How Anthropic enabling it’s India GTM strategy, explains Anthropic India MD Irina Ghose

0:00 India's complexity li lies in the diversity of the population it serves the scale the multilingual language that we really have to solve for and when you look at that claude from the way that we putting it across to the hands of the users in banking say for example hundreds of millions of users are interfacing and trying to access and get inclusive in the bank banking spectrum over a long period of time now claude is being built across and we are getting better every day by ensuring that the largest spoken languages more than 10 of them how do we get better at that and even building proficiency on the training systems that it is kind of getting built for so that's one side now on the other side India's core priority sectors whether it's agriculture whether it's education healthcare skilling in each of them it really matters because the minute you kind of try to solve population scale problems by making it available to the last mile. That's where the real difference lies. Just some time back we announced that the entire thing about trust about where the data lies, where the inferencing lies, that's soon coming across via Amazon bedrock in India and that also gives us a new way of opening up the vistas. for example Ashokaswani the MG of Kotuk bank said that the when the frontier model is available in the proximity of the Indian soil for the ecosystem that will be a gamecher again NPCI which is our host when they are building across AI next their platform on claude that's a great way of kind of ensuring that we are solving larger ecosystems making it centrally available for people to build things which are connectors and getting them a lot more pervasive >> right so Arena you mentioned so many sectors as well as the companies which are really using it but a large part of the business community still hasn't taken that particular decision or really convert it what is the boardroom discussion at the current moment and what are you seeing across the board in terms of the requirements to really convert and adopt AI like their other peers >> absolutely And at the highest level there are two kinds of organizations.

2:27 One which has decided to take a bet and are moving ahead on the curve and the journey and the others who are a little cautious to see that is it the right time is it the right manner is it the right methodology to go in for and in all my conversations and my experience in the dialogues that I've had I find two things. The ones who have kind of gone ahead of the curve are looking at AI to see as to how they can touch two ends of the spectrum.

2:58 >> Essentially AI can do two things right one everybody is fastidiously measuring the efficiency how efficient have you made the existing system how have you kind of brought across efficiencies in resources etc. When you're measuring that the other part we always encourage and the ones who are ahead of the curve are measuring efficiency is bringing in headroom for capacity for innovation. Now are you really measuring the innovation because that innovation is what will be giving the quantum levers and the multipliers of the growth. So on one side if you look at what efficiency is bringing across to everybody in the team in the department and the organization. Second also look at possibly those big rocks or the large moonshots that you would want to go after and look at how are you trending on those milestones for driving innovation that's driving the growth. So that's what the ones ahead of the curve are doing. the ones who are kind of still gauging whether or not to go. I think one of the things which can always be a driver is to have the leader in the organization in the room in the team to adopt it himself or herself. If the one who's kind of looking at making the change finds the biggest problem which they are not asking what can AI do but says what good outcome would look like.

4:28 they literally have that clarity of saying the problem definition that they're wanting to solve and they put AI across to iteratively work alongside it. I think that's the way the entire culture will change. So Arena of course efficiency is one thing but also cost competitiveness, cost efficiency as well as return on the investment deployed in technology is also one of the very big factors for decision making. How competitive is frontier model and the solutions that are available which are much more high-tech and how is the competitive edge versus point solutions which are now available because of open-source AI.

5:14 >> Let me put it in in a manner that you kind of understand as to where the entire value is coming out from. At the highest end, the intelligence when you put across to the largest complex set of problems is what defines what the ROI is kind of coming across from. Now when you look at the frontier models and claude really shines across when you look at discovering doing things which you don't know for example in the sphere of healthcare drug discovery in the sphere of banking everybody sitting here whosoever has been associated with banks would be knowing of reams and legacy code which have been existing in the banks people have not touched it. When you put across these problems to frontier models, things that you don't know, that's where the intelligence shines. Because look at this, the legacy code, even the conceptualization of it, which would have taken months, which would have taken a very high level of experienced users to do it can now be done in days.

