Latest / Thursday Think Bites / Episode21 - 21 Weeks, 21 Lessons, One Decision
Transcript
- Prashanthi Nuthi: Hello friends, usually say if you do anything for 21 times, you a habit. And every week I'm going to spotlight an impact maker inviting them a guest to my podcast. ⁓ Today's episode is really very special because we have a very special guest Ganesh Regmi 21 weeks back when I was clueless about how to launch Thursday ThinkBites, he was very kind enough to help me with tips and tricks and tools about how to launch the podcast, how to write the blog, so on and so forth. So from the bottom of my heart, Thanks to Ganesh for all the guidance that you provided me. If it not for you, Thursday Think Bites wouldn't have been a reality now. Ganesh Joshi Regmi: thank you Prashanthi. And before I introduce myself, I just want to make sure that you me too much of credit for the impact boost thing. ⁓ was only there to help you with some small tooling ⁓ that you need, but all credit goes you. This is a huge I ⁓ I know you as great leader from the past, and this is a ⁓ next step in your interest that you're looking to build a community and this is going to take a long way. Your learnings are really valuable. That certainly helped me and I know it's going to help the wider community as well. My name is Ganesh Joshiragmi. I work for Amazon at the moment. And what I do at Amazon is I'm one of their modernization specialists. So I work with the enterprise customers in the Prashanthi Nuthi: additional micro boos for you today. Impact doesn't check ⁓ age, gender, background or ability to create impact. If you believe you are an impact maker and want to join a community who believes in making real impact, please use link below to join the community. Until we meet again next time, let's keep making impact together. Thank you. Ganesh Joshi Regmi: FSI healthcare sector and I help them redesign and re-architect their application infrastructure, application architecture so that their application can make use of the Gen.AI movement that's going on, all the Gen.AI innovation. So my job in nutshell is between legacy application and Gen.AI adoption as a bridge. Prashanthi Nuthi: been hearing so things this AI, Gen-AI, Gen-AI movement and everything and also I've heard a lot of popular topics about AWS bedrock So what is this AWS bedrock? Ganesh Joshi Regmi: Yeah, that's an interesting thing, Prashanthi. These days I talk a lot about Gen.ai services and specifically because I work for AWS ⁓ as a part of my work and as a personal interest as well. get a lot of those questions about, what's AWS doing in Gen.ai? And they might have heard about a service or two, one of which is very popular, which is AWS Bedrock. So let me start with what Bedrock is ⁓ and why should our business leaders actually care about it. So ⁓ look at it this way. I'm going to take an analogy here. So when cloud computing actually arrived, the smartest organizations didn't ask, how do we manage servers better? What they ask rather is, what becomes possible when we no longer have to manage those servers? So that's the mindset shift separated the leaders from the others in that decade. Now, we are at the exact same moment in GNI now. And Bedrock is the infrastructure layer that enables the shift. Now what Bedrock does is Bedrock removes the need to build and manage AI systems from scratch. Now you get access to the world's leading AI models hosted securely governed properly and ready to be enabled into your businesses. Now ⁓ if I talk to a leader I always say three things so that Bedrocks give leaders. It is three C's choice, control, and confidence and I'm going to dip, ⁓ diver slightly deeper into each of those. So let's start with choice. So what Bedrock gives is Bedrock gives you a choice of multiple ⁓ AI models, leading AI model and frontier AI models. So when we know that in our, ⁓ JNI use cases, it is not that one model fits for all the business cases. ⁓ we might have to experiment with multiple AI models. Now, ⁓ Different problems need different tools. More importantly, in a market where more model capability are advancing every few months, what Bedrock does is Bedrock lets you evolve without rebuilding everything. So now that's a flexibility and it's not a technical necessity only, but a strategic hedge. Second C I spoke about was control. Now because enterprise AI without governance isn't enterprise AI, So the question I hear most from the senior leaders ⁓ isn't, ⁓ can AI do this or is AI capable of doing certain tasks? But it's, can we deploy this safely and compliantly at scale? Now, what Bedrock does is Bedrock is built inside AWS security parameter. So your data doesn't leave your environment. It's not used to train someone else's model. and your existing compliance investments extend naturally to all your AI workloads. Now, for regulatory industries like healthcare, FSIs, this isn't a detail. It's the precondition for adoption. So that's the second scene. Now I'm going to third stage, which confidence. Now, because we know a model alone isn't enough, ⁓ a foundational model is generally intelligent. It knows a great deal about the world, but it doesn't know ⁓ anything about your organization. Now, what Bedrock does is it closes that gap. It lets you connect the AI model to your own proprietary knowledge. As an example of healthcare, your clinical protocols, your customer data, your institutional expertise. So the outputs are now grounded in specific contexts, not just general training data. That's what moves AI from interesting to genuinely useful. But the reason leaders should care right now is about timing, not technology. We are at a genuine inflection point. From the past two years, most organizations were experimenting, running pilots, exploring what AI could do, asking what's possible. That phase is gone. That phase is closed already. Now the organizations pulling ahead today are the ones that have moved out curiosity to embedding AI into the fabrics of how they operate. So I would say the cost of waiting is actually larger than the cost of failing. So I would rather say at this point, a leader should care about trying and moving into their one or two business cases or business use case for the GenAI movement. Prashanthi Nuthi: like fail fast rather than failing to try. Wonderful. And Ganesh, what is your recommendation to the companies who are trying to build AI applications but do not want to invest massively in the infrastructure? Ganesh Joshi Regmi: That's right. Yes, Prashant, the larger companies and larger enterprises, they do have those capex to invest into larger, into massive infrastructure and cutting technologies, but not everyone does. Or even the larger organizations might not always have that budget, but they want to start small. So