Although the quality of the underlying

shafi654@gmail.

One reason is the limitations of large language models themselves. Organizing business knowledge into corpus for training large models is a huge project, because a lot of knowledge is tacit and not documented or recorded in writing, but is passed on by word of mouth or face-to-face communication. Organizing this tacit knowledge into training corpus that the model can understand and digest is a major task. model is good at the level of general transactions and knowledge application, the reality is very skinny when it


comes to enterprise-level applications and creating new products. If the enterprise's business knowledge is not effectively transformed into the context required by the model, then using AI for business processing in daily work will not yield the desired results. B Afghanistan WhatsApp Number product managers build products based on the company's business activities and business pain points. In the current product development process, product managers abstract and summarize business logic, and then provide it to R&D personnel for implementation in products.





My failed product As a product manager, my team and I have tried various application scenarios of AI products on the company’s self-developed business platform. It is a pity that only a few succeed. It can be said that you sow but reap little. From a product success perspective, it is indeed lackluster. But from an engineering perspective, although our repeated failures cannot tell us what success is, it can at least tell us under what circumstances some products that appear to be very "correct" and "valuable" on the surface are unsuccessful.