6:22 and Frontier AI can make that context available to people who've not possibly dealt with the experience and can start doing it very quickly. So I would say that look at the kinds of problems that you're solving look at things which are kind of going ahead in terms of knowing things across in the sphere of healthcare whatever you want to do across in the financial segment and that can make it h happen. I'll give you a few examples. For example, in the case of financial segments in India, you've got say Razer Pay which built across an agentic studio. Now this agentic studio can do a few things. It can do merchant reclamation which was not happening earlier. It can do a great dispute respondent which was not happening earlier. It can also kind of ensure that it is doing a lot of cash forecasting which was not happening. Now this is being done by a very lean and mean team. Now when you look at people who are putting a premium to this that's where it makes a difference. Now the same level of knowledge by which this is getting built by say the fintech industries out here. When we put this across to the larger banking community out there we created something called claude for financial services. Now what cloud for financial services does it's created what we look at as a overall coms analysis is put across a DCF analyzer is kind of put across data where it is pulling things across from as well as connectors to the various places such as the Moody's such as the SMS organizations and it is already kind of being used by places such as cities such as Brex such as Coinbase etc. Now when you look at this, you've kind of created an institutionalized way in which you can kind of put things across in the financial segment. So putting a harness and giving that across to other organization is where it kind of really comes into play.

8:25 >> Right. So Arena from a lot of what you said this is not a templated model that really can do a little bit of tailormade approach especially when it comes to large enterprises and the solutions that you talked about that is also something which is the requirement at this particular time while there are AI native companies as well as sectors. So is that also incorporated in the new consultancy services which most of the AI companies are now going towards? Is that a move and a trend in the industry that you're seeing?

9:04 >> Absolutely. And when you look at the largest of the gsis or the largest of the consulting organizations whether it's the likes of the deoids, the TCSS, the infosces, they are all kind of looking at the way that they are redeploying or recreating their playbooks and making themselves AI first, AI native in the way that they are addressing their customers. And this takes the form in a few ways. First, how are they re-imagining the solutions that they would be going to their customers with? The edge that they always have is the domain knowledge of their customers. The second is the last mile that they really know. And the third is the prowess of the technology.

9:48 The second what they're doing is that they are not looking at AI as a thing that they're bolting across and asking their employees. It's a large cultural tide shift that they're looking at. So when Cognizance gives say Claude across to 350,000 of of its employees, they are rem reimagining things for things such as what they can do in the space of healthcare, what they can do in the space of anything to do with insurance etc. And third, what they're also doing is that the platforms that they had, they are infusing AI into them to ensure the agility and the responsiveness of what can be more futureproof and future ready is getting built alongside too, >> right? so many companies have the vision of being future ready. A lot of the companies may be also very price sensitive and margin sensitive especially in the listed space where they have to give an answer to their stakeholders every quarter arena. So how do you cross that bridge with the companies which are right on the edge of making a decision and the the requirement is definitely there but it's the competitive edge when it comes to the financials and return of investments. What is the tipping point over there in your view? I think a lot of times our customers ask us a question that what is the distribute what is the differentiation and what is the real mode and in my mind the differentiation and the mode comes across by thinking ahead of the curve and ensuring that how are you kind of building things that have been with you and few things which always kind of help them get ahead on this is one the distribution channel. So a distribution channel which has been created by an organization over a period of time that literally is a great value.

11:41 The second is the trust that they have built across with the organizations with their customers over a period of time and the third is the data. So to the question that you were asking how are they kind of managing things around the cost and the efficiencies at the end of the day the models or what they can do with the models is actually a conduit towards what they're serving with the customers with these distribution and the data and the trust models that they are building across >> right. so when we talk about decision making and the companies which have crossed over what are the kind of benefits as well as productivity and the returns in terms of revenues as well as customer being more satisfied and they taking a lead in the sector. What are the instances that you have seen already?

12:32 >> Quite a few and varied actually. So depending on the nature of the customers there are things which get measured on a few fronts. One at the very early stages people would really want to see what's happening as an employee benefit or things within the organization. So when you're going out what am I doing to the people power of my organization? So things which are kind of going for a change from a productivity for both the developers and the non-developers for example in anthropic now claude can merge across more than 80% of the merge lines of things that we are writing what that gives in a sense is a lot of capacity for people to kind of innovate and co-work was one such product which was written in about 10 days because people had the capacity. So the first is the productivity benefit that you're looking at in your organization. The second is a lot to do with the efficiencies that you're looking at in in the way things have been working. So how are you streamlining things which have been working across between organizations, manual processes, orchestrations etc which were happening.

