Post by account_disabled on Feb 25, 2024 1:04:27 GMT -6
Once we achieve scale, we can maintain good gross profit margins, and only then can we have money to invest in research and development and advertising, so that more people can use it and form a positive cycle. Is it necessary to have LLM capabilities to make super applications? To put the question another way, can we make super applications with LLM capabilities? Neither. Super applications are inherently difficult, and consumer electronics is also competing in all aspects. It is better to focus on things we can focus on. Apple did not manufacture chips for Day It first used Intel, and then spent years replacing that chip with its own MC chip. Therefore, consumer electronics is an all-round competition.
Don’t think that LLM’s awesomeness is everything. Private data is very important, and it is especially important to give LLM enough context. Browser plug-ins are a good form. It depends on what the user has said in the past, what Colombia Phone Number List he has checked, what he has saved... (Of course, even this may still not win). Perplexity’s product design capabilities Perplexity's underlying search uses Bing's API, and result processing uses GPT's API. It is a pure shell. His users are one ten thousandth that of Google, but he is often mentioned in this category, which proves that he has a certain imagination in terms of product design. Its search box is a text area, which is multi-line, while Google is single-line. The implication behind it is that Google search may require users to enter a phrase or a short sentence;
while perplexity allows users to enter a long paragraph, which can help you break down the matter and help you do a good search. It is more in line with human expectations. Problem imagining. Second, it is very AI Native and has no burden. of links. Even if there are generated answers, the links will always be in the main position. The design of Perplexity is to publish a long answer first, and the links are in the form of small annotations. The answers come from these places, and you can trace the source by yourself. From the perspective of product design, it does not have the burden of traditional search bidding rankings or advertising sales. It can redefine its product design.
Don’t think that LLM’s awesomeness is everything. Private data is very important, and it is especially important to give LLM enough context. Browser plug-ins are a good form. It depends on what the user has said in the past, what Colombia Phone Number List he has checked, what he has saved... (Of course, even this may still not win). Perplexity’s product design capabilities Perplexity's underlying search uses Bing's API, and result processing uses GPT's API. It is a pure shell. His users are one ten thousandth that of Google, but he is often mentioned in this category, which proves that he has a certain imagination in terms of product design. Its search box is a text area, which is multi-line, while Google is single-line. The implication behind it is that Google search may require users to enter a phrase or a short sentence;
while perplexity allows users to enter a long paragraph, which can help you break down the matter and help you do a good search. It is more in line with human expectations. Problem imagining. Second, it is very AI Native and has no burden. of links. Even if there are generated answers, the links will always be in the main position. The design of Perplexity is to publish a long answer first, and the links are in the form of small annotations. The answers come from these places, and you can trace the source by yourself. From the perspective of product design, it does not have the burden of traditional search bidding rankings or advertising sales. It can redefine its product design.