How to use AI to redo existing

shafi654@gmail.

It is just a reasonable inference based on the content shared by his team. If there is any misleading or wrong information, please point it out in time. You may ask: It’s not even online, you haven’t experienced it, and you’re not an actual participant, so why are you sharing it? The reason is to ensure the comprehensiveness of the case and avoid limiting thinking. The author's previous idea of ​​using AI to redo products was preconceived by


Microsoft Copilot, Youzan's intelligent operation Afghanistan WhatsApp Number assistant and other products, and he believed that the best product form (or even the only product form) can only be a full-featured set. Central type, while ignoring the embedded link nodes, as well as the product positioning and product form of different assistants. Summary and inspiration ) Judging from current information, the biggest consensus among almost all companies that are willing to “all in AI” (such as Microsoft, DingTalk, Youzan, etc.) is to follow the implementation plan of “remaking existing products with AI”. Instead of using AI to create a new product. )  products (including B-side and C-side)? Currently, there are two relatively mature





product architectures, corresponding to two different product forms. ) Regarding how SaaS products implement AI (including implemented scenarios), it is still relatively early, investment is not enough, and product architecture and positioning are not particularly clear. ) If SaaS products are to be implemented in AI, the above analysis shows that it can still follow two product architectures and two product forms. Early stage: [Link node embedded] can be implemented first to empower existing product capabilities. At the same time, a [full-featured centralized center] approach is adopted, with only customer service assistants implemented in the early stage; Mid-term: Reorganize, position and integrate various internal AI assistants;