a lot of different components. It is workshops work quite well. On the flip side, many industries, majority of businesses are struggling with AI deployment. It's the biggest elephant in the room.

13:10Well, AI promised all this and the delivery is so, so bad. The majority of these are using GPT, drafting emails and thinking they're AI-first companies. So that journey is typically workshop. Then it pushes on to, we're now doing paid pilots where we have to, otherwise turn it into contracts, essentially. And the goal becomes helping unlock your non-technical employees, your operators, partnering with our AI operators on our team, helping them essentially get agents to work effectively inside businesses. Have you got any particular niche you go for in terms of your customers? So there's a sweet spot around fewer than 500 employees is our preference.

13:52The biggest barrier normally is access to data. And when you get a problem on access to data, we much rather have clients that are using SaaS tools. SaaS tools, by definite, Notion, Confluence, Slack, and those types. That tends to mean the startup and scale-up community, where sort of my own feeder background comes from, we just prefer that part of because of the network, part of because where we essentially come from. They just move much faster. and they have a lot of work to be done. So their internal team, they almost all have an engineering team, but the non-technical operators, we want to turn those into 10X operators.

14:30And for them, we want to help them without them needing the engineering support. So we've been structured in a way that once we go into a client in the business, the first step is connecting to all their tools and systems. And essentially, that is where some engineering is needed, but that takes an hour or less, because those integrations are straightforward because they are typically using SaaS tools. And then I'm from an identity background, right? Permissioning is access management. So they set the rules on access rights, and then the engineering can leave the non-engineers to just go wild if they want to.

15:04It's primarily sandbox environments. And that unlocks the non-technical, the actual operators, then do what they need to do to succeed, but they need targeted support and targeted training. And that's what we're there for. And that complexity is continuing to go down. So it's going to be less time needed from us to help them actually get reliable agents into production.

15:26Amardeep Parmar:Hello, hello. I hope you've enjoyed the show so far. I'm Amadeep Parma, co-founder of BayHQ and the host of this show. For those of you who don't know, BayHQ is the community for high growth Asian heritage founders and investors in the UK. Over 7 ,500 people have attended our events. 200 people have been through our impact programs. And obviously there's been over 250 episodes of this podcast. If you want to join us and take part in the programs and come to the events, go to bayhq.com forward slash join. We also just released our first ever book called Startups for Outsiders, which you can get on Amazon now and the link is in the bio.

16:07Amardeep Parmar:Hope to see you soon at our event. Hope you enjoy the rest of the episode. And you've got the team pretty lean, like you said, right? You said there's a certain amount that you think is where the team is both agile, but also getting things done. What do you look for in those team members? What is it about those people that make them so good? So we're four team people. Half of them are former founders. And that's, I think, the three things we look for is hungry, humble, and high potential. Hungry is something that you can't buy it's that person often from adverse backgrounds uh who've gone through adversity who are have a fire in them and they want to pursue excellence and they want to continuously raise the bar and they have that hunger and you operate differently when you're hungry so that is really really important there's a massive need and desire to to to win the second Second is humble.

17:07And hunger, we kind of look at as adversity quotients. Have they gone through adversity a lot and may have to go jump through these hoops and survive anyway. And now they get a kick out of putting out fires because goodness knows there's always going to be fires in the startup. Second is humble in that it's more EQ. Are they good to work with? Will they train their colleagues? Will there be an example? Will they set an example? Are they going to be selfish and focus on themselves and things of that nature? And third is high potential, which essentially means the propensity to learn and enthusiasm for learning and continuous improvement in that side.

17:43So those three things, I think, are the most important thing that I've held throughout. And I think particularly for a startup, you need. Now, you kind of need a different skill set, early stage, mid and later. But definitely for the early stage, these matter even more.

18:00Amardeep Parmar:And I think this is the last thing you came on as well, right? So your voice has grown massive in the startup industry. And not that you didn't have a voice before, but now your policy and you're also involved in so many different areas and also the London AI hub as well. Can you tell us what made you go for the London AI hub and to contribute to the whole ecosystem and the way you are?

18:24So basically I started writing a little bit more, especially on LinkedIn, because I found myself getting frustrated at what was being put out. part of it was the AI hype in 2023 and summer 2024 every other piece was AGI every other piece was like 100 % of the workforce is going to be automated every other piece was like the model providers are going to dominate everything and I just I've been building an AI since 2010 like I just knew that this is far from the truth I didn't bother me we always have hype there's a blockchain hype crypto hype sharing economy we were just this is a very hyped up weird weird industry that we're in.

