Internal examination AI assistant implementation method and system based on large model and workflow

Abstract

The invention discloses an internal auditing AI assistant implementation method and system based on a large model and workflow. The method comprises the following steps: setting AI assistant information of AEO internal auditing; wherein the setting comprises setting an assistant prompt word, an assistant model, an assistant knowledge base and a preset workflow; adding external data account configuration required to be used for auditing into the assistant knowledge base; uploading the archived auditing data in a dialogue form, performing integrity judgment through a preset workflow and a large model, and replying an auditing opinion; if the external data is not complete, accessing the external data through the configuration, automatically collecting the data, and automatically archiving the data into the audited data; the method has the advantages that compared with an existing manual auditing mode, efficiency is greatly improved, and manpower and time cost is reduced; and meanwhile, external data can be accessed through configuration, and the data can be automatically archived into the audited data, so that the problem can be more effectively solved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method and system for implementing an internal audit AI assistant based on a large model and workflow. Background Art

[0002] The field of international trade is an important part of the global economy, involving the cross-border flow of goods and services as well as relevant laws, regulations and standards; companies or enterprises involved in it need to undergo relevant certification. Authorized Economic Operator (AEO) is a system in which the customs certifies and approves enterprises with high compliance, credit status and safety levels, and grants customs clearance convenience to certified enterprises.

[0003] The existing solutions mainly audit the import and export trade of enterprises through manual review. This method requires a lot of manpower and time. In addition, the collation and collection of enterprise data is also a huge task. The certification personnel of small and medium-sized enterprises often do not know what information should be submitted and need to conduct corresponding review consultation, which leads to the following problems:

[0004] First, in the policy consultation process, websites, public accounts, video promotion and other methods are often used, but this kind of information transmission is one-way. If you encounter something you don’t understand, you still need to go to the site for consultation or make a phone call for consultation;

[0005] Secondly, in the data collation stage, although most trading companies and trade management departments have already carried out information construction, the information is often distributed in different systems, and the applicants of small and medium-sized enterprises are not clear about which data should be downloaded and collected;

[0006] Furthermore, during the data review process, each applicant also checks each item one by one, which is inefficient and the amount of data is quite large. For example, some documents contain thousands of copies of data per year, and manual downloading is bound to result in omissions. Summary of the invention

[0007] In order to overcome the deficiencies in the prior art, the purpose of the present invention is to provide an internal audit AI assistant implementation method and system based on a large model and workflow.

[0008] The technical solution provided by the present invention is:

[0009] In the first aspect, a method for implementing an internal audit AI assistant based on a large model and workflow includes the following steps:

[0010] Set up AI assistant information for AEO internal audits; the settings include assistant prompt words, assistant models, assistant knowledge base, and preset workflows;

[0011] Add the external data account configuration required for the audit to the assistant knowledge base;

[0012] Upload archived audit materials through dialogue, make integrity judgments through preset workflows and big models, and respond to audit comments;

[0013] If it is incomplete, the external data is accessed through the configuration, the data is automatically collected, and the data is automatically archived into the data for this review.

[0014] Preferably, the method further comprises:

[0015] Perform automatic detection again. If automatic collection cannot meet the authentication requirements, the system will actively prompt the missing information and remind the user to upload it manually.

[0016] Preferably, the method further comprises:

[0017] Respond to user inquiries;

[0018] Perform similarity query on the question being consulted in the assistant knowledge base by using RAG technology;

[0019] Return the query results to the big model;

[0020] Combined with the query results, a large model is assembled and presented to the user.

[0021] Preferably, the method further comprises:

[0022] Receive individual files uploaded by users and review them against AEO standards;

[0023] Feedback the review results to the user.

[0024] Preferably, the method further comprises:

[0025] The data in the assistant knowledge base is vectorized using the vector model in the assistant model.

