@Sangeet - very comprehensive and builds nicely on your previous work. At some point you may want to get to the question of variance of doing the same things across enterprises and how it will impact this framework - because why services should even exist. Why isn't it all software even today. I think it is more than the managerial role that rebundle the work. Will be curious to get your ideas on it
Good point. At a first pass, I expect that this leads to one of three scenarios:
1. Some workflows will lend themselves to standardization and clients won't care about variance if economics and risk sharing are better with standardization
2. Some workflows will only be delivered as managed services i.e. services wrapped around software, and competition will shift to a margin improvement game based on how much your model can expand the margin on the service
3. For some workflows, the variance will shift upstream or downstream and continue as a service bottleneck. The question then is (1) whether minor configuration can solve for the variance and (2) whether the service bottleneck sits in a valuable part of the workflow.
@Sangeet, if several critical service bottlenecks are identified across multiple client workflows, how do you prioritize which bottleneck to address first? Furthermore, how do you ensure that the solutions are standardized and scalable, rather than remaining as custom implementations?
Fantastic approach and thought process Sangeet, appreciate you taking the time out to explain. I do have 1 question - is it not hard for companies to define what a job looks like? At the end of the day, the job is performed by a role (human) and it is not easy to identify the tasks, purely due to the complexity in which large enterprises function.. is it not more advisable to think in terms of business workflows, where more documentation/guidance is available to understand which tasks within a workflow can be automated vs augmented? Sorry if I am missing the point.
Hi @Sangeet, Thank you for your comprehensive post. Is it possible to include at least one case study and run through all your methodologies within it? I think this might be more convincing. I am currently working on a GenAI efficiency improvement plan for a wealth management company, and I am not very sure if your approach can really work in such a highly specialized field..
Hey @Sangeet, thanks a lot for putting this together. While evaluating the opportunities in this space I have formulated a view that we would need a Services + Product Model to be able to deliver value to Enterprises/SMBs etc. You have given a great name to this Services as a Software.
Would love check your views on AI native ERPs/ Enterprise system like how cloud native systems kicked in? Do you see potential in this? The reason I am highlighting this we can now re-imagine the entire systems esp. the UX + DBs given that the enterprise data is to be leveraged for new AI native systems.
I believe we're going to see a new AI-native architecture alongside one where AI is co-opted into existing cloud-native architecture. Cloud-native players who've invested in building knowledge graphs over the past several years but not easily figured a way to deliver that value into the UX can now do it much better with AI.
Unlike the cloud, where value accumulated at horizontal layers and created higher modularity, I feel AI is moving us into vertical integration because there are clear benefits to greater integration across compute, data, model, UX etc. That, more than anything else, will yield an AI-native architecture because the best AI applications will not be layers on top, they will be deeply and vertically integrated.
@Sangeet - very comprehensive and builds nicely on your previous work. At some point you may want to get to the question of variance of doing the same things across enterprises and how it will impact this framework - because why services should even exist. Why isn't it all software even today. I think it is more than the managerial role that rebundle the work. Will be curious to get your ideas on it
Good point. At a first pass, I expect that this leads to one of three scenarios:
1. Some workflows will lend themselves to standardization and clients won't care about variance if economics and risk sharing are better with standardization
2. Some workflows will only be delivered as managed services i.e. services wrapped around software, and competition will shift to a margin improvement game based on how much your model can expand the margin on the service
3. For some workflows, the variance will shift upstream or downstream and continue as a service bottleneck. The question then is (1) whether minor configuration can solve for the variance and (2) whether the service bottleneck sits in a valuable part of the workflow.
@Sangeet, if several critical service bottlenecks are identified across multiple client workflows, how do you prioritize which bottleneck to address first? Furthermore, how do you ensure that the solutions are standardized and scalable, rather than remaining as custom implementations?
Fantastic approach and thought process Sangeet, appreciate you taking the time out to explain. I do have 1 question - is it not hard for companies to define what a job looks like? At the end of the day, the job is performed by a role (human) and it is not easy to identify the tasks, purely due to the complexity in which large enterprises function.. is it not more advisable to think in terms of business workflows, where more documentation/guidance is available to understand which tasks within a workflow can be automated vs augmented? Sorry if I am missing the point.
Regards,
Dev.
No, you're absolutely right. Eventually, it is about workflows and the sequence of tasks through which they are executed. This article is probably a better explanation of how these views fit together: https://platforms.substack.com/p/ai-wont-eat-your-job-but-it-will-880
Hi @Sangeet, Thank you for your comprehensive post. Is it possible to include at least one case study and run through all your methodologies within it? I think this might be more convincing. I am currently working on a GenAI efficiency improvement plan for a wealth management company, and I am not very sure if your approach can really work in such a highly specialized field..
Hey @Sangeet, thanks a lot for putting this together. While evaluating the opportunities in this space I have formulated a view that we would need a Services + Product Model to be able to deliver value to Enterprises/SMBs etc. You have given a great name to this Services as a Software.
Would love check your views on AI native ERPs/ Enterprise system like how cloud native systems kicked in? Do you see potential in this? The reason I am highlighting this we can now re-imagine the entire systems esp. the UX + DBs given that the enterprise data is to be leveraged for new AI native systems.
I believe we're going to see a new AI-native architecture alongside one where AI is co-opted into existing cloud-native architecture. Cloud-native players who've invested in building knowledge graphs over the past several years but not easily figured a way to deliver that value into the UX can now do it much better with AI.
Unlike the cloud, where value accumulated at horizontal layers and created higher modularity, I feel AI is moving us into vertical integration because there are clear benefits to greater integration across compute, data, model, UX etc. That, more than anything else, will yield an AI-native architecture because the best AI applications will not be layers on top, they will be deeply and vertically integrated.