In the final episode of Series 2, we meet Maryam Torshizi, the founder of DeepSearch. Maryam's enduring fascination with data and its management led her to realize the potential of AI tools in integrating and analyzing vast data banks, thereby providing compelling insights for her clients. Despite the challenges she faced in establishing the company, including significant financial constraints, she has persevered. Her openness to learn from others' experiences and to reflect on her own development has led to valuable insights. Her new watchword is patience, and she has also learnt the art of selling - both to clients and to employees.
Connect with Maryam Torshizi on Linkedin: https://www.linkedin.com/in/maryamtorshizi/
We hope you enjoy this episode! Give it a like and subscribe if you'd like more content like this :)
From
The Top Floor Team
#ceointerview #businessleadership #businessleaders #ceo #ceotalks #businesstalks #ceos #ceosdesk #ceoadvice #podcast #podcasts #podcastshow #podcasting #podcastclips #podcastseries #thetopfloor #topfloorpodcast #foryou #foryoupage #fyp #fypシ #fypシ゚viral
welcome everybody to the last in my second series of stressing values and I think we're going to end it with quite a bang. Today we have Maryam Tarshizi, who's the founder and CEO of DeepSearch Limited, which is a very interesting, a very topical company that's taken forward AI in some really quite creative ways. So I think you'll join me in being very interested in what she's got to say and how she's running her business in this new world. So welcome Maryam, thank you for coming on stressing values. I don't think I've done you a full service by introducing you merely as an AI business. So perhaps you could tell us a little bit about yourself and about DeepSearch and what it does. We started as a thematic solution but then we realized we are actually a search engine because we ingest a lot of different data sources, media, audio, video, images, text basically and then we try to connect all of these data sets together and we create this map of information and then this map you can basically query or contextualize with large language models in three different ways. A fourth way will come soon. So the first one is you can summarize points on this map. Second one you can interact with it like just you would with a chat box and a third way is to turn this entire map into a report that you could basically use to give you a consensus view of a topic you're interested in. Wow and how is that different to the popular perception of things like chat GPT? Yeah so we spent a lot of time on data engineering so when we say data engineering we don't mean like just inputting data in Excel files. It's machine learning basically end to end. So we use image to text, table to text, PDF to text, audio to text so everything turns into a textual format and then we apply that's computational linguistics on top to clean this and then we ingest that data for training purposes in our large language model and as a result of the way we work with the data we don't get hallucinations. Yes you can avoid all of that. Okay because that's what chat GPT is famous for isn't it kind of making up new information? But as you don't you get proper references like existing references not hallucinated references and it's it gives control to the end user so the end user can see a chat interface or can see this map of information and this map of information is every information that is out there about your topic and so that gives you control so you can say oh okay well I'm looking at for example Donald Trump and what else can I see with Donald Trump and it will give you all the ideas that you should perhaps use to do your search and it's fantastic for ideation. It also allows you to get accurate answers on the questions you might have or an accurate reports like a consensus report on a topic you're interested in. So these things actually chat GPT doesn't do so we are more targeted at research and insights for people who get who want to get insights into things or investigators basically so it's three different fields and and research can be anything can be done by a portfolio manager can be done by a head of research in some firm or can be done by an intelligence officer so there are like different different types of people who use it for reset investigations can be done by media companies by intelligence offices again and then you've got the in-between which is content generation and ideas. So when you say intelligence officers do you mean the enforcement agencies? Not enforcement agencies but more of other kinds of government agencies so they work with defense predominantly. Interesting, wow so very diverse and so you had this vision you pushed it forward through many months of without any money you get the funding there's a story in its own right there's a whole interview just for that I think but and you got it started where do you see it going? I have been so I genuinely ignore the noise that is in the market as long as it is not affecting us a lot so for example when Tat GPT first created its front ends and went to markets I think was November 2022 people suddenly got excited and I thought it was a smog and mirror but basically just ignore it for now because nobody knows how they're going to use these there's no applications just chat post and I was right so I think a lot of people still struggle to find a use case for these sorts of applications and there isn't an application it's a model basically and I think people are struggling to integrate it so I believe that the way we started going for where we were unstructured predominantly so we were just ingesting unstructured data but because of the set of clients we got we went down the path of structured semi-structured unstructured