Why most companies are not ready for AI agents

The next big thing in AI seems to be AI agents in which pieces of AI code will help you or companies get tasks done with less human input but our company's equipped for this next phase we're going to find out on this episode of today and [Music] Tech hi everybody Welcome to today in Tech I'm Keith Shaw joining me on the show today is Anneka Gupta she is the chief product officer at rubic but also a lecturer at the Stanford Graduate School of Business welcome to the show .000,

.640 anakah great to be here thanks for having me and I also want to mention that you were a guest on our .640,

.840 demo show which is the product demonstration episode that we have on cio.com did a great job of .840,

.560 showing all of the cool features of rubric we're not going to talk about rubic after this right .560,

.360 correct all right so I want I wanted to ask about a lot of companies that are you know deploying .360,

.440 generative AI at their companies now when you've interacted with them what are you seeing in terms .440,

.560 of what's working what are you seeing that companies might be hesitant about in this whole .560,

.120 generative AI ecosystem before we even get to the the agentic part or AI agents so I think .120,

.000 it's a really exciting moment right now in AI I would say Business Leaders across the board .000,

.360 see the opportunity to deliver vast productivity gains to their organizations um through AI but the .360,

.040 reality is is that it's still very early days so where we've seen some success ready um in talking .040,

.080 to organizations is that many organizations especially with large call center Footprints have .080,

.960 implemented generative AI um alongside machine learning to really shorten the time it takes to .960,

.360 resolve call center calls and just make that whole process a lot smoother and really deliver .360,

.160 bottomline results by reducing the costs um of managing a call center and I think the reason why .160,

.800 this is a use case that has taken off is because the ROI is so clear and the benefits dropped .800,

.000 immediately to to the bottom line what we haven't seen is much broader scale adoption of use cases .000,

.520 um at scale and I think where the hesitation is coming from is one is you know how it's figuring .520,

.960 out all the technology pieces and how do you put them together to deliver the outcome but second .960,

.440 is really a concern around data privacy and data security today there's a lot of data that is .440,

.560 within a an Enterprise where it's very difficult to find that data um and so if you accidentally .560,

.160 have permission set improperly where you have confidential um financial data that is actually .160,

.120 accessible to everyone in the the company it's probably quite difficult for some random person .120,

.640 in the company to find that information but once you start building generative AI products all of a .640,

.400 sudden um search and discovery of this data could end up leading to unintended data exposure and .400,

.520 that's really scaring a lot of Security leaders privacy leaders um and and people thinking about .520,

.760 risk because this opens up a whole new Avenue of risk that they hadn't really they haven't figured .760,

.760 out how to actually manage right so so could you could you say this as an analogy of like the the .760,

.720 individual employee might not know where a lot of this data sits but if you start adding an AI tool .720,

.960 that's automatically will find it and that's when it gets leaked right yeah exactly yeah um so so .960,

.040 these are more of the general AI tools maybe like a Microsoft co-pilot or some of these you know .040,

.600 chat GPT type tools um versus something where you can really hone in and almost add it to an .600,

.640 an existing product that's where we're seeing all of the people that I talk to on this show .640,

.040 and on demo have grabbed that what we call it the lwh hanging fruit like they have a specific .040,

.840 idea of how to use it um rather than opening it up to the entire employee base where where .840,

.240 all of this stuff can happen I haven't heard I I haven't heard yet um concerns about like .240,

.000 what you just brought up where if they're worried about these tools inadvertently leaking stuff you .000,

.800 know because of what the employees are using it for we're we're seeing this even with Microsoft :06.800,

.120 co-pilot because if you would think of like a one driver SharePoint you might accidentally .120,

.280 you might have an employee who had access to confidential or sensitive data create a copy of .280,

.880 a presentation or document and then accidentally set up the wrong sharing permissions today if .880,

.480 you were to go try to search for that you're probably not going to find it but then if you .480,

.320 start layering in even co-pilot Technologies there someone might ask a question about you know what's .320,

.200 the what's the salary of the um you know head of HR and that may end up getting published um .200,

.680 to that person because they inadvertently had access to that data when they weren't supposed .680,

.760 to oh you just gave me a new idea of what I can do what I can use at my own companies I .760,

.760 can start looking up I don't know I don't know if our our company's even um deployed any internal .760,

.400 you know chat GPT type functionality yet um but so so so you know in the in the media landscape .400,

.680 we're already seeing discussions and analysts and journalists and stories all about agent you know .680,

.760 agenic Ai and you know because again all we can do is like we're like what's the next big thing .760,

