Apple Podcasts
Spotify
Castro
RSS Feed
Breaker
PocketCasts
YouTube Channel
The player is loading ...
January 25, 2024
Harnessing Generative AI to accelerate Software Development with Bito | Episode #77
This episode features a discussion with Amar Goel, co-founder and CEO of Bito, a company revolutionizing the software development process through AI-driven tools. Focusing on increasing developer productivity and code quality, Bito integrates ChatGPT into IDEs and CLIs, streamlining the coding process. Amar and I discuss Bito's role in enhancing software development, its …
This episode features a discussion with Amar Goel, co-founder and CEO of Bito, a company revolutionizing the software development process through AI-driven tools. Focusing on increasing developer productivity and code quality, Bito integrates ChatGPT into IDEs and CLIs, streamlining the coding process. Amar and I discuss Bito's role in enhancing software development, its key features, and the future trajectory of AI in coding. The conversation also touches on job security in the evolving landscape of AI and software development.
This episode features a discussion with Amar Goel, co-founder and CEO of Bito, a company revolutionizing the software development process through AI-driven tools. Focusing on increasing developer productivity and code quality, Bito integrates ChatGPT into IDEs and CLIs, streamlining the coding process. Amar and I discuss Bito's role in enhancing software development, its key features, and the future trajectory of AI in coding. The conversation also touches on job security in the evolving landscape of AI and software development.
Founded by seasoned entrepreneurs and technologists in the heart of Silicon Valley in 2020, Bito has emerged as a pioneering AI-driven software development company.
Topics Covered:
- Amar Goel's background and journey to founding Bito.
- The inception of Bito and its mission to transform software development with Gen AI.
- Bito's integration with IDEs for enhanced coding efficiency.
- How Bito AI assists in understanding, writing, testing, and documenting code.
- The significance of Bito for non-developers, including product managers.
- Discussion on Bito's AI models and their selection process for specific tasks.
- The concept and future development of AI agents in Bito for automating coding workflows.
- Exploring the impact of generative AI on job security and the software industry.
Personal insights and plans for Bito in 2024, focusing on developing AI agents for code review and unit testing.
☑️ Web: Bito AI Official Website
☑️ Crunchbase: Bito AI Crunchbase Profile
☑️ Support the Channel by buying a coffee? - https://ko-fi.com/gtwgt
☑️ Technology and Topics Mentioned: Bito, Generative AI, ChatGPT, IDE Integration, CLI Tools, Software Development Efficiency, AI-Powered Coding, Code Quality Enhancement, Job Security in AI Era, DevOps, AI Code Completions, AI Agents, Vector Database, AI Model Selection, Automation in Coding.
☑️ Interested in being on #GTwGT? Contact via Twitter @GTwGTPodcast or visit https://www.gtwgt.com ☑️ Subscribe to YouTube: https://www.youtube.com/@GTwGTPodcast?sub_confirmation=1
☑️ Subscribe to Spotify: https://open.spotify.com/show/5Y1Fgl4DgGpFd5Z4dHulVX
- Web - https://gtwgt.com
- Twitter - https://twitter.com/GTwGTPodcast
- Apple Podcasts - https://podcasts.apple.com/us/podcast/id1519439787?mt=2&ls=1
☑️ Music: https://www.bensound.com
Transcript
RAW TRANSCRIPT
I think like what this is going to result in is probably like almost more software developers or at least software developers who are like more and more productive you know and producing output you know hello and welcome to episode 77 of great things with great Tech the podcast highlighting companies doing great things with great technology I'm your host Anthony sper and in this episode we're talking to a company that positions itself as an indispens tool for modern developers aiming to revolutionize how code is written and
managed with a focused on AI powered efficiency and a strong user base it provides a significant boost to developers productivity and code quality that company's Bido and I'm talking to the co-founder and CEO of Bido Amma go how you doing hey thanks for having me excited to be here and uh yeah love your love your podcast excellent thank you very much and hopefully we're going to even make it even better when we just kick off a 2024 talking about Ai and something that you know I think everyone wants to hear about especially when it
comes to the devop world to coding and how generative AI can help and you guys are at the for Forefront of that but just before we get into all that just as a reminder if you're into great things with great Tech would like to feature in future episodes please click on the link on the show notes head to GT wgt.
