DevReal: Nikita Ivanov Interview
Recording: DevReal: Nikita Ivanov Interview
hello everybody I'm Alexi kabro the founder and organizer of bay eii the longest running deepest technical a up in the world here our in location Cloud flare with Nikita ianov founder and investor on human ton of which I'm also an adviser tell us Nikita what does heatron do well heatron basically develops sort of a new concept of a software Define Workforce we call the hireable AI so I be able to build our workers and most importantly hire them pretty much the same way as you hire people in your company and uh I know you for at least 10 years you presented a text by the Bay Y right with data linga yeah right the previous startup which did query based on precisely DEC composing yeah called domistic NLP yeah that's right and then you did a uh AP LP craft right the open source package business and what I noticed you are probably one of the few coding cosos right like I you share the scolar code base for that one and like every day commit commit commit commit and now you're coding for the heatron so how do you do that well it's it's never me alone there's always a team yep but I always keep my hands sort of a dirty yeah so uh it's just where I am um I'm a I'm a playing coach and so that's how it rolls and so then you you used angular for for that Ling I remember Yeah well yeah most of the like couple startups I'm involved right now both humron and Alpha prism it's stack is very similar it's all nextjs in the front end which is a server side react if you will and it's obviously python in the back end so I I noticed some people doing typescript that's unique you know most of the back ends today are python for obvious reasons you know most of the Lang chain stuff is python first M so you kind of uh we still use Scala somewhere I know the team is using Scola but sort of a desire guys for the whole back end development is a python right now for obvious reasons so do you use any EOP pilots on you C of course yeah how can not you know everybody Co away with two or three co-pilots in the idees but you know I always keep saying the co-pilots you know help a lot to sort of an entro level Junior Engineers for this for the senior Engineers for professionals who have been there for my they basically give you maybe five maybe 10% improvements here and there but it's it's not something that dramatically changes anything for you so you know software industry Inside Out basically right you were doing it and so I think you were a field CTO for data Stakes yeah right so like you really see the whole Spectrum so I think it's I have to ask you like where do you see the software industri is going you I'm actually one of those who have pretty contrarian view I don't think it goes anywhere dramatically different uh just like if you look at what people would do and you know 40 years ago it dramatically changed it will change again but I don't think Engineers go anywhere the the job the engineers do the productivity they'll get much much better but fundamentally you know yours and my casid will still be right and code if they want to and they'll be doing a different level different abstractions with a different productivity and efficiency but we're still going to be writing code so hum still will be in charge yeah listen my presentation today I I'll basically make a claim that in 20 years from now 90% of employees will be still human right but you want to sell horrible AI right the agents and they have names what made you kind of give them names give them personalities right because you want to basically make sure that jail would exist in team my my contention is that for a very long time in the future majority of employees will be humans yeah that's therefore it's ai's job to adapt to the how we work you know how do we work every day we chat over you know Slack Zoom we send email we send SMS we have names you know we can drop on a video call and whatnot so the AI has to be able to do the same things mhm uh so they can Jael nicely but obviously provide all this AI driven benefits and you also s them times owns right so humans will not have to wake up in the middle of the night exactly those little things yeah you know we take for granted but when you think about how humans work it's it's obviously you know there's a lot of those details that go in our interactions that if you don't Implement them on the ey side it's always going to look awkward right I think this is all smart and I think the strategy uh is basically to make this system compatible with HR Solutions as they exist with people I mean the way I'm looking at is that I want to get to the point where you can hire AI you can work with AI dayto day and if you don't squid you don't really know who is this in front of you MH you're probably going to see some some some badge in ey and something MH for probably otory reasons right but if you don't really squ it should not be really seeing a difference okay um so and you basically I think one of the early startups we should start to think what is the infrastructure for highi what does it take right to like get this agent so how do you see this Evol what needs to change in the existing Society uh do you see kind of more interoperation like apis for inter like DHR agenes for instance have to develop new apis can you integrate the workday already or do you need you know Human Side of employment to develop new I think the later one I think there's a still certain level of passive rejection and I'll be talking about this by saying look we're not hiring AI we're hiring people it's kind of immoral and I and there is a certain Truth for that but I always keep reminding people like what happens to horse scage drivers what happened to typists right from 40 50 years ago was it immoral that this you know industries were completely wiped out no it we just moved on and you know there's a lot of Automation and that's pretty much the same way yeah they got you know repurposed as uh locomotive drivers and they're still there but you know yeah it's not it's no longer this highend profession that it was liking about 100 years ago so uh you basically like I going back to the Ling NLP like you develop a whole bunch of skills and experience with NLP right so what I know at some point we thought okay statistical AI threw it all out the window now we found the statistical AI has L lamps which lie and we need some human knowledge and Rule based systems to verify so for instance knowledge graphs are back so do you see kind of right like the worth of experience people Acquired and software engineering right and system design what You' seen databases do you think like the like all of that will still play the role and where should people who build startups now right like what kind of skills do you think they should have uh right what like what kind of hard software skills I still you know maintain that I'm you know I'm pretty contrarian in this I mean I'm not subscribed to this news I guys that you know everything's going to change I think you know as like almost everybody saying we absolutely going to see a onean you know uh unicorn companies that's become impossible with all the tooling you can be a one person and really develop a very large business yep that's technically become impossible is it going to be often enough I highly doubt but like 20 years ago that was practically impossible in Tech today it's becoming technically feasible so we're going to see that but again as a Founder technical skills number one number two number three and then the softare skills obviously come into play later on but again a lot of those can be supplemented with you know whatever the agents or something that human developing something else but technical skills still a rule of the game right you have to develop a product without product there's just nothing else here to that ha technical skills and so what's next for heatron what's the plan for the well the team is working you know we have you know team has a pretty you know large road map and uh I think it's about 9 to 12 months of development so there going to me some fundraising in between but we'll see what happens all right thanks for sharing looking forward to your talk thanks you out you for