13:46 And the third is reimagining recreating products which are customerf facing. So when you're looking at all these elements which people are looking at tracking on the customer side is a lot to do like you said towards the customer satisfaction is it agile is it accurate how many turns am I kind of taking towards looking at complex solution when I'm looking at growth have I kind of made a breakthrough in a least possible time because I had 10 ideas I'm able to test 10 ideas because of AI and then look at the best idea and take it forward. And third, am I able to unlock new avenues of revenue which I was not able to do earlier. All right. So these are some of those important examples that you've mentioned. but also Arena the cycle of technological evolution is much shorter than what we have seen in the brick and motor and the old economy sectors. we are also seeing boom and bust and the evolution very fast and especially in the artificial intelligence space. When we talk about the version upgrades, how easy or difficult is it to push down the upgrades and how well is that also accommodated in the companies or are the companies which have already invested say we are okay with the first version as long as it lasts like we do with iPhone version where till the last one lasts we are okay using that. I would say India being a builder economy and India's developers and the developers who sitting across the large ITIS organizations as well as the ones who are sitting in the entire startup ecosystem. They are the ones who have surprised everybody with everybody with the highest levels of tech intensity.

15:37 They are forever experimenting and we've always found that as soon as any model, any feature, any point of view comes across from us, they are the first ones who are kind of taking any of these in their hands, giving us feedback. So they're ingenuity and intuition to experiment with it is very fast. That's one. You're right towards saying that the way the versions are emerging is very fast. So what we also encourage is to ensure that in any organization the ways evaluations were happening earlier long PC cycle long milestones checking that out and then kind of taking a decision that has crunched a lot right now for organizations which are ahead of the curve because the fallacy of keeping a longer evaluation cycle is that by the time a net new model might have come out. I mean, we released the last one about 9 days back and we always encourage our customers to be on the move to ensure that you're doing shorter cycles, testing them quickly, putting out something and learning from it and then scaling. So that's the second part that we kind of encourage everybody to go along with us.

16:45 >> So the long story short are the adoption of upgraded versions also which is a critical part of the journey is that really taking place already? Absolutely. People always are wanting to experiment with the latest versions of the model, the latest versions of whatever we are putting across as the feature and they're also kind of putting across any of the point of views that we are engaging with and that's a very very vibrant cycle. Absolutely. >> Right. Also in the trends in the sector arena when enterprises are using AI solutions and are building the new framework for higher technological progress. When we talk about that, how many enterprises in your view really are keen on apart from the top layer are really keen on huge investments on very high level advanced technology. A large part of them as reports some reports suggest are okay with some point solutions and also lower versions of the technological advantage to get a better efficiency in their system. What is your view on this as an industry? I would say there's been a shift from what was happening about a couple of years back and to where the industry has moved in the last n months.

18:07 >> About a couple of years back the question was a lot to do with what can I do with AI? There was a lot of experimentation happening because the validation of the outcomes wasn't there. From then to now the validations, the outcomes, the results are proven and hence the intent of the industry whether it's banking, financials, indust insurance, healthcare across the board, the overall intent is to engage in areas which can make a material difference and that has gone across dramatically. So like I called out the segments earlier each of these are in very good mature stages of the evaluation cycle investing ahead of the curve and ensuring it's not just their own employees what they're doing across with it for their customers also is kind of coming across in a very very inclusive manner. How do you see the ecosystem in terms of the upskilling arena of the employees the human capital movement in terms of choice of companies they want to really join in terms of doing mundane daily operations versus something which uses their skill set and elevates it with the use of technology. What is the overall trend in ecosystem in terms of the employee movement use of human capital and choice of talent in AIdriven companies which are AI native?