.920 what's the next big thing um so it feels like there would be some concerns because you you .920,

.440 haven't even probably seen a lot of companies that are completely on board with you know the .440,

.280 last generation of generative AI so like that that should be raising a lot of red flags as companies .280,

.920 then you know explore the idea of of a gentic AI absolutely and when we talk about generations of .920,

.400 AI we're not talking about one new generation a year now the you know underlying llms yeah there's .400,

.480 some big changes that are happening you know once a year every six months but if you look at the .480,

.040 other technologies that are developed on top of this like the agentic AI or the the AI agents that .040,

.120 people are developing startups are developing for a variety of different use cases the rate .120,

.760 of change and Innovation is probably happening every 3 months mons um and so what we're seeing .760,

.880 is that the the level of innovation and what is possible today versus 6 months ago and what will .880,

.920 be possible six months from now is just radically different and I don't think that Business Leaders :13.920,

.440 have really gred the fact that there's going to be this level of change and really what is .440,

.400 the power of a gentic AI really deployed at scale for a variety of different use cases across your .400,

.760 organization do you see the same companies um pushing a gentic AI that are that were p .760,

.400 ing you know general purpose AI tools you know are you know I think Microsoft is is looking to .400,

.520 do something I know open AI certainly is um so it feels like the same people that were pushing you .520,

.760 know these large um tools are now the same people that are doing aent and could we see something .760,

.240 similar where maybe you take that Niche purpose and a smaller company perhaps that could that .240,

.560 could handle an agenta case for that specific purpose or tool right you know or is it just .560,

.880 going to be a whole big big storm of of mess my belief is that it's going to be a whole big storm .880,

.800 of mess because there's the phrase there yeah yeah yeah exactly I don't whatever that phrase .800,

.200 entails um whatever comes to mind because yes like the big companies are all um investing in agentic .200,

.920 AI you see it in all of their releases you know Salesforce at dreamforce was talking about agent .920,

.040 force it's all about agentic AI but then you look at the burgeoning startup ecosystem and .040,

.840 the companies that are building agentic AI for spefic specific use cases and the ability for .840,

.120 them to rapidly iterate um and not necessarily be held back by the same security privacy legal .120,

.880 copyright um constraints that larger companies feel because the risk is higher for them whereas .880,

.240 for an a a startup the risk isn't there it's all the rewards so the amount of innovation that's .240,

.720 happening in the startup ecosystem I think is unparalleled and unrivaled especially when you .720,

.920 talk about really bringing together not just the llm but truly building workflows and agent .920,

.600 and it's so much more about the packaging and delivery of a highquality use case not just the .600,

.720 underlying llm technology which I think will end up being owned by the the larger companies do the .720,

.240 agentic tools that you see have the same concerns and the same issues around privacy or data leakage .240,

.520 that that some of these these General AI tools or are there even new concerns that maybe an agentic .520,

.800 tool has that that a regular gen tool might not have well I think it's a question about .800,

.600 who is using the tool and what concerns do they have right so there's a lot of consumer tools .600,

.760 right now for video editing um image editing taking notes on meetings like tons of different .760,

.840 use cases like that that are very powerful and an average consumer could probably con like use that .840,

.720 and is not going to have a lot of complaints once you start bringing that into a large Enterprise :48.720,

.160 and especially a highly regulated Enterprise like a bank or something like that the bank is .160,

.280 going to have a lot of questions about what's the architecture of this what are all the data sources .280,

.760 what am I allowing this agent to go do um and the AI agent to go do on behalf of my employee is that .760,

.120 going to fit under the regulatory standards of my industry so then that's where a lot of the .120,

.640 the concerns come up but I think like many of these companies are starting with the consumer .640,

.840 first approach and then rapidly iterating and I'm sure the ones that win will figure out how to then .840,

.560 evolve their security practices to really be able to surface um a much broader industry do you feel .560,

.200 like the agentic AI is going to be attached to existing platforms again like what we saw with .200,

.800 generative AI or are we going to see outside tools maybe that people start using outside agents .800,

.480 whether it's in a business case or even a consumer case I think we'll see both because I think .480,

.680 most of the big platforms that that organizations are using can be made a lot better through agentic .680,

.280 Ai and those companies will take the opportunity to to make those improvements such that there's .280,

.800 more automation there's more um personalization there's just less like dashboard fatigue things .800,

.040 like that and alert fatigue coming from your products but I also think that there's going to .040,

.800 be a whole new crop of startups and new companies that really go hone in on specific agents that .800,