com hit like And subscribe go to YouTube as well we're on there all past episodes also on all podcast platforms the Google Apple Spotify whatever it is all hosted and distributed by Spotify podcast all right em with all that let's talk a little bit about Bido actually let's let's talk deeply into Bido because it's a fascinating company um representative of where we are in 2024 with AI but firstly give us a little bit of background about yourself and how you came to co-found the company yeah so um I'm I live in the Bay
area and I'm originally from the bay area but sort of grew up kind of traveling around I I lived in New York for a while my wife and I we actually moved to Mumbai and lived or Bombay depending on what you say it's funny technically the terms mumai but I noticed everyone who lives there says Bombay so yeah I'm actually go I'm going there in a couple of months so I'll let you know what what people say oh nice cool well have a great have a great time let me know if you need any uh you know tips or anything um but any case yeah so
um moved back to the Bay Area in 2017 like I said which is where I'm from and uh I've been an entrepreneur for you know quite a long time I started an e-commerce company when I was in college in the late 90s internet boom and then um you know spent some time in Corporate America at McKenzie and Microsoft and then started a company ptic um actually um one of my co-founders at Bido was one of my co-founders at ptic and we were very fortunate that company grew significantly and went public in 2020 um and uh was sort of a software
company focused on digital advertising and um an in then myself my co-founder and our third co-founder mug we all kind of really thought it would be great to build a new company and so we started Bido and we've always been focused on developers but we kind of ended up migrating our our focus a little bit over over time so how does that I'm fascinated in that you've wrapped up what seems like a pretty interesting career in about in about a minute and a half right up to up to bid but um which is the way but you know
what before we get into it what what dragged you over to go back to India um at the time yeah yeah um so it was you know it was in 200 um six and uh my wife and I we were living in New York she's from New York um and both of us I was 30 she was 27 we just thought oh would be fun to go you know life in New York was great and we're having a good time but we thought it would be fun to go live abroad go live somewhere else and so we thought about moving to Europe and then we you know she's also of Indian descent
she's from New York but her parents you know were Indian so we thought oh we maybe India would be an interesting place to live and so we just kind of started thinking about London or somewhere else in Europe and then we thought you know there's a lot going on in India and you know we're Indian like it would be great to go live there so we moved there in 2006 and we ended up we thought we were going to live there 3 to five years ended up being there for 10 years and um it's a little longer than
we planned both of our boys were B born there um crazy enough which is something I never would have thought and we moved back um in 2016 to the US yeah and and obviously you you would have picked up a lot of you know India is obviously well renowned for being strong in in in the development sector right so I'm guessing you were working pretty hard there and picking up a lot of the skills to be able to lead you to you know the Microsoft and and then the founding of the first company that went public right
so you've obviously HED yourself from that trip as well yeah you know it was interesting I mean I actually studied computer science in in in college but I wasn't um you know and I I did an internship as a software engineer at Netscape um the or wow there you go now we're getting into it that's cool yeah the The Young Ones probably would they would be like what's that you know but that was that was like the hottest company on the planet you know back in absolutely the way back years of 199 six and 97 right um uh I have a
really funny anecdote if I don't know if I should say it but funny so I was an intern at Netscape and um I um Jim barale was the CEO at the time and uh Mark andrees was the uh you know one of the co-founders he was the CTO and so Mark Jim barkell Held a like an intern summer cocktail or something like like that at his house so you know there was probably like 50 inters or something we were all in his backyard and everyone was just chatting away Mark andreon was there and Jim and you know like other people in the
insurance and then these kind of two dogs were there and then they they jump into the pool um and then um you know it's kind of like they jump in but it's not like a big deal and then all of a sudden Mark andrion goes oh bleep those are my dogs they can't swim they're just kind of you know like sinking to the bottom of the pool and so he just like fullon jumps into the pool like full-on clothes you know his like whatever cell phones we had back then wallet whatever and just jumps in the pool and goes to the bottom drags his