my advice is to most executive team is to start with a problem, not a platform. The organizations that get stuck are the ones that spent six months designing the perfect AI strategy before even building anything. And the ones that break through pick one high value bounded problem. Something that is painful today, which is measurable and human intensive, and they ship something real against it within weeks, not quarters. Prashanth, you and I delivered a lot of such ⁓ innovation and transformation projects in the past. We understand that instead of just sitting and trying to build a strategy, why don't we pick one small use case and then pilot it? So let's take an example of a healthcare industry. The entry points are obvious once you look from them. So things like patient intake and triage, things like clinical documentation, things like revenue cycle management, discharge planning. So these are processes that are labor intensive, error prone, and have direct downstream impact on the customer, on their patient experience. Now, AI doesn't replace the clinical judgment. We leave that to the clinical experts, to the medical experts, to the doctors and nurses. What it does is it handles the surrounding work so that the judgment can be applied where it matters the most. The key insight I would offer to leaders as part of my day-to-day interaction is that you don't need to boil the ocean to generate a real value. Pick one workflow, prove the return. build the organizational confidence and then scale. So what makes this moment different from previous technology cycle is a speed of value realization. We're not talking about a multi-year transformation program before you see outcomes. So organizations are going from idea to working capabilities in weeks and that changes the risk calculus entirely. So the question isn't can we afford to invest in AI? The question is Can we afford the compounding disadvantage of not moving at all? Prashanthi Nuthi: well said Ganesh. what's advice technology leaders who want to move ⁓ AI curiosity to AI Ganesh Joshi Regmi: Yeah, so when I work with large organizations and we do a pilot, usually we see the pilot is conducted and processed in a very controlled facility. Now, when we try to expand that into production, we see a lot of problems. So the question I spend most of my time is on with the senior leaders. And the answer is almost never about technology. The organizations stuck in the pilot purgatory share a pattern, and that's what I see. They treat AI as an IT initiative. And the pattern works, so the pilot works technically well, but it never connects to a business outcome. It never reaches the real user at scale and never generates the momentum that turns a proof of concept into capability. So what we did is we sat together, when I say we, it's my team, we sat together to see What are those organizations who move from pilot to real production AI application doing in common? And we see they do three things in common. So I'll start with the first one. So the first one they do is they have an outcome owner, not just a sponsor, not a VR exploring AI, mandate from the top. So a named leader who's accountable for a specific measurable result. As an example, if I take an example again of a healthcare system, that might be by the end of this physical year, we will reduce clinical documentation time by 30 % in these three different departments. So you know what to do, you know what areas to target, and you know when would that be achievable. So that's called specificity, which creates accountability, and accountability creates momentum, and momentum gives you outcomes. So that's first thing. Second thing we saw is they treat their own data as their competitive advantage. The AI model itself is increasingly a commodity. So when OpenAI released their GPT 2.0 model back in 2023, 22-ish, we all jumped into OpenAI. We went to chat, GPT, start writing prompts, and we saw great results. Now that's the GPT 2.0 from there to where we are today with all ⁓ the models from Amazon, like the NOAA models from Amazon, or be it one of the core anthropic models, all these models are doing wonderfully well, but they're all commodity. What differentiates you is the context you bring to it. So your organization data, your clinical protocols, institutional knowledge, that's been built over decades. So the organization's winnings are the one who's investing in making that knowledge accessible and usable. And as part of my job, I work with this large enterprise to see how that data which is lying in silos, which is running on a legacy infrastructure can be brought closer to GenAI systems so that they can make use of these models. Now, because an AI system grounded in your specific context will always perform a generic one, so it's always useful to make sure that your data is working in your competitive advantage. Now, the third one is that They're designed for human alongside the machine. The best AI developments I have ever seen aren't about replacing people. They are about fundamentally changing what your people spend time on, and spend the most time on. As an example, if a nurse who used to spend two hours on shift documentation, handover documentation, uses AI, now she has the time back for her direct patient care. or a revenue cycle analyst who used to manual review claims now reviews AI-generated summaries and applies that judgment to exception. Now the productivity unlock is real, but it requires leaders to intentionally redesign the entire workflow, not just drop another AI on top of one of the working tool. So to sum it up all, Pranthi, the shift from curiosity to Capability is ultimately a leadership decision. The technology is ready. The question is whether the organization is willing to commit, move with urgency and learn introduction rather than waiting for perfection. Prashanthi Nuthi: In fact, you touched a very great point, Ganesh, so people have asked me whether AI will replace my job. It's not the AI that will replace your job. AI is there to add to your productivity. If anybody is going to replace your job, it's another person who's leveraging AI. ⁓ whose productivity is better than you. So friends, here's the microboost for today. If you have not yet embraced AI, yet about a tool that you are curious about. Is it chat GPT? Is it cloud code? Is it perplexity? ⁓ start taking one step at a time. By the time you hit your 21st step, you will see how big of an impact it has made on you. remember small actions, big impact ⁓ And thank you Ganesh for being the guest today. Thank you. Ganesh Joshi Regmi: Thank you, Prashanthi. My pleasure. Prashanthi Nuthi: After 21 weeks of Thursday ThinkBites, I'm about to create a curated and trusted network of impact makers called the Impact BoostX Community.