.880 maybe kind of cross barriers in a way that the technology that organizations have purchased today .880,

.560 don't for instance it's like you know you have sales technology you have marketing technology you .560,

.320 have collaboration technology those things don't really like besides collaboration which is for the .320,

.920 entire organization most other technology is for a specific function but what is so powerful about .920,

.240 a gentic AI is that you can actually cross those barriers and you don't have to worry so much about .240,

.480 you know trying to serve only one one Persona one use case it's really something that can span all .480,

.120 of the work you do all the Technologies and that's where I don't think an necessarily an existing .120,

.600 company is going to come up with an agentic solution that really does that super well and it's .600,

.560 going to be some of these upand comers that really figure out how to develop a super inuitive user .560,

.240 experience really figure out how to make sure that the the actions that the the agents are taking are .240,

.040 right and correct and um those are going to be the ones that end up being like new technologies that .040,

.960 emerge um for organizations to adopt right and and I think um for an example you were talking to me .960,

.080 about um before the show we were talking about the you know not only do companies have to worry .080,

.000 about the tools that they deploy but then all of these external tools that that that employees .000,

.560 might be using and and I think you gave a great example you you you know you surprised me cuz .560,

.600 I didn't even know that this existed but there was a there's an AI tool that will help people .600,

.600 give answers to interview questions right yes yeah walk me through that a little bit about about some .600,

.960 of these tools and and why this could should be concerning to people that might be using or that .960,

.600 people might be hiring other people right yeah it's um it's super interesting so the the first .600,

.560 version of AI agents um if you were for instance on a video call and you had an AI agent that .560,

.520 was taking notes doing transcription providing recommendations um taking AIS like they would be .520,

.480 a member of the the call with you like if you're on a zoom call they'd be a member of the zoom call .480,

.960 and you can see that they're actively there now the way the technology has developed is that this .960,

.760 um this technology can just run silently on your device and the other person on the call may not .760,

.800 won't won't have any idea that um there's a AI agent in the background really interesting use .800,

.560 cases that are emerging in this space are around recruiting um so specifically like a lot of um now .560,

.000 people that are interviewing will have these AI agents running in the background and they're .000,

.640 getting so good because they're getting trained off of lots of different interviews that are .640,

.080 happening um especially for like the larger companies where they have their set interview .080,

.000 questions and if you get enough people enough candidates putting this on their computer you can .000,

.640 get all of the Corpus of interview questions you could understand what was a good answer and bad .640,

.720 and now these AI agents are actually sitting in the background and interview the candidates can .720,

.840 actually in real time when someone asks them a question um get recommendations of how to .840,

.840 respond to that question more effectively so that kind of throws off the entire interview .840,

.720 process because you don't know if like you're actually getting the real answers from the .720,

.840 candidate or if it's coming from an AI I saw I saw an Instagram video that showed exactly what .840,

.000 you were talking about the guy had a he had his phone and he had chat GPT underneath right right .000,

.600 underneath where the camera was and so you you were seeing the zoom call the guy the recruiter .600,

.240 or whoever was asking the interview question and then within seconds chat GPT had an answer and he .240,

.760 then the guy just started repeating the answer or or at least you know trying to make it sound .760,

.680 like it was his answer and um you know that was that was well I mean it's impressive and .680,

.400 scary at the same time it's impressive and scary and it's that's technology is today so you can .400,

.760 imagine like how that could develop in the future especially with video and audio and images where .760,

.400 it's like you may end up having an interview with an AI itself and not know that it's not .400,

.400 a real person but it's actually an AI agent so yeah I mean I'll I'll be I'll be replacing me .400,

.320 with my agentic AI Avatar and then let them do the interview and then I'll get hired at some .320,

.440 point or or my avatar will get hired and then I'll just be sitting at home like Wall-E um or .440,

.760 or those people that were in Wall-E um all right so I want I want to switch gears just a little bit .760,

.360 so um you know we we've talked a lot about how threat actors are are going to use generative .360,

.360 AI in new ways you know you know early on it was all about you know making the the fishing email .360,

.520 better have better grammar spelling Corrections and all that kind of stuff um do you do you see in .520,

.200 the space that threat actors might be taking advantage of atic AI too and you're going to .200,

.160 start seeing even more attacks or it it feels like they're going to be doing more attacks .160,

.200 rather rather than actually more sophisticated attacks they're just going to be able to do more .200,

.080 of them right yeah I think it's both they're going to both be able to do more sophisticated .080,