drags his dogs back out luckily they were okay and everyone was okay or whatever but it was kind of yeah it's an interesting story right like those dogs obviously wanted some attention and you know obviously from his point of you back back then where he's got to now in the industry like absolute legendary status with what he does today and all that investing and whatnot that's pretty crazy that's an awesome experience for you as well so and then and then getting into why did you found Bido like what was what was
the origin of the company what what was the problem statement and maybe actually before that just explain a little bit more about what Bido is and what it does like I sometimes forget to ask that question but you know let's start there and then go to the actual founding reason for it yeah yeah so you know Bido um you know our kind of belief and what we're focused on is that we really think gen is going to transform how software development software development happens and how software developers work um and
we think that over the next five to 10 years um software developers are going to become you know 10 to 20 times more productive um and we're going to build better code and you know like it's it's really going to be very transformative it isn't going to be like oh yeah like it helps me a little bit you know um and so you know we think it's kind of like the next next it's here today but the next big kind of abstraction you know layer for um software developers and sort of how software is you know made I mean
sorry to go back to like mark andrees for a second I guess he is a Legend um you know his whole essay right about software is eating the world like yep kind of feels like we haven't seen anything yet actually you know like scary the amount of software that will be possible to be built now and tested and be robust and be you know be high quality the I mean one way to think about it could be if you think about your like in software right you like tracking your requirements you you have a backlog of like all these features you want to
get to but you never got to them because you're like well I only have so many software engineers and they can only do so much right and so this thing just isn't worth us building at this point because only 42 people are going to use it and this other feature 10,000 people are going to use but now with generative AI you could be like actually we can build that really quickly and 42 people is enough to make it worthwhile you know um and so that's kind of you know one like model I think about is like it's not just about
writing code but the entire that means that you need to test more code that means you need to do more code reviews that needs that means you need to do you know like you need to build more software so it's going to put it's going to change kind of how you know everything everything kind of happens and so we think that like the tooling that's going to be required for software developers to live in this new world you know is is is going to really change and we also think you know I think that like
we're at gbt 4 today we're at anthropics Cloud 2 we're at Gemini Pro or supposedly Ultra you know whenever it comes out I mean there's going to be an ultra squared and Ultra cubed and there's going to be gpt7 and so all these models are also improving you know at a significant rate and so I think all of these things kind of combined to make like a perfect storm that will you know really help move the whole art forward yeah so in terms of how bit plays into that um what is it and and obviously
you're working with Ides um you know development environments to help coders be efficient you talked about code reviews you know QA all that kind of jazz so what is Bido itself like how how does how does a coder or devop guy who's looking at it today actually start to interact with it and use the product yeah yeah so we we launched the product in December of 22 December of 2022 and we started out as a chat product in your IDE so kind of like we sort of say chat GPT in your in your IDE right so we work with Visual Studio code
we work with all jet brains idees and so you could use it to write code you could use it to you know write test cases you could ask it to explain code that's actually the number one use case 21% of the time code can you explain this code for me um which is which is kind of interesting and actually almost 10% of the time people are saying can use it to document this code so 30% of the time people are using it to understand or explain to others what's going on right which I found very interesting which is you know not
something you would have you know would have thought everyone always focuses on like writing code but yeah think about most developers like a lot of times you're writing code related to something else that exists and you're like I don't really understand what the heck this does or how it works you know can I expand that as well and like from my perspective my background working on the other side like I I started at University at doing computer science but I quickly realized that I didn't want to
code for life um and so I went into the infrastructure hosting world but I still needed to understand the code I still needed to see the code that was being written that was going on my servers because if it was if it was bad code it was going to impact the way that we did the Ops right so I've always been interested in that so and that's that's still today it's a lot of people do that you know it's not just the developers that are writing it's people that need to understand what's going on on so if