19:02But I started seeing having, I started to notice it having negative consequences. Then one episode like really had an impact on me. It was a dinner with a policy advisor to the government, especially on education. And this lady was not too interested in AI, the realities of AI and not the reality of AI, because she was quite convinced that in the future, there will be no jobs. So why are we invest in education for for for the future and i i worried i thought like how can we have advisors to ministers who are talking about education in a way that's not needed because my number one thing is under 18s are getting the worst education they've ever had and we're really really going to fall behind and the education they need is not just proper classrooms proper resources proper schools proper teachers that are paid properly and all that sort of stuff but also the the right hemisphere of the brain stuff you know after school clubs um libraries swimming pools all the things that we're closing down all the bad that's happening i want to reverse i mean if i've had a magic wand and equally and you know there's nutrition and the benefits and not like the need for it and so forth so i kind of got a bit disturbed by that these trends are are the mainstream media and social media and big tech are all pushing these narratives that are starting to have very adverse consequences.

20:25So, you know, my tiny voice, I just put a, try to start to put it like a counter narrative. And the main themes for me are like macro economy type things around these, especially AI trends, micro economy and inside businesses, you need to invest in training your people. You need to own the technology. You need to have independent all these those kind of kind of things are essentially at the heart of it and I particularly care about education for the under 18s because I think as a country we don't have a necessarily a shortage of financial capital going into startups and scale-ups we have a shortage of human capital we have a shortage of founders and operators around them to build the amazing technologies that we all need so just being able to write a bit more I see it as like therapeutic on the

21:13Amardeep Parmar:weekends and with the London air as well the community sorry so so I am a big big believer of communities as you know I'm a member of yours I'm a very happy member and so I was a big beneficiary of the fintech community in the 2012 2013 period and we had level 39 in Canary Wharf the 39th floor of Barclays building Revolut with tenants there the regulators the FCA would come all the time and all the cool events would happen there and on feeder was kind of like a provide a rectum, I would attend where I could, but you just know the quality would be first class. And I saw firsthand how this helped us get regulation rights, get innovation rights, and get competition rights and everything else.

21:55And so we became the FinTech capital here in London. And as such, I wanted to move my team into the AI equivalent. So it just didn't exist. I just assumed for a period it doesn't exist, but no one wants to tell me where it is. And then lo and behold it doesn't so with a few partners founders forum group technician and and Morantix so Morantix have the most well thought through playbook successful in Europe and they deploy that in their space in Berlin first-class AI center of excellence and founders forum group and technician are just so well networked and are known for delivering amazing value for founders and as such it was a case of they we all kind of agreed this is needed and we got lucky to start the community about 15 16 months ago and it's very much designed it says no it's non-profit it's just designed for our ambition is to help make it easy for ai companies to come and work together and learn from each other and become and all the events that we do are free every two weeks we have um dbt come and ofcom and all the sort of ai minister come and do events there we've got anthropic 11 labs doing hackathons there and we have a pretty significant program where people can apply and come and pitch for government procurement or pitching and all of it is sort of you you apply and you come and we specifically focus on having those who either are in stealth or have not raised or may not have a profile but you don't have to have any of that to come and pitch and and um so i'm really glad i'm really glad i'm it's it's starting to ice a space i'm helping create that i really want to exist for my own team so we posted a couple events there ourselves as well so

23:33Amardeep Parmar:It was also, I think I remember I was walking around the building and I just kept bumping into people I knew because there's so many founders there who are doing... That's the vibe we had back in the fintech days. But the space is small. So the hope is to just have it be even bigger and for good things to come from it. One of the key things, the hope is when you have proximity generates luck. So I've kind of had a view that, let's say YC, very special. No one can question the success. It's actually not, in my mind, the playbook that YC has that makes people, these companies successful. The playbooks are online.

24:17They're free. Like the version you can have, a startup school and others. Now, it's certainly valuable. But what they have is essentially an insight into trends that are coming. And they publish those. And then that creates like a honeypot for people to apply. and so they get people going in building on the things that they think is going to be big then what they really do is it creates this proximity and like loads of collisions and connect you to anyone you want to be connected to and that increases the likelihood of luck I'm a massive believer in luck and that these groups and these communities rapidly increase the proportion or the propensity to get lucky so that's wonderful but I was rejected by YC so yes I've actually never said that and so one of the ones they regret one away yes yes certainly but um I you don't need that like if you're a founder that doesn't have access to these things you can you can put in the hard work and go to the events events are published online right and I want to create a space where all of those who are want to build and want to build something special are able to start the process of being in the like-minded the community because that's when sort of like magic happens i think that's the thing as well

