An individualized ai execution system and method based on cloud vectorization memory

Abstract

The application discloses a kind of individualized AI execution system and method based on cloud vectorization memory, it is related to artificial intelligence, cloud computing and automation execution field.The application constructs global user vector library in cloud, carries out unified vectorization coding, storage and management to the user data, behavior data, scene data collected to multiple terminals;Receive execution request from different terminals, generate global consistent individualized execution strategy based on cloud vector similarity retrieval;The individualized execution strategy is issued to the corresponding terminal, or directly completes API scheduling, process arrangement, task execution, RPA automation operation, automated payment, order settlement, aggregation payment processing in cloud;Task execution result and user feedback are updated to cloud global vector library synchronously, realize cross-device memory unification and iterative optimization.The application is suitable for multiple terminal cooperation, local life service, take-out delivery, payment aggregation, enterprise automation and the like scene, protection range covers cloud system, terminal, method and storage medium, infringement feature is clear, easy to take evidence, with strong commercial and right protection value.

Description

Technical Field

[0001] This invention relates to the fields of artificial intelligence, cloud computing, and automated execution technology, specifically to an application system and method that realizes vectorized storage and semantic retrieval of user historical data on a cloud server, and combines a cloud task planning engine and an automated execution unit to complete personalized and highly efficient automated task execution. Background Technology

[0002] This application is a further improvement and extension of the technical solution of the prior application with application number 2026102583924. That prior application discloses a technical solution that performs vectorized memory storage and semantic retrieval locally on the terminal device, and uploads only a simplified context to the cloud, effectively protecting privacy and reducing cloud costs.

[0003] However, the prior art solutions mainly target edge devices, and their memory storage and retrieval functions are limited by the terminal's computing power and storage space, making it difficult to handle massive amounts of user data and execute complex, personalized tasks across devices and on a large scale. Existing automated execution systems based on AI Agents and RPA typically employ the following methods to achieve personalized services: 1. Pure end-user processing: All data is stored and processed on the user's terminal device. While this can protect privacy, it is limited by hardware computing power and storage, making it difficult to handle massive amounts of data and perform complex tasks.

[0004] 2. Full data upload: Upload all user interaction history and business data to the cloud for unified processing. Although it can utilize the powerful computing power of the cloud, it has problems such as large data transmission volume, high bandwidth cost, and privacy leakage risk.

[0005] 3. Simple end-to-cloud collaboration: The terminal only acts as a data collector, and all data processing and execution are completed in the cloud. It fails to effectively utilize local terminal information for optimization, resulting in high cloud processing pressure and insufficient personalization.

[0006] Therefore, there is an urgent need for a technical solution that can fully utilize the powerful computing and storage advantages of the cloud while ensuring user data security and achieving efficient and personalized execution. This invention aims to migrate the core ideas of the prior art to the cloud, leveraging the powerful computing capabilities of the cloud to achieve more efficient and comprehensive personalized services, thus providing a powerful supplement and extension to the prior art. Summary of the Invention

[0007] (a) Technical problems to be solved This invention aims to address the technical problems faced by existing AI automated execution systems when utilizing cloud computing power, such as high privacy and data security risks, high cloud processing costs, and low personalized execution efficiency.

[0008] (II) Technical Solution A personalized AI execution system based on cloud-based vectorized memory, characterized in that it includes: The cloud-based memory system, deployed on a cloud server, includes: • A cloud-based vector database is used to store the original text of user historical interaction data and its corresponding feature vectors. The historical interaction data comes from multiple different terminal devices owned by the user, realizing unified user preference memory across devices; • The cloud-based semantic retrieval unit is used to convert the received user input data into a current feature vector, and retrieve the N historical interaction data with the highest semantic similarity from the cloud-based vector database to generate simplified context information.

[0009] Cloud-based execution systems include: • A personalized task planning engine is used to receive the simplified context information and dynamically generate a task execution path that matches the user's personalized needs based on the user's historical preference information therein. The task execution path is dynamically optimized according to the user's historical execution records. • An automated execution scheduling unit is used to complete automated operations based on the personalized task execution path, using methods including but not limited to RPA, API interface, intelligent agent, timed scheduling, script execution, workflow engine, and rule engine. The operations include but are not limited to enterprise-level office process automation, customer service system response, cross-platform data integration, and third-party service calls. • The execution feedback unit is used to store the execution status and final result data in the cloud vector database and feed them back to the user terminal.

[0010] A personalized AI execution method based on the above system includes the following steps: Step 1: The cloud-based memory system receives input data sent by the user through any terminal and converts it into the current feature vector; Step 2: The cloud-based semantic retrieval unit retrieves the N historical interaction data with the highest semantic similarity from the cloud-based vector database based on the current feature vector, and generates concise context information; Step 3: The personalized task planning engine plans a personalized task execution path based on the user's historical preferences in the simplified context; Step 4: The automated execution scheduling unit completes the automated operation according to the personalized task execution path; Step 5: The execution feedback unit stores the execution results in the cloud vector database and returns them to the user terminal.

[0011] (III) Beneficial Effects 1. Decoupling of computing power and data: By deploying a vectorized memory system in the cloud, centralized storage and efficient processing of user data can be achieved. The powerful computing resources of the cloud can be fully utilized to support the execution of complex and large-scale automated tasks, breaking through the physical limitations of edge devices.