there's a way that we can interpret and understand code easier like that would have saved tons of my time back in the day right um and also what it does it would have allowed me and allows people potentially to get more up to speed and quicker and learn you know what certain code segments are doing um through exp explanation and also documentation so to your point I think this is more than just a a writer a coders tool because I see this being valuable for the other side of the obars yeah I mean product managers also you
know we hear from product managers I mean they're not like our big user base but people are like oh now I can kind of understand what this commit that an engineer made is doing or you know hey I'm trying to figure out how something works in our codebase so I can I'd love to have it like something that can explain me that code or I can interrogate the code base or ask questions or whatever you know so um so yeah so going going back to kind of what what BTO does so yeah so we started out with this you know chat product um and
we also launch the code what we call AI code completion so as you're typing a line of code it you know completes it or you give us a comment in your IDE and it'll write a whole block of code and one of the big things that we launched last year um in the summer is um we we index your codebase and so U we understand your code and we're able to let you you know perform AI on that right so you can actually say hey can you change this function or you know I want to um you know I want to ask a question about my codebase um
or I want to like make a change to this database can you tell me all the changes I need to update so it can do all of that because it can understand all the parts of your codebase that are you know um we use we use a vector database to do that so we build like embeddings like these large scale vectors and then were're able to use that to kind of understand your your codebase um and that's actually one of the areas that we're focused on for the future is really under understanding your code really really well and it turns out it's
like a I think it's a very important issue but it's also kind of difficult and so a lot of companies we've seen haven't really gone deeply into that they've kind of stayed out of it but we actually think it's you know one of our differentiators but more more importantly to help developers I think you can't just ignore all the code that they have that's been written you know especially I think it's yeah in the Enterprise use case I think it's important because you know knowing a lot
of developers it's I I I've called I've called like copy copy devs so you know devs that basically take snipp of a code from here a snippet a code from there put it together kind of mash it up and kind of make it their own and I think there's a lot of that that happens there's there's very few developers out there that kind of are writing pure code from you know the really brainy ones right that are just suran at it so I feel like most coders kind of Leverage other people's code to understand learn
and adapt that to to create what they need to create and wrong with that I'm not saying all coders are you know copies or whatever but that's just the reality of it right so to your point if you got a system that can effectively validate and make sure that What They're copying is legit Works in company policy in scope of project um all that kind of jazz I think that's really valuable in itself right yeah no 100% like I think it's super you know valuable and and and important um I mean most of the time in
a in a in a in an employment context like in a job like developers are not writing new code you know they're like or they're taking some existing code and then changing it they might be adding to it or whatever but they're they're taking something they're not just like oh I'm just starting from new and writing a whole new module or class or you know component you know for us so um just coming back to BTO for one second so I think one of the big Focus areas for us now we so one of the things we've seen
over the last year which has been really interesting is um that we think that all these tools out there like code completions and all that they're really interesting and they're quite useful for developers but we think that like a lot more is possible and and really that we think is going to come from Agents um and so we're spending a lot of energy kind of focus developing agents that we think can really help you know achieve that goal of helping developers become 10 20x more productive right so we gone from 10x to 20 to a th like yeah
that's that's that acceleration that you're talking about because there's definite that what we see in the industry in general is that this this um this technology is self-perpetuating and just creates a flywheel of innovation and Technical advancement right that's just what it does and that's kind of been the prediction for AI and this sort of technology for a number of years and like you said before it was more mainstream it was kind of always the realm of the data scientist the ml guy
and the company yeah yeah but now it's kind of at people's fingertips and to that point the acceleration is going to get you know quicker and quicker and it's a little bit scary to that point you know you mentioned software in the world um is a great book that I've read life 3.
