I have to disagree with the basic your premise of your 10.18.24 post, which (as I read it) asserts that AI will be able to subsume a business application's capabilities and functionality sometime in the near future. AI and generative AI will be able to a lot of things, especially in assessing and managing unstructured data, but the technology is nowhere near ready to deal with the complexity and ambiguity of enterprise-level processes currently handled within the structure of a specific business application. Decades of development have gone into these core applications to be able to deal with the mission critical details and nuances. Agents can do a lot of things, but we are about to discover their practical limits as well as the growing chore of curating and maintaining them. Sure, Salesforce might be more easily replaced by AI than an ERP or supply chain management application where 'sort-of' and 'on-the-job training' doesn't work for critical processes. Even so, an agentic Salesforce (already announced and on its way) will accomplish the same thing and available many years before a generic AI developer can have something better to offer. The key difference between AI and past disruptive technologies is that AI is an additive technology for business software, not a foundational one like GUIs or the cloud.
The emphasis on who UI's are built for is a super interesting point. At the end of the day, it makes sense to me that humans tell their preferences at regular intervals to their own personal UI (Call it a personal AI), and then that AI works off that to interface with all different applications/UIs. My example from today is how google now fills in your credit card information, or one click shopping
I have to disagree with the basic your premise of your 10.18.24 post, which (as I read it) asserts that AI will be able to subsume a business application's capabilities and functionality sometime in the near future. AI and generative AI will be able to a lot of things, especially in assessing and managing unstructured data, but the technology is nowhere near ready to deal with the complexity and ambiguity of enterprise-level processes currently handled within the structure of a specific business application. Decades of development have gone into these core applications to be able to deal with the mission critical details and nuances. Agents can do a lot of things, but we are about to discover their practical limits as well as the growing chore of curating and maintaining them. Sure, Salesforce might be more easily replaced by AI than an ERP or supply chain management application where 'sort-of' and 'on-the-job training' doesn't work for critical processes. Even so, an agentic Salesforce (already announced and on its way) will accomplish the same thing and available many years before a generic AI developer can have something better to offer. The key difference between AI and past disruptive technologies is that AI is an additive technology for business software, not a foundational one like GUIs or the cloud.
The emphasis on who UI's are built for is a super interesting point. At the end of the day, it makes sense to me that humans tell their preferences at regular intervals to their own personal UI (Call it a personal AI), and then that AI works off that to interface with all different applications/UIs. My example from today is how google now fills in your credit card information, or one click shopping
Would be great to add TRST.L to the comp set