data which means text images and videos on audios plus excel files that have got just numbers plus excel files that have got numbers and words so you now ingest everything and anything basically and you can then we link them we link everything up and then you can contextualize it basically using the large language models and it's a totally different system to what is in the market and I think through this we are able to grab more of the market than many other firms because we're not just an unstructured focus company we are focusing on many different data sets and data qualities and that's why I said we are as a core I would put our USB data engineering because data is the most important thing and there are so many variations of it and actually that's kind of brings us to where I see the future we have got consultancies on the applications intelligence firms on the applications intelligence and corporate intelligence we've got asset management firms and then we've got consumer goods companies and they each have got different needs but the core tech that we've built is so flexible that they can all use it the same way but getting what they want to use it yeah you use it the basic process is the same but they're using it for different wow exactly it could be replicable and applicable for years to come the basic process keep on growing in different industries so that's where I want to go like for example we were approached by media companies very large media companies recently they want to use the application because it gives them privacy and they can they can choose whether their data is going to be ingested or not so they have and then their editors can use it to go back in time to do research and to connect different magazines they have so many different magazines this company and to connect the outputs from these different like magazines to provide more holistic kind of view of a specific topic that they want to write about and research on yes yeah and but of course it's now confidential and it's not shared all the new stuff is not just shared out yeah actually the big risk with the other models yeah absolutely so you're an engineer by profession what new skills have you had to evolve to become a chief executive oh a lot um I would say the most important one was patience I think um my major very impatience and um that's and also the work that I did after university was also in finance which is an impatience industry so they don't have patience either so it either has to be perfect and on the first so and quick yes so um I had to learn to be very patient because this is not a sprints unlike in like the industries that I worked at it's more of a marathon so you need to like take your like take um calculated measures basically and steps in order to survive the long game and um one of them was patience and I had to learn the hard way and then I think other things is for example there are many others like marketing how do you do marketing I didn't know how to do marketing initially and I had to learn what would you but um I had to learn that marketing is not just putting information out there but actually trying to target your end clients yes and then so there's a whole circle of connecting marketing to sales that's I initially I completely missed because I didn't know. Yes yes it's like a cyclical thing so you you market to get clients but then you get clients but then you got to keep marketing so you don't just stop no no you can't stop no no it's constant effort yes yes and that that was another one and I think um I'm very good with engineers and engineering so I understand how to manage them and that's that's something that's has been like I think is through my background to be honest and hiring sales people was a difficult one because completely different yes we need different uh style um they get different pays like even their their packages are different different and for me I had to get like I had to get familiarized with that so like initial chat that I had with the first sales person I was like oh no he's robbing me you know that was that went through my hands and then I realized actually no this is the this is a norm in industry and I didn't know my expectation yes absolutely yes and you've got to match that otherwise you you you won't recruit yes absolutely exactly wow but it's such fun that's so interesting so underneath that when I've gone on to your website I've seen quite a lot of the use of the word ethic ethical data ethical stuff so for you personally what does that mean what are your core values what what lines were you not cross we actually came up with our first so we have a whole document on how to collect data and that was the first thing that I came up with is rules and regulations for the company people and how collections should be done and what what should we do with and and clients data was something that was added later but initially it was all about scraping and crawling the web so the web has got its own rules so and these rules are generally written in little compliance section on the websites they're always like in like very small phones you can find them and then you can read about them or there's something called robot texts and these robot texts Jeremy the media companies would have them they would allow you to scrape like they would tell you what what you're not allowed to scrape and what you're allowed to scrape but you so that is really important to read and to understand and not to collect things that you're not supposed to collect and on top of that social media was another one that we needed to think about like how would you collect data on individuals and individuals should be given permission should give you permission you know so to to collect their individual data that's one thing and second