0 by Max tegmark he talks about um this um Omega project and it's effectively an AI that was built by the company Special Ops part of the company and effectively they started it and it became super self-aware really quickly by it just effectively got smarter by rewriting itself and rewriting writing itself and eventually basically long story short went out and consumed the world and in a good way made the world a better place it was one of those good stories it wasn't the bad stories of AI and robots and whatnot but that initial learning and the self embedment is I
think where we're at today with a lot of coders in in this world right where they are getting better it's going to be you know more more acceleration and that's where this is really exciting it's scary but exciting at the same time um we'll talk about agents in a second I just wanted to get the the listeners to understand and you know you plug into the IDE and then it's is it then talking to a a service that you're running somewhere you mentioned the vector database is that is it is a SAS platform
that's running that interprets the code that's running on the local IDE and then kind of does its thing or how does the interaction work between the IDE and the actual smarts of your platform yeah yeah no great question um the answer turns out to be a little bit complicated because um developers and organization are very concerned about their code security and what I was getting yeah access to their their code and so um today in our in our IDE the way it works is that um when we actually index your codebase we actually deploy like very
seamlessly you don't really notice it as a user but we deploy a vector database on your machine and we we uh all the vectors the embeddings are stored in there so everything is local on your machine so we don't store any of your code in our system in our Cloud we don't use any of your data for model training none of our AI Partners we work with a bunch of the big kind of public llms um like open AI anthropic Google um but we none of them you know do any model training off of your off of the consumer's re the user's
requests but anyway so a request you make we might like access that uh Vector database on your machine grab the relevant context send comes into our you know service running in our cloud and then we send the request to an llm um and then get back the result maybe format it and send it send it back to your machine oh yeah okay it so it's kind of like a multi-tiered kind of architecture but we we don't um like I said store anything in our we store a little bit of metadata like to understand for example oh you were
trying to ask us to explain code or you know some logging about where this request came from but we don't store any specifics about the request that's really interesting and a good point right because I think obviously privacy is so huge a topic when it comes to Ai and generative AI in general what IP is there I mean what what is the protection of of the of the data that's in the model where's it been sourced from um so the fact that you've thought about that from the start you kind of talking about
you know that PR it's almost like a private a private model to a certain extent the data is kept privately through the vector database and the smarts to put through those those models up there it it probably makes um people's minds rest easier than say if they're pumping it directly into a open AI platform or whatever the other ones are and where the data is intrinsically just pumped into it um that's really good so here's my thing so I I I've I told you earlier that I've been building
this bit of a side funky project you know using chat gbt and it's been really fun and interesting but frustrating at the same time right because what I'll do is I've got I've got Visual Studio open and then what I'll do is I'll tell it something I I want why draw me this I feel like this should be a feature gives me the code I translate it into Visual Studio see if it works and then I'm copying pasting copy and pasting so sound sounds like I should use B right absolutely and I'm probably G to give it
a go right but what what is the difference in that so effectively what you've done here is you've made that seamless right that's kind of what I'm trying to get at so because there is a ton of frustration you know you get you kind of get used to it to a certain extent and you it but there's a way right and this is what you guys are basically ping here with bid yeah yeah no exactly so you know one is like by having it built in your IDE right it's just really easy to kind of immediately
copy the code over you can easily iterate on it um we have a lot of templates that are built to like you know document the code you know write test cases right so you just literally select code there's a prompt that's already been created so you just one click and you write a whole test case versus like copying in the code typing something in um but I think maybe even more interesting is the fact that you know we understand your codebase right so if you think about that I mean you don't cat gbd doesn't understand
anything about your you're constantly re reiterating what it's doing and then sometimes the window will will time out so you got to start again then you got to spend the time to go hey this is what we're doing you were doing it before but you don't remember it but I'm going to tell you again what you're doing so there's a lot of relearning and reinforcement that needs to to go into chat jbt as an example which is quite frustrating and it's quite inefficient yeah so you can say like hey