19:41 >> That's a great question and I would say that a lot of times when I meet people they ask me as to how are you seeing the talent shift happening. I think what is valued very much in today's world is from a overall skill set is the person's curious is the person willing to let go of what they knew and be very curious to embrace what they don't know. So what people want to kind of do when they're looking at a change in the industry is join or look at organizations which will make them more capable at the highest level. Right now what does that mean? AI is doing a compression on a few things which is taking away the mundane, taking away the day-to-day rigor which is coming across. But there is a compounding premium which is always there on judgment on outcome clarity on systems thinking. Now when people are having an ability to do this in AI organization or in a world to have these levels of higher levels of capability being demonstrated and they can learn alongside what an AIE ecosystem can do, they really kind of are jumping on it.

20:56 various scenarios which kind of come across into play. Examples of these are when people are joining across say a banking organization or a fintech organization. They really want to kind of see the way a problem which was not solved earlier. Do I with my mental problem solving model have an ability to define what good looks like and then let the model as an AI work alongside me to work on it. >> Right? Another industry trend is also the data privacy which we all are now grappling with. governments, regulations, everything is in an evolutionary stage right now arena and very critical to this particular sector.

21:39 Now cloud versus use of own device and with laptops now becoming much more efficient and with the use of technology much cheaper. How are you seeing this particular scenario developing for the AI world? I would say the need for having data privacy and the need for having data reside within the boundaries was always an ask for the industry >> and towards that we absolutely want to ensure that the way we are serving the institutions people in the country we ensure the data privacy for the data residing in the country and inferencing via like I talked about making it available with the hyperscalers is available. It's a very valid ask because at the end of the day, anything which is getting built requires that levels of security, privacy and ensuring that you're having the right levels of governance. For us at Anthropic, frankly, we consider safety and capability two sides of the same coin. And whatever we are building across, the audience that we are serving it across to and the way that we are putting the distribution that takes into consideration this entire circle. So to answer your question, these are absolutely inbuilt in the way a model should be kind of coming across and we are always wanting to work across with the customers, the regulators and the entire founding ecosystem to make it work in that manner >> right and also Arena while now technology has made the world borderless we are using technology across the world companies sitting here are being served by you from the US while we are still formulating regulations and you know country alliances ecosystem for fair play. How big is the regulatory risk of crossborder diplomatic and geopolitical relations when we develop the AI culture across the world? I would say just like any other technology that you'll be kind of putting across and deploying there's always a need to kind of ensure that we work across at the highest levels in the ma in the manner the technology is accessible working across in the form that the global good is put across anthropic like like you know is a public benefits company how we look at the mission towards ensuring that we solve the last mile for the problems of humanity as at large is the biggest mission that we strive for and towards that the work that we do in every country that we engage with and especially in India working very closely with the government and the country is one of our prime force too >> and also now you lead India arena how is India positioned in the overall scheme of things for anthropic as a growth market and as an AI native market >> very core to everything that we're building across from here not just for what we serve for the people who are consuming here but also So what we learn from here create the playbook and ensure that is serving the rest of the global good as well.

24:49 >> Right. So good to hear that about India and just to also say that you know all the sci-fi movies we have seen 20 years back and I don't know if you are a matrix fan but you know more than two decades back in matrix it was said AI will rule the world and transformers has had many such occasions of how machine as well as humans interact. How do you see this ecosystem developing with AI and human capital working together?

25:20 >> Two parts to it and like Dario has our founder has put across there are two papers and I would just encourage everybody to read them in case you've not had an opportunity. The first one is called machines of loving grace in which we call about a future and in which machines and humanity will be coexisting side by side and also being cleareyed about what are the ways it can pan out. There's the second paper which is called adolescence of technology again to think about what are the ways that it can pan out which can have any impact which could which could which we should be thinking about and then going in that journey. So these are the two bookends in my view. so that we can a think about what are the absolute benefits and be clear about cleareyide about where it can go across to and then take the journey alongside. Thank you so much Elina for very patiently taking all my questions about the Indian market and how India can be an AI native market and how the new financial order is getting built with the use of AI technology. Thank you so much Arena for joining us. Thanks to an engaging audience. Thank you.