thing is if they are in a public domain like for example there's a debate happening in a group and that Jeremy is considered by many as a public data because you're putting yourself out there and you're putting your opinion out there so your opinion can be so if your opinion can be quoted therefore your opinion can be taken so this is this these rules I think it's not very clear to consumers but it's it's at clear when it comes to data regulations so we had to write a lot about that as well and then another one was about end client's data so how do we handle that right to ensure that there's no data breach to ensure that they're comfortable with whoever is accessing their data so in the context that we write to engineers name plus my name so that's we are taken accountable if something goes wrong with their data so not just a company but us three individually yeah and I think that is really important because that gives the end client's confidence that you're not just playing with their data or taking their data without their concerns and they have the ability to delete things and read like inputs new data basically in the system so they're in full control of their data and so you're not keeping it you you don't have it as such it's theirs and they amend it you're just holding it to reflect back to them having you do the sort of the search and then reflect back the information but it's not your information it's forever the client and it's written clearly in the application that in the country that we write them that data does not belong to us it is theirs and they have full responsibility for that data and if something goes wrong for example there is a data breach we are responsible for that if something like for example if there is a data that is not supposed to be in the application they are responsible for it so yes yes so it sounds like integrity would be one of your core words very important I think integrity ethics of collecting data like all their importance without these in place I think especially with the regulations that are probably going to come out on their labor government I think we would not be in a good place but we are currently in a good place and as a startup I think it's really important to build the foundation correctly because otherwise you never know maybe you want to sell your company later and then to somebody and then that would be a whole due diligence done and then if there is any breach any sort of problems you are in trouble basically you're going to be in trouble absolutely so has holding to these values caused you any issues in the time you've been running the business has it had there been any conflicts issues like for example when we when the clients go on the application some of them they ask for can we see like the sort the media sources that you've included for example okay sometimes five sets of media sources so then we have to remove them from the system and then it's it's a bit of a hassle on that front and they the media source have said that we want to be anonymous or not anonymous not once the specific because we only scrape public information that allows this great maybe a website that we are scraping does not is not a website that they like so then they say can you remove all the content from this for example specific website we don't want to see anything and so that's the whole of a hassle right there and that's that's an issue but with regards to contents there's always contents like you won't believe this but the articles that ft producers like they they are rewritten by different media sources but with like one or two sentences difference okay i do believe yeah people poaching other people's journalism yes it has their own which is quite interested in your on the website it said as a for instance you said that a client could say we don't want the BBC or something like that but if you're trying to present to the client all the options and to help them really fully understand the theme and the depth of their topic how does it help that client to take to enable them to take out one whole stream of information and do you push back and say you really need to keep that in be well how do you manage that so i spoke to one of my advisors William Natum belt about this at length actually and one of the things that we said was that we always say to the end clients that we are meant to give you the whole view of a topic so if you're just removing something because you don't like it or a few things that you don't like then you are your customizations and you're creating basically the bubbles like you're on you're creating your own bubble yeah and and by creating this bubble you might miss crucial information and i've been asked by clients why don't you just wait some like journals like media companies or sources more than others and i said but but i'm introducing my own bias and would i like to you to have my bias or should we give you to we allow you to have your own bias for example in the system i don't think so i think you should be unbiased when you're getting and information and you should get everything about that piece of information and i always give them an example like when clove it happens the news was out there like guardian was reporting on it telegraph was reporting on it in 2019 i remember november i was reading it was out wasn't it understanding the whole width of the debate is really important yes so but nevertheless some clients will not accept certain streams of information yes sometimes some clients will ask us to remove certain things like for example some clients don't even want to see the data that we've collected they just want to see their own data so they they want to have that echo chamber of like information like um especially consumer goods companies they say oh