you know I want to update my authentication you know methods to to use Google ooth to you know whatever let's say use that right and then so it will go and find all the authentication methods in your system you know in your codebase and then you know pass that to the llm with your instructions and rewrite all that you know and so then you get all that output and so I mean I'm not saying it's like perfect every time or anything like that but you know you get 80 90 100% of the way there in about 10 seconds right versus like
trying to hunt through your code and figure out all the different things and um so I think by understanding your you know your code base we think that's like a really you know kind of powerful thing the other thing we have is a CLI so we have like a um you can access BTO through your C and so we see people using that for Automation and that's kind of where you know where some of our real thinking around agent started to come from is like we've seen Dev teams like QA teams for example say oh we're
going to run take the code base you know or take the you know changes that somebody's made and ask it to generate you know test cases actually one of the things that um one uh QA team showed us is that they they take a an set of code and they'll ask BTO like show me ways to break this code um and then they'll get the respon responses from uh you know from Bido from an llm and then they will take that and say now write me unit tests you know that will test these issues right and so then they they s of
are hardening you know using testing as a way to like Harden their code base it's really interesting I was thinking about in this particular I mean it's I've built it as a kind of semi platform it's got authentication but you know and it's got elements of a database where as I go long I think I'll actually got to add a new a new table or a new entry into the database and then what I've got to do is got to remember myself um again hacker here right but I've got to think to myself okay every part that I've
referenced the database entries when I when I get it out to the website I've got to actually go back and now um you know add the new the new tables and new new data points and where is it all you know and then so if it's not in my mind and if I'm not me me if if it's still small it's okay when it gets bigger it's like it's crazy so this is where this can help as well so even for the demo the demo video that we have on our homepage actually that was one of the kind of use cases we showed which was
like Hey I want to add this column into this database like into this table but like can you show me everywhere I need to update my code base update it okay that wasn't even planned we didn't even plan that but that's that's that's worked out well right so yeah so I'm thinking how valuable this is and that this is just me as a side project I'm now people that are actually do this for a living so to speak you know this this is this is huge right so how does it decide what model to use how do you
dictate that is is it I no there a couple of pricing plans there that differentiate some level of access but in general if someone's doing it through the IDE do you just use the best model that's available at the time for the question or is there something that that goes in in the background there yeah so we um we have a couple of different things so one so we have like the product's free you know you can start out free and use it for free and that gives you access to what we call B basic AI models like GPT 3.5 turbo and um uh
Google um some of their um more basic models like um uh CH bison is kind of one of their um you know core kind of chat models anyways and then we have an advanced we call Advanced AI models which is GPT 4 and um anthropics CLA 2 um and so those are accessible in our paid plan which is like $15 a month um and so but the other thing that we do then even within basic and advanced is we try to find the best fit model you know for what you're trying to do so like a simple example of that would be uh traditionally gp4 you know only had like
a 16k you know context limit now G they had 32k but actually almost nobody ever really had access to it they've announced like uh you know gp4 turbo has a 100K context limit which is you know a lot better and great um but it's not like really broadly available today and thropic Claud do has a 200,000k contact window which you know just to provide some you know that's about a 250 pages so you know pretty much any code file can fit you know in that or you know even some small repos could fit in that
and so um you know basically based on the the context of the question and what you're asking we will like find the best model that we think will you know for example that expanding context window you know can affect us so if you're just giving like a very short thing or you give us two functions and you're like hey help me edit this then you know that's like we can use gp4 but if you give us five code files and you say I want to do XYZ then you know we might need to use clouds you know 2.1 model so
those are kind of some of the different you know types of um things that we're doing so we try to manage some of that for the user so you don't have to think about that that being said we have of course people saying oh I only want to use this model so we're also building capabilities to let you like dictate the model that you want but most of the time most users don't want they don't want to think about it to be honest okay yes so you take that away it's a bit of a backend sort of decision that's made