.800 attacks um create you know deep fakes are not just about you know better fishing emails but .800,

.920 text audio video right all of that plays into um U into attacks where you're trying to compromise .920,

.920 identity and at the same time what a gentic AI is also going to be able to do is because it can .920,

.040 learn while doing various tasks it can launch a bunch of different types of attacks against .040,

.680 a system figure out what's working what's not adapt to that and and get really good .680,

.480 at exposing vulnerabilities so when you talk about like you know zero day vulnerabilities .480,

.040 for instance um being found in systems you know today it might take a bit longer for someone to .040,

.960 actually be able to exploit that zero day vulnerability but now like it could happen .960,

.680 instantaneously and that it becomes really scary because how quickly can an organization discover .680,

.040 these these vulnerabilities and Patch them now again the technology will also be used within the .040,

.080 security products that companies buy so it will kind of be this this race we're back to a space .080,

.160 race between the good guys and the bad guys yeah y um yeah because a lot of the a lot of the SEC .160,

.680 tools now you know you you see software that that can detect if an attack is going on and you .680,

.280 know there's all of these alerts that get but those Alerts get sent to a human and then the .280,

.400 human has to decide if this is a real attack or not a real Attack so it's almost like the defense .400,

.840 of that allow my agent AI to say is this real is this not real and then and then really just .840,

.480 either fight it if it's if it's if it if it can be fought like at what point do you you still need .480,

.280 humans to to have some kind of decision-making power right yeah I mean at the end end of the .280,

.640 day like a a human is going to have to decide for instance if you're a retail stores set of .640,

.200 retail stores and you're under a security attack do you close the retail storage right someone has .200,

.120 to make that decision but like can the AI go and say well I found this um potential I got got .120,

.920 this alert they go investigate whether the alert is real what caused it was it actually an attack .920,

.760 do they do all of those steps and then do they ultimately come to a recommend like come back with .760,

.080 a recommendation with all the data supporting it to a human that's probably what's going to happen .080,

.000 but I don't know that like to what extent for the really big decisions that impact an organization .000,

.600 whether someone's going to trust an AI to make that decision on their behalf yeah are are you .600,

.200 seeing a lot of these tools um being deployed at this point have they been created yet or are we .200,

.280 still talking about a lot of what if scenarios I mean I I would say the security industry has .280,

.960 always been on The Cutting Edge of using AI but truly doing these agentic AI use cases where .960,

.000 you're essentially pulling together signals from multiple different tools investigating .000,

.840 things that's still early days and I don't know you know I I I don't know if the technology is .840,

.000 quite there yet but I fully believe it will get there because you can see again these kinds of .000,

.000 use use cases being solved across many different types of um you know whether like collaboration .000,

.080 produ like other developer productivity things like that so you can imagine that it's it's you .080,

.080 know it's being worked on right now and um these Solutions will come up where you're not going to .080,

.840 need a whole bunch of sock analysts to go figure out what is is um what's real and what's not and .840,

.160 like triage through your millions of alerts right do you do you think companies um that are thinking .160,

.160 about a gentic AI um should they take a step back and maybe look at see you know what advice would .160,

.920 you give them should they take a step back and go let's get a handle on our on our first phase .920,

.480 of of generative Ai and make me feel comfortable about the data that that might be getting leaked .480,

.120 or at least preventing the the data from being leaked um before they get into agenic AI or can .120,

.160 they can they explore that while they're also doing that that protection I think it has to .160,

.400 be a dual pronged approach so one is yes like how do I put in place the the um Technologies :40.400,

.240 capabilities Etc to um reduce the downside risk of deploying these Technologies and and inadvertently .240,

.360 exposing data or exposing um things that you don't want to expose about your in your organization .360,

.360 internally or externally but I also think that organizations that really want be ahead have to be .360,

.920 immerse themselves in where technology is going and they have to experiment at least maybe they .920,

.480 do it on their personal laptop at home right but someone needs to be really thinking about this .480,

.400 because again the landscape is changing so fast that if you just wait you will be left behind now .400,

.320 I'm not saying if you're a super traditional organization like go find like take a hundred .320,

.800 people and put them on the The Cutting like figure out the Cutting Edge tools but I do think that .800,

.360 it's worthwhile like really having someone in the organization have a finger on the pulse to like .360,

.120 really see what the art of the possible is because I guarantee your competitors are doing that and so .120,

.920 if you don't do it you will be left behind today the productivity gains that we typically see with .920,

.080 like kind of the the versions of of gen that are being used are like something like 20 to .080,