we don't want to see your data we just want to see our own data but actually it's a bit of a bizarre thing because as a consumer goods company you should care about what the consumer of you yeah absolutely the wider view yes interesting yeah so in the going forward how do you think you've managed that because that seems to be a potential conflict you're you're setting up to provide people this all seeing i almost and some people are choosing to almost be blind but you can't help them to the full extent if they adopt that to your choice that's a bit of an irony really um have you any thoughts about that going forward sorry we said that it's okay if a client doesn't want to use all the data that is available to them so if they just want to use their own data and just optimize their search on their own data that's fine i think a lot of companies have got more data that exists online basically so i was reading i think um JP Morgan has got four trillion data or something like that and crazy amount of data um or something more than that i can't remember the exact number but it was very like in the that's external internal to them you're saying yes they've got enough that they've got more than enough to cope with yes and they can't access a lot of companies can't access their data like i was talking to um a company consumer goods again and they they have issues with linking data from different databases so there's like so many different databases and linking everything together is a huge huge task and i think they're struggling with that and without that linked data you're not going to get good outputs when you're using these machine learning algorithms so um a lot of the companies are struggling with that and that's where we are seeing like our usp playing really well and um we're going and saying it's okay if you don't want to use our data but this the date the application comes with its own default data and we can shut it down for you and you can just access your own data basically and that's also a possibility so it becomes a search engine but on internal data based on their own yes that makes a lot of sense to me yeah absolutely so your team is expanding you've gone from i know one or two of you how many of people are you are you now we are about 17 people with advisors counted and counted them okay and what challenges have you met with are you did mention the sales force the sales team recruitment has it brought with you it brought any other challenges because it's quite rapid growth that's only in a year or two isn't it that's correct um no i think bringing people and and on board is a styles process basically you've got to sell to them how how good it will be to work with you basically you're always selling you're selling in three different ways when you are a founder you're selling to your investors you're selling to your clients and then you're selling to your employees because your employees are also they need to be sold to as well it's not just not just like all like they would see the value and they would be onboarded so i spoke to a few people a few founders and when they first started their company they were quite rude actually to their employees and then they realized that that's not the way to and you're so focused sometimes when you start off you can't see any any any any width you're just you just want to get to where you want to go yeah i can understand that i focus that predominantly like a marathon like it's a long game we are together for a long time so let's let's make it a happy so that's what i do with my employees i try to make sure that they're happy they are taken care of and they're doing like if they need a resource i try to get that for them so that they can get on with their job and not like that yeah because that's so important without that you're not going to have a healthy environment you're not going to make progress like the progress we make clients gets on a call with us every now and then and they say it's actually shocking how much progress you've made from two months ago because it means that the like we are continuously getting feedback and we adjust but that but that's because people in the company are feeling relaxed and they're not feeling stressed and they are excited about what we are building and on top of that i have created this culture where initially i said to people if you don't know something you are more than welcome to tell people that you don't know it it's okay like yes yes you want to frown upon you nobody's going to say bad job you know you're fired like this is not a lot of companies so we will allow you to learn and i think that's whole you can learn as you progress in the job and that's very exciting for people because and they don't want to stagnate in their role no especially in your field where it's changing almost daily they want to grow with it so how would you sum up your sales pitch to employ potential employees then i always say that um i actually got a very good advice once with um from one of the founders and he said when you hire people you should sit them down and say this might not be the end of your journey this might be at the beginning of your journey or maybe we are the middle path but the aim is so that we both can progress so that's what we always say we are here to progress together and there might be things that i don't know that i would learn in the process of um working with you and you would teach me and there are things that you don't know that i would teach you so it's it's a it's like an educational place rather than like a research focus place rather than just a company and that's exactly what we say to people and we always say that we say research at the forefront and then everything else at the back and i push my software engineers because for