25:31Amardeep Parmar:with community like say with us where when you're supposed to be tracking things a lot of times you don't really know like there's just people just meet each other and like later on some like oh i got my job because of you guys or oh i got like 500k investment because i sat next to that person at the event it's like it's like you said it's like i think something when people over engineer it so much of it is just that proximity and just being around people and you just never know because sometimes you know people are very transactional like is this person useful to or not and sometimes the people you don't expect who can open up the doors right yeah and i i feel that and i believe that and uh there is the yc community connects people all the time and we have founders here and it's becoming a thing where we kind of like help each other um so you cannot know that transactional approach doesn't really work in my mind the better approach is and and paul Graham has said this like help and don't expect anything and do that so often that magic starts to happen um it's just so much easier to do when you're kind of physically around each other yeah so one of the things we try to do now the community is that I increasingly became the bottleneck right everybody messaged me about this or messaged me about that and then you've got the same thing it's like you get stuck right you can get agents to connect and help yeah so now we've got the platform so there's like I think over 300 people who volunteered to become mentors on the platform so now you can just see everybody now connecting with each other so you can volunteer become a mentor you can also see who's looking for mentorship and people need to reach out to each other and i think the idea behind that is that we don't want to be the people in the way and sometimes like for whatever reason people can't come in person i always think the best way is in person that's why i'm always at events because that's how like so many of our lucky creators have come i just was at an event i chatted to this person you then got a relationship you saw another event and it kind of compounds from there right the in-person thing the attention is a big factor so i might be in front of someone that we can help each other they might reintroduce me to someone i hire i might connect them to a company that might help them improve their design and you get each other's attention it's more memorable and you get live feedback and you start forming a relationship and it's actually a series of relationships um that that matter more than anything else it's what was the saying I liked recently it's not who you know it's who knows you and you just have to put in the work you have to be present you have to go to the conferences you have to go to the events you have to put in the work and as founders it's kind of part of your part of your job and it's hard because you can sometimes spend a week like I did three four things should I be back in my office like working away or can I be spending sometimes with events and it's not easy to get the balance right but there's your endopathy and you have to be known and be involved to frankly for you to 90 % of revenue for B2B companies that are found like you know at the early stages the revenue comes from founder generated leads and in this crowded world noisy world in person I think is really the way to go yeah I probably get about 100 spam emails a day now right from people trying to send me some kind of software yeah and that's only going to amplify yeah but you're not going to get 100 people coming to you on the street pitching you but not yet yeah yeah all right and like it's interesting as well because like you're like one of the voices about like making the change in the positive way right but you've got all these people the doom saying things like that so people listening who are like maybe looking for us by the next few years and what like the uk and the ai ecosystem looks like where do you hope this all goes like where do you hope that the uk is just in what we're able to achieve together the first and most fundamental change that will have the biggest impact over the next like 10 year period is the extent to which we get education for under 18s rights and that is holistic so you know classroom schools but also nutrition and and all the things that kind of go go with it um there is lots to be pessimistic about including globally the things that really get me down though are not the ones that i see on my linkedin feed for the most part there's so much to like the UK has got so much going for it there's healthcare service, even the post office is the treasurer, the BBC is or maybe was, it's a controversial one the talent base the universities, the ability to for someone to come from a school from an underprivileged background and build the biggest AI company of the decade or whatever there's like all these things that can really happen but the sort of The mainstream media, for the most part, is something that frustrates me because I see it as an extremely damaging force for the country and for the mental well-being of a share of the country, often in retirement age, that should be enjoying life and not have to be concerned over things that they really don't need to be concerned about because it's mostly exaggerated and false.

30:23And political activists that are exploiting the often economic inequalities that we have for their own ends, this is like far right in particular, and it's on the rise. That's what really, really gets me. AI, there's areas of it that's awful and hasn't realized with any new technology, but it's nothing I don't think like regulation and communities coming together and collaborating and getting the right balance can't resolve. There's many components of it that in my mind are great potential equalizers. Health, access to health and access to education is like pretty, pretty key and crucial ones for me.

30:57But on AI, I think in 10 years, you're one of two camps. If you believe in AGI, then by definition, you kind of believe that a handful of these model providers, especially Anthropik and OpenAI, are going to rule and own and control everything. That's a pretty grim view or outcome if we get that. But if you don't believe in AGI, you kind of think there's room for startups. And you think there is a world where it's more decentralized. And there's the power, and it's all about where the power lies, lies with startups and even corporates that they have much more control over their own stack and as such aren't beholden to big tech.