.080 30% which is still really good but I believe that the productivity gains that we really can get from .080,

.720 AI are massively larger than that I'm I'm still I'm still waiting to a hear a bunch of scenarios .720,

.280 that will help me in my own job um but then you know with all of these productivity gains that .280,

.680 are happening um I think people are left with like well what am I supposed to do with the rest of the .680,

.720 time that's been saved um and I'm and it it just feels like the immedi it just get back filled with .720,

.000 more mundane stuff or more or more meetings I'm hoping like they all everybody says oh yeah it .000,

.160 you know it allow your people to be more creative and you know more engaged and and all of this this .160,

.760 this stuff but I sometimes think that's a utopian wh if too of you know and and it won't be back .760,

.760 fi with other with other tasks or more tasks or something like that well people will always .760,

.520 find a way to fill their own time right and and organizations will find a way to fill people's .520,

.000 time I don't know if it will be backfilled with more mundane tasks but I don't think that the .000,

.320 outcome is like this utopian Paradise for everyone because the reality is is anytime you have a big .320,

.160 shift in productivity um typically like the labor markets the regulation the government support .160,

.120 systems don't change fast enough to keep up and there's with the rate of change of AI I think like .120,

.720 I'm even more worried about what is a societal upheaval that happens as you start saying well .720,

.280 okay a lot of the work that can be done today um that it can now be done by AI so like where does .280,

.960 that put people in terms of H you know what how does that displace the workforce right and .960,

.800 then and then we yeah we'll be back to that whole discussion about whether or not an AI is replacing .800,

.200 a job versus replacing a task um right yeah I you know so I me I I'm sure that these discussions .200,

.360 will happen as we go through the the year and and see even more agenic AI um Ju Just a while I've .360,

.080 got you here I I wanted to ask about um consumer use of of agentic AI are you seeing anything out .080,

.840 there that you know it feels like this is going to be a business First Technology um but it may .840,

.880 trickle down into the the consumer space I haven't seen any good examples yet of of of of a use case .880,

.880 where I go wow this is great um and then and every time I try to think about a a scenario where it .880,

.720 might be useful to me I I then start thinking about how much data am I giving up to this agenic .720,

.160 Ai and that concerns me because again yeah go ahead so I think actually the biggest leaps .160,

.800 right now in gen are being made in the consumer space oh really so so it's okay yes um well then .800,

.160 I'm talking to the right person then so tell me where you're seeing this so I I mean I'll give .160,

.280 you a couple of examples one one is um and this could you could see how this could translate into .280,

.520 the business space quite um quite successfully there are applications where you essentially like .520,

.120 set this up as a note taker in the background of your meetings yeah um it will take notes for you .120,

.680 you can also take notes yourself it will merge the notes together it will generate the list of .680,

.720 action items based on the meeting and then kind of the next step of it I think it's still early .720,

.680 days in these Technologies is it'll actually go do the actions for you so it's like oh send an email .680,

.240 to coordinate a meeting or schedule a meeting to do this or send a slack message to this person go .240,

.880 figure out what like the answer to this technical question is so we can return to the customer it .880,

.360 will actually be able to do all of those steps for you um and that is like hugely productive .360,

.880 it's like how much of my day am I spending like I'm taking meetings and then it's basically all .880,

.120 of the follow-ups all of the action items all of the notes could automatically be done for .120,

.520 me but that's still a business use case because it's still personal use case yeah maybe a better .520,

.920 example of a consumer use case is just around like Photo Video creation um and the just the ease at .920,

.200 which now you can create a whole video or create yourself or any like family member into a photo um .200,

.240 and it looks super realistic and you can basically put yourself in any scenario you can create video .240,

.080 content really easily um you can create um just any sort of web content really easily too like .080,

.000 those are things I mean again those translate into to business use cases but today the predominant .000,

.080 uses of this technology is in consumer where where creators are using this artists are using it um .080,

.960 people who would just like want to be tinkering around with AI are using it um and there's a lot .960,

.000 of like really interesting art and other things that are coming out of AI okay but but that but .000,

.200 that's just the regular AI stuff too not not necessarily the agentic AI right or yeah it's .200,

.440 a combination I think it's a combin you're right it's like it's a bit of um the evolution of llms .440,

.440 and being able to fine-tune them but it's also the workflow that you build on top of it um but but .440,

.600 yeah I mean I think you're right like a lot of the agentic AI stuff you can essentially think of as .600,

.000 like having a personal assistant to do all of your to do all of your household taxes etc for you or .000,