[0026] In a second aspect, the present invention further provides an internal audit AI assistant system based on a large model and workflow, comprising:

[0027] The setting module is used to set the AI ​​assistant information for AEO internal audit; the setting includes setting assistant prompt words, assistant model, assistant knowledge base and preset workflow;

[0028] An association module, used to add the external data account configuration required for the audit to the assistant knowledge base;

[0029] Audit module for:

[0030] Upload archived audit materials through dialogue, make integrity judgments through preset workflows and big models, and respond to audit comments;

[0031] If it is incomplete, the external data is accessed through the configuration, the data is automatically collected, and the data is automatically archived into the data for this review.

[0032] Through the above technical solution, the present invention can bring the following beneficial effects:

[0033] The present invention constructs an intelligent trade consultation Q&A application, which can automatically conduct preliminary review, respond to review opinions, and interact with users in the form of dialogue, so that users can more conveniently understand the specific information and improvement suggestions in the review opinions; compared with the existing manual review method, this greatly improves efficiency and reduces manpower and time costs;

[0034] At the same time, the present invention can also access external data through configuration, collect data, and automatically archive the data to the data of this audit. Compared with existing automation tools, this has stronger data processing capabilities and can more effectively identify and solve problems;

[0035] AI can be used to detect whether the data meets the requirements. If the data is incomplete, it can proactively indicate which data has gaps. This helps improve the fairness of trade audits and avoid deviations caused by human factors.

[0036] Vectorizing data through vector models facilitates data retrieval and analysis, and has strong data processing capabilities;

[0037] As well as being helpful to the development of small and medium-sized enterprises, the solution of the present invention can help small and medium-sized enterprises to collect and improve data, thereby increasing their competitiveness and helping to promote the fairness and balanced development of international trade. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 A flowchart of a method for implementing an internal audit AI assistant based on a large model and workflow provided by an embodiment of the present invention;

[0039] Figure 2 A schematic diagram of a service description provided by an embodiment of the present invention;

[0040] Figure 3 A process flow chart for setting up an AI assistant provided by an embodiment of the present invention;

[0041] Figure 4 A processing flow chart of a preset workflow provided by an embodiment of the present invention;

[0042] Figure 5A process flow chart of a user consultation provided by an embodiment of the present invention;

[0043] Figure 6 A processing flow chart of a single file review provided by an embodiment of the present invention;

[0044] Figure 7 A structural block diagram of an internal audit AI assistant system based on a large model and workflow provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0045] The specific embodiments of the present invention will be described in detail below. It should be noted that the embodiments described herein are only for illustration and are not intended to limit the present invention. In the following description, a large number of specific details are set forth in order to provide a thorough understanding of the present invention. However, it is apparent to those of ordinary skill in the art that these specific details need not be adopted to implement the present invention.

[0046] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0047] In the present application, the terms "upper", "lower", "left", "right", "front", "back", "top", "bottom", "inner", "outer", "middle", "vertical", "horizontal", "lateral", "longitudinal" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings. These terms are mainly used to better describe the present application and its embodiments, and are not used to limit the indicated devices, elements or components to have a specific orientation, or to be constructed and operated in a specific orientation.

[0048] In addition, some of the above terms may be used to express other meanings in addition to indicating orientation or positional relationship. For example, the term "on" may also be used to express a certain dependency or connection relationship in some cases. For those of ordinary skill in the art, the specific meanings of these terms in this application can be understood according to specific circumstances.

[0049] Throughout the specification, references to "one embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in conjunction with the embodiment or example is included in at least one embodiment of the present invention. Therefore, the phrases "in one embodiment," "in an embodiment," "an example," or "an example" appearing in various places throughout the specification do not necessarily all refer to the same embodiment or example. Furthermore, particular features, structures, or characteristics may be combined in one or more embodiments or examples in any appropriate combinations and / or subcombinations.

[0050] It should be noted that, unless otherwise specified, the technical terms in this embodiment have the common meanings understood in the relevant technical field.

[0051] like Figures 1 to 4 As shown, an implementation method of an internal audit AI assistant based on a large model and workflow provided by an embodiment of the present