example full stack engineers they just like to build they don't actually like to research so actually they have it's really hard to get them into the research i can't stop doing yeah i've got to just carry on building yes but they've gotten into that mindset and they love it they're they're like oh this is challenging and like you know they okay okay so you can get them to take a day off from from building and then that actually do some proper research yeah exactly and they need to do that because otherwise like one day of like doing proper research is not going to delay the products it's going to actually advance the products because it's yes research properly then they would have solutions that they can use to optimize to build faster to do something more exciting whereas like if they don't take that day then they're not going to be able to progress basically i isn't that an Einstein quote or something if asked he says if i was asked to solve a problem i'd spend 55 if i had an hour i'd spend 55 minutes doing the research and five minutes solving it and i think it's that's like that's what you're trying to embed in in deep search i love that we always say that data is the most important thing so i've always said that to everyone like without data we are nothing because in machine learning field you do like 70 to 80 percent of the work you do is data engineering and then the 20 percent is the algorithms basically and it's okay to bear that in mind because without that you're not going to build great algorithms so data is so important it needs to be accurate it needs to be of good quality and without that we're not going to do it is rubbish in rubbish out isn't it you you can't it's the same rule it's that's been the rule forever isn't it exactly so with that in mind you've recruited this team it even in such a short time you must have had to have some quite difficult conversations with people who have come on board but maybe not lived to those those values of integrity and or interest in research and so how have you managed those situations i think the churn in the company is quite low we've had in the beginning it was a bit of a chaos actually because that was the beginning of the company and i think i had one difficult chat with one person because i just knew that they could they could pull them pull like you know push themselves a bit more yeah they were just not doing and it was quite disappointing to be honest and i give them a lot of chance but then they saw it as instead of i one thing i learned when i was younger was that when you have a coach or you have a teacher that is riding you hard is because they love you they want you to believe in you you know so it's it's really important to know that and sometimes i am hard with some of my employees but because i just know their potential and i think that that's knowing their potential if i didn't believe in them i wouldn't spend a second with them but if yes it was supporting it spends like time with them and i think a lot of people don't know that they don't see that as a good thing and i learned to add the hardware as well because when i was doing swimming like when i was little i used to competitions like swimming competitions and sports and then i got into piano and lots of different things but with swimming like your teacher basically your coach will walk alongside the swimming pool and would tell you go on you can do it you can do it yes that's right yes keep going keep going but if they didn't believe you they would not they would not tell you these things they would just say like oh you're not going to go far so i'm just going to take it easy yeah well am i going to push myself and yeah because it's a lot of energy they're putting in yes exactly but when they believe in you they push you they push you to your limits they're like oh you're not doing our hands right like i've taught you this like last week why is it not working you know and so like this sort of repetition is really important the repetitive information coming through is important but also dedicating and being dedicated to a task and to progress yes but not everybody gets that they they take it as criticism and they take it in a more negative way yeah i see that more in gen z's like the new generation they tend to have different life work balance yes i've heard this yes but um without skill sets so i understand life work balance when you have skill sets but when you don't have a skill set you need to like put in some hours maybe afterwards to read some sort of like tech that you are using because otherwise within work so when are you gonna have the time to do it right yes catch up catch up on things and i remember there was one one person who joined the company initially and they refused to join um after after six o'clock meetings with one of the guys who was teaching them how to use cloud infrastructure and then and then she was given a project and she wasn't able to do so and at that moment i had to sit down with her and say like the reason you can't do is because you didn't join those meetings if you had joined those meetings then you would be in the same level as the other people in the firm you go through after that some time and um and she didn't like it so and she thought i was i was too harsh and i was embarrassing her that was one more other thing that i was told and i said no that's not the case i'm just trying to make you do better like you know and yeah that just didn't work well so that ended quite rapidly that was in the beginning but now i work differently i'm more patient with people so if they can't do i'm like okay well it's okay i'll give you another few days i'm sorry let's try and find a different way and a bit more time but and then i'll have people to spend time with them so i'm like