based on what what the what the entry is what the data is that's going in yeah yeah exactly oh good stuff okay so now that we've got that sorted out talk to me about agents because you mentioned agents so what are agents and how are they different to you know what people are doing today yeah so we think you know agents are important and are going to be the future um so agents we Define them and a lot of people are talking about agents and new words like agentic are coming up but agents we Define as something that
really completes an entire workflow or or task um and so you know we really see vidto as like building these Dev agents which are agents for devs to help them complete these everyday tasks and workflows um and what what we to give you some examples for that like um you know like um we just launched like a code review agent um and so that agent will sort of integrates into git GitHub and gitlab you know big buckets coming soon and basically when you submit a PR and developer says Hey I want somebody to review this code it runs an entire
kind kind of code review you know on that on that code and it also incorporates in static analysis tools and you know also open source vulnerability um scans and so it takes all those recommendations from those tools and then suggests you like line level changes and so our thought there is that a lot of time and energy gets spent in code review and Dev organizations but this agent we think can you know reduce that time by you know 50% probably um and as as our you know agents get better and the AI gets better you know we think that that can
go down to like be reduced 80% or we've even had some you know organizations tell us like for example we give you a score around what's the quality you know of this um of this um PR poll request and we also give you a a score on the complexity of the pr so some organizations that hey if it's a really high quality PR like the code submitted is really good quality and well done and it's a really low complexity thing imagine like a oneline change then maybe we don't even need a human to review it
the AI could just say hey just Auto merge this you know so there's a lot of and that's the agent that's the agent component so so because when I think of agent I think of say that sits on a little computer and does does a little background work and whatnot but you're talking about this in a literal sense like a help like almost like a helper guy like someone yeah the AI is replacing a human to do something right like they are literally an agent sitting there and going yes no that's good that's bad let's go through it again
yeah exactly and so it runs and says hey you should do all these things or you know or and and so it's something that again fits into kind of a workflow that you have or you know a process that you have and and kind of completes you know that task or gives like an initial kind of pass through and we actually want to make these agents available in your IDE as well so that before you go to submit to PR you can run this you know code review agent and say oh yeah it's it's in my IDE and so like it gave me the
feedback I can change before even get to P request so those are you know some of the types of you know agents that that are out there that agent that we built yeah hey good stuff there I think I want to spend the last sort of eight or so minutes talking about just the state of generative AI especially in in the devop world in the developer World um and then job security right because you know I know that a lot of the sort of big headlines were like that's it developers are dead there's going to be job losses
um you know 2024 is going to come in and we're going to all this all this craziness in the market I don't think we've seen that right um what's what's your thoughts on where we are today and how this sort of technology is going to help like we talked about generative AI I mentioned earlier like assistive AI as well so you know where are we at there and what's your thoughts on how it's going to impact jobs moving forward yeah yeah so I think um if you think back to when chat gbt was launched right I think November of
2020 too um there was a lot of talk right I mean I I was on text threads With Friends There Was You Know CEOs of software companies were like oh my God everyone's going to lose their job like blah blah blah blah blah and I think that's fast forward you know whatever 15 months a year basically I don't think too many people have lost their job you know from from AI um and in fact I think if you talk to most people in most companies they would say I don't really use AI to do anything you know um so I think that you know
sort of some of those very short-term concerns have been a bit like overblown um but I think that I think like like in the medium term I think there could be some you know job loss as people get more efficient yeah but I think like if you think back to like when we have had other you know the these kind of Step changes and you know we can debate whether this is a step change or not but if you just assume it is for a second like compared to like let's say the Industrial Revolution you know it's like you know it's like okay well yeah we
went moveed to factories and so you say oh there is some job loss because like oh now this Factory produces more stuff more efficiently but I think if you look out over time like a lot more jobs were created because humans are really Innovative and they just keep consuming more and more you know kind of stuff and I think you know the the way I like to think about it a little bit is if you think about most companies you know they have a backlog of features that they never get to um you know that backlog you know it's kind of almost a joke
