I have been chatting with a few e-commerce friends for a long time these days. They are doing copywriting and planning on Taobao ppt、 For short video services, similar to the intermediary model, orders are placed on Taobao and corresponding designers and writers are hired to complete the delivery. Ai had a great impact on them, forcing them to learn from Ai Then a very practical problem arose, that adapting Ai to one's own business is far from simple. Most of the time is actually spent debugging Ai, but they are not good at this People in the cryptocurrency industry like to try new things, and it is not easy to truly reach the level of professional players. The large number of individual businesses that truly need AI to complete business optimization are finding it difficult to adapt to the arrival of the AI era When I saw the appearance of xUbble for @ dappOS_com, there was finally a project trying to solve this overlooked problem. Ai itself was meant to save everyone's time, but Ai learning itself requires a lot of time For example, a milk tea shop owner who is completely unfamiliar with video production and wants to create a promotional video for lemon tea can directly input on xUbble: "Create a 20 second video for me to promote our newly released lemon tea, which should look refreshing and summer like, accompanied by lively background music You can watch the video below for the finished video effect, which can be directly used for commercial purposes Why can giving a vague prompt word achieve this level? 2 sets of technical templates behind xUbble 1 Bubble Pilot Pilot is similar to an intelligent navigation recommendation. After the user writes a very short or even vague prompt word, Bubble Pilot will read the user's short request, identify the task type, and automatically match it from the SOP library, giving you the SOP with the highest matching degree The so-called SOP is the standard operating procedure, which is essentially a skill. For example, in this video demand for lemon tea, which is a specific beverage field, the product close-up, production process display, and atmosphere creation required for this type of video are similar. Targeted optimal templates can be used, and SOP is this standard template If there is no existing SOP, it will enter the universal standard path, and repeated rollback requests will become candidate signals for the next step of building SOP in Bubble Engine 2 Bubble Engine Engine is a SOP generation engine. For example, initially xUbble did not have SOPs for beverages. Engine, like an AI programmer, would automatically establish a standard, learn from competitors, and iterate repeatedly to solidify the optimal path into reusable SOPs As long as there is a demand for beverage videos in the future, the generated SOP can be directly used XUbble also has two operating environments Bubble Computer End to end project workspace. When Bubble Pilot detects a multi-step task, it routes it here, and the sandbox automatically starts and loads skills on demand, without the need for user management intermediate steps throughout the process. Can complete research, writing, design, verification, and deliver final results within a single run In most cases, computers can be used Bubble Personal Local environment mode. Can operate users' local files, browsers, applications, and schedules. Installation, download, and system level changes are executed in cloud containers and destroyed after task completion; Users can only perform operations that have been explicitly authorized on their own devices and do not need to install software For privacy related matters, such as some files and information that are highly private and do not trust any platform, then choose the Personal mode In fact, the core of xUbble's product is its SOP. For most demanders, they cannot effectively and accurately describe prompt words, nor do they understand terms such as close-up shots, slow motion, and product close ups. If they use the commonly used AI in the market, AI can only guess the requirements. Whatever you input to them, AI outputs, and the delivered product naturally does not meet expectations Input end: Fuzzy requirements Output end: Fuzzy product This is the advantage of xUbble. Users do not need to give Ai a precise requirement. If you want to make a beverage video, just one sentence is enough. Bubble Pilot will automatically find existing SOPs for beverages, and vague requirements can be found with precise SOPs to generate accurate products XUbble actually did one thing, identifying ambiguity, finding precise requirements, matching targeted SOPs, and generating precise products The pain point of AI's popularization is that it can continuously fool people and accurately extract vague human language on its own Experience link for xUbble: https://(dappos.com)/xbubble/zh-CN