[0012] 2. High efficiency in personalized execution: The cloud-based planning engine can dynamically optimize the execution path based on the user's complete historical data. Compared with edge-side solutions that rely on local data, the task planning is more accurate and efficient, and is especially suitable for scenarios that require long-term memory and complex decision-making.

[0013] 3. Data security and controllability: All user data is centrally stored in the cloud and managed by the system, which facilitates the implementation of high-standard data security and privacy protection measures and meets the stringent data compliance requirements of enterprise-level applications.

[0014] 4. Wide range of applications: This invention can not only serve individual users (such as cross-device personalized assistants), but also be widely used in enterprise-level services, such as enterprise knowledge base retrieval, process automation, customer service systems, etc., and is the core foundation for building cloud-based intelligent services.

[0015] 5. Full coverage of execution methods: Supports all mainstream automated execution technologies, with no blind spots in protection scope, and any automated execution method falls within the protection scope. Attached Figure Description Figure 1 This is a system architecture diagram of the cloud version of the present invention. Figure 2 This is a flowchart of the cloud-based method of the present invention. Detailed Implementation

[0016] Example 1: Enterprise-level cross-device personalization assistant An employee of a certain company uses multiple terminal devices such as mobile phones, computers, and tablets. The employee uses any of these devices to voice input: "Help me organize the meeting minutes from last Friday regarding Project A and send them to the team."

[0017] 1. Data Reception and Retrieval: The cloud-based memory system receives instructions, converts them into feature vectors, and then retrieves relevant historical interaction data such as "last Friday," "Project A," and "meeting minutes" from the cloud-based vector database.

[0018] 2. Planning and Execution: The personalized task planning engine plans the task path based on retrieved historical preferences: calling the office software API to generate a minutes document → retrieving the last sent list → calling the email API to send.

[0019] 3. Automated execution and feedback: The automated execution scheduling unit completes operations through API, RPA or Agent, and the execution results are returned to the employee terminal and synchronously stored in the cloud vector database.

[0020] Example 2: Cloud-based automated customer service The company integrates its customer service system with this invention.

[0021] 1. When a customer initiates an inquiry through a webpage or app, the cloud-based memory system retrieves historical inquiry and purchase records based on the customer's ID.

[0022] 2. The cloud-based semantic retrieval unit generates a concise context based on the current question.

[0023] 3. The personalized task planning engine plans the optimal response path.

[0024] 4. The automated execution scheduling unit is responsible for automatically retrieving data, filling out work orders, sending replies, and transferring to human agents, which greatly improves customer service efficiency.

Claims

1. A personalized AI execution system based on cloud-based vectorized memory, characterized in that, include: The cloud-based memory system, deployed on a cloud server, includes: A cloud-based vector database is used to store the original text and feature vectors of users' historical interaction data. The cloud-based semantic retrieval unit is used to retrieve the most relevant historical interaction data from the cloud-based vector database based on user input data and generate concise contextual information. Cloud-based execution systems include: A personalized task planning engine is used to dynamically generate personalized task execution paths based on the user's historical preference information in the simplified context. An automated execution scheduling unit is used to complete automated operations based on the personalized execution path, using methods including but not limited to RPA, API interfaces, intelligent agents, timed scheduling, script execution, workflow engines, and rule engines. These operations include, but are not limited to, enterprise-level office process automation, customer service system response, cross-platform data integration, and third-party service calls. The execution feedback unit is used to store the execution results in the cloud vector database and then provide feedback.

2. The system according to claim 1, characterized in that, The cloud-based semantic retrieval unit is specifically used to: convert user input data into a current feature vector, calculate the similarity with historical feature vectors in the cloud-based vector database, select the N historical interaction data with the highest similarity, and package them together with the current user input to form the simplified context information.

3. The system according to claim 1, characterized in that, The personalized task planning engine generates personalized task execution paths by dynamically adjusting the order of task execution, branch selection, or parameter configuration based on the user's historical operating habits, decision-making tendencies, preference settings, or historical execution results.

4. The system according to claim 1, characterized in that, The cloud-based vector database is used to store historical interaction data from multiple different terminal devices owned by the user, enabling unified user preference memory across devices.

5. The system according to claim 1, characterized in that, The user input data is collected by any type of user terminal device and sent to the cloud.

6. A personalized AI execution method based on cloud-based vectorized memory, applied to the system described in any one of claims 1-5, characterized in that, Includes the following steps: Step 1: The cloud-based memory system receives user input data and generates concise contextual information through the cloud-based semantic retrieval unit; Step 2: The personalized task planning engine generates personalized task execution paths based on the user's historical preferences in the simplified context; Step 3: The automated execution scheduling unit completes the automated operation according to the personalized execution path; Step 4: The execution feedback unit stores the execution results in the cloud vector database and returns them.

7. The method according to claim 6, characterized in that, The step 1 of generating simplified context information includes: converting user input data into the current feature vector, and retrieving the N historical interaction data with the highest semantic similarity from the cloud vector database, and packaging them together with the current user input.

8. The method according to claim 6, characterized in that, Step 2 generates a personalized task execution path, including dynamically optimizing the task execution order, branch selection, or parameter configuration based on the user's historical preferences.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the method described in any one of claims 6-8.

Images