sometimes right oh it's in the backlog and it's like okay so maybe I'm never going to see that again you know but you know if if you look think about like Mark Andre's um uh quote you know about um uh software is eating the World um I mean I sort of feel like you haven't really seen anything yet you know if we're talking about 10 20x Improvement um in in capabilities for developers like we're going to produce a lot more software so you're going to be able to go a lot further down on that backlog
you know and I think if you think about where you live in a competitive world yeah you could say oh I don't need any software developers anymore but I think the reality is the pace of innovation is just going to go at a much much faster Pace because that's kind of what's happened throughout time is like Innovation just fly yeah flye always kicks in and everything quicker more accept yeah it gets bigger so customers want more and more you know and so you're like well I can now as a company for-profit company you're like well if I
build this feature faster I can get that customer segment or I can sell it to this new thing and and then your competitors are doing that too by the way so I think like what this is going to result in is probably like almost more software developers or at least software developers who are like more and more productive you know and producing output you know I mean you could argue that I'm now I'm now a software developer right you could that that's scary in itself you know what I mean I like that we didn't even drink
about that that's all good I mean let's go back like 40 years in software development you were writing like machine you know Machine level code right and you know maybe machine and then maybe assembly right so you're you're writing assembly and you know then those like C++ came out right and then Java came out I was like oh my God now I don't have to do garbage collection right takes care of that for me it's like oh my God like this is a higher and higher levels of abstraction and so I think this is kind of a next
big level of abstraction where you know you don't may have to manage all the syntax all the time and you know you've got tools to help you do that and agents to help you kind of do some of that stuff so I just think like a lot of this is going to help the pace of innovation you know go faster you know I was talking to a CIO at a big drug company uh a Pharma company and he was kind of saying that like one of the things that they're you know looking at is like okay we we this is not it for developer but
just to give us sense in another domain yeah he's like okay well we have a new drug and we produce a marketing collateral you know in English big use it yeah we do need to produce that collateral in like you know 100 different languages because we're in a hundred different countries so can we use AI to help us do that right and so now they're like okay that process used to take four months now maybe we can get done in two weeks right and imery and you know collateral that's produced and like all these different
things so you know I just think the world going to go faster and faster yeah that's right and I think to that example you know that that might impact so my belief is that in the dev world devop the software development world it's only going to make the uh people more productive and it will generate more competition we've seen that the proliferation of AI companies that have popped up almost as much as what crypto companies popped up crypto was big right so there'll be a consolidation at some
point with that absolutely um but I think what you've just talked about there is probably the the more concern around this world where you know a marketer would have had to sit there and translate and people they would have had to hire a translator go through some cash and and and that might have taken a few months where else now they probably don't have to do that so outside of the the realm of software development and devs and whatnot there is potential for some for some job security issue but um like you said life finds away Humanity
finds away and we as a stepped approach whether it be Industrial Revolution you know whatever it might be all these revolutions that come at us people survive and thrive that's kind of what we do right so yeah it's all good hey hey we've got like about a minute left um just give us a little bit in a very quick sort of succinct way what's the next steps for Bido what are you looking to achieve in 2024 yeah so like I said I mean agents is a big Focus for our for us we've come out with this code review agent we've
just kind of soft launched it we' got got bunch of companies trying it out we're excited about that and then we're working on a unit test agent um and so we really think that these things could really help companies a lot in 2024 and we want to make them great for people great stuff em this has been a really great conversation I've enjoyed it wish we could talk longer but I'm really excited for what BD's doing and you know what I'm going to install it I'm going to play with it and I'm going to see how
good it is so you can probably and what I'll do is I'll tell people about it as well so not that's that's really exciting for me moving forward so hey thanks for being on the show um this is a reminder if you love great things with great Tech and like to feature in future episodes hit GTW gt.
com or at jtw JT podcast on Twitter X or on YouTube and with that thanks for being on episode 77 and we'll see you next time on great things with great Tech awesome thank you so [Music] much
Join the Email List
Listen On
Support On
Top Episodes
Here are some great episodes to start with. Or, check out episodes by topic