.120 having like an assistant at work to do all of that that that work for you well yeah so so that always .120,

.160 makes me think of if I'm going to have a personal assistant I'm going to want one personal assistant .160,

.960 to do seven different things like you know do my taxes but also send out emails um plan these trips .960,

.480 go shopping so that's at least four different um tasks within that agentic AI so do you think that .480,

.200 there's going to be a a general purpose agentic AI that people are going to use or they going to have .200,

.200 to explore four different unique individual you know agentic a there'll be there'll be one AI that .200,

.160 can do all of that already what are they do is it going to do it good or well yeah I mean there's .160,

.960 already like the voice technology on AI is so good that many for like many logistics companies and .960,

.040 things like that they're actually replacing their the coordination the people that you call to like .040,

.880 coordinate things with AI and they're able to do the job yeah and like adapt to the situation so .880,

.720 it's pretty pretty good like I do think that you can have a personal assistant that'll be able to .720,

.720 do all of those tasks does it require that users give up you know a lot of their information or is .720,

.320 there an assumption that all of this is already known by by by most people I mean I think it .320,

.480 depends on like what you're what the use case that you're trying to to go after and do right if you .480,

.960 end up wanting your agentic AI to like make Bank transfers I don't know how you do that without .960,

.280 giving them access to your bank account but like that's a decision that that one could make of .280,

.880 like do I want to actually enable that kind of use case what we've seen time and time again is .880,

.680 that if the benefit is high enough people will be willing to give up their data to do it now I .680,

.840 think they're going to want the safeguards around that too and these companies are going to have to .840,

.080 figure out how to build those right safeguards so that protecting that very precious data because .080,

.920 again you're just like as you collect more and more data you become a target for the attackers .920,

.120 right and again you can you can say the same thing about about business usage of this too .120,

.880 how much data is a business going to have to give up in order for these agents to work and then and .880,

.040 then prove that that these are trustworthy as well so you know because again there's a .040,

.120 trustworthiness Factor on the consumer side too totally like if if I gave my agent access to my .120,

.120 bank account it would it would come back with wow you spend a lot of time at McDonald's um maybe .120,

.960 you know and then I ask it like oh give me some dinner it's going to say well you know you you've .960,

.080 been you've been going to McDonald's a lot so I'll get you McDonald's and I'll be like I don't want .080,

.960 McDonald's I want something else and that that's you know what I mean like that's where it gets all .960,

.160 it's it gets all messy and Head Start hurt I think there's always going to be some if you have like .160,

.920 a personal assistant gentic AI or AI agent you're going to have you're going to have to do some .920,

.800 work in giving them some context now that may be they they you may be able to take all of the you .800,

.440 know your phone call data text Data whatever that already exists whatever you already have .440,

.320 and dump it into this AI very easily so you may not have to sit there and like really train it .320,

.240 but there's a lot of like data that we're creating exhaust data that we create in our da daily lives .240,

.960 whether it's your grocery list or whatever like all that stuff is in in electronic form at .960,

.640 least for me and so someone could learn from that like an agent could um could learn from that very .640,

.040 quickly and understand what my preferences are and how I operate and what I want and when I'm .040,

.840 talking about what to have to for dinner with my husband like what's the conversation that we .840,

.800 have so you can imagine that the the level of effort it could actually be quite low because .800,

.880 a lot of people are already kind of creating a lot of this data it's just not in a very usable .880,

.880 form today but um llms don't need it in a usable form yeah yeah okay so do you think that this is .880,

.200 going to be the year 2025 will be the year that that we're going to see a lot of this moving or .200,

.040 you think it's going to be slow and cautious you I think that's still a adoption within Enterprise :29.040,

.680 is going to be somewhat slow and cautious but I think what we're going to really see this year .680,

.040 is true adoption of use cases that go beyond the call center or some of these really really obvious .040,

.840 places um to start seeing that and I think in the consumer world like individuals let's say .840,

.080 I think we're going to see a massive explosion in things like these personal agents to do all .080,

.040 of your household chores all right we're going to have you back we'll have you back at the beginning .040,

.240 of next year and then we'll see we'll see how how we did on on this sounds great yeah anuka .240,

.160 Gupta again she's the chief product officer at Rubik thanks again for being on the show um .160,

.160 and um you know thanks for those insights all right that's all the time we have for today's .160,

.720 episode be sure to like the video subscribe to the channel add any thoughts you have below join us .720,

.715 every week for new episodes of today and Tech I'm Keith Shaw thanks for watching.