hey like for example um as a dean like you know how to work on this can you please go and sit down with this person and teach them because they need to like come up to the level of skill set that they can do on their own so now we teach people in the company like when they struggle we are like okay don't hesitate call us we will teach you if you can't get to the solution but initially it was more of a every person was on their own so they were all learning and we were all learning together yeah of course and it required some push from the individuals and i think when those individuals didn't have that push it was really irritating because yeah yeah if they're not self motivated then it's really hard to sort of gee them up isn't it to get them to come yeah absolutely so it sounds like you're working quite long hours then how do you manage your time to be honest i'm obsessed with my company so it's it's something i'm obsessed with so i actually don't have a break from it ever like i'm always working on it and there are times when i just close my laptop but journey happens very late at night so like around 10 o'clock there's no laptop time so like close the laptop close the lid no more work but just no more devices either just sit down and silence basically for a bit and i think that is really important because you do need that kind of a break but not so much like long hours of break i don't really like that i find i find it you take weekends off i work Saturdays Sundays i do take off but it's when i say i take off is i'm actually doing reading on the industry and how things are progressing but that is my time i'm sure leisure yes and i like it i enjoy it um if i didn't enjoy it then it would have been there would be be harder but do you think it's sustainable over a longer period yeah i think so i i think um as we grow i think some of the roles that i'm doing probably going to be replaced by other people so other people will be doing them and so i would be left with more of a strategy focused and resource allocation kind of role and initially i think you you do a lot of things but as the company grows your responsibilities changes but then your work becomes maybe even heavier so that's what i think because then there are other bigger problems that you need to solve because at the moment the problems are small but they're continuous whereas like i think then the problems will be bigger but less less frequent so that's what i think and let's take up the same amount of time will be taken out probably probably true actually yeah that's very wise observation so if you were looking back um well i don't know three or four years what would be you seem to have learned a lot in such a short time what would you what tips would you pass on to somebody following in your footsteps i would say just be patient um put your health first health is so important accidentally i got into this thing called 75 heart which is okay it's not i don't agree with all of the things that they mentioned but one thing that i loved about it was discipline it teaches you discipline um so you do two 45 minutes exercises one in the morning and one in the evening when you're end of the day and you can choose what you want to do but one has to be outdoors so but that means that you need to plan your day so that you have the time to do those exercises totally and also eat well and drink water because yes water do that then you're not going to be able to achieve these two 45 minutes exercises every day so that really did help like that discipline creation through that's okay so you're not on the laptop all the time you are doing i would um like tell myself to be more patient or tell the person to be more patient okay because it's an marathon not a sprint so you're not it's a marathon i think that's a lovely piece of advice yes yes and you're not going to compete with a unicorn company you are early stage you should always bear that in mind when you're starting is at early stage there are like so much you can do with your resources you're not a 300 plus company no yeah you can't do so much in a day and with the money you've got yeah absolutely and you can't you can't burn out your team members because they need you need them and they need you so you need to be patients and like take it slowly as as they go it's by but the speed is really important one of the things that my manager taught me as uh my app to ask a measurement that i worked at was the speed of change because it's fast then people don't like it they they become resistant towards they push back and or they burn out right because they they just don't know they need to do something and then they burn out however if it is like at a good pace i think then you have people understanding what changes are appearing why they're appearing what needs to happen to make them happen and and that's really important understanding of why change is needed is really important i think that needs to happen at a good pace not super fast not super slow because if it's slow you're going to lose your resources yeah yeah you got yes absolutely go through to your resources and not have changed and and people particularly in your industry want to see change they do want to see the evolution don't they yeah well marym that's a fantastic end point to close on so we've got patience which is a virtue exercise and health and try to weigh out and have a steady pace of change yes absolutely well that that's been enthralling thank you i'm sure everybody watching this is going to sort it reappraise their ideas about technology and AI and i see how to start a business so marym toshizi thank you very much indeed i hope i have been helpful to somebody i'm sure you have i absolutely convinced it thank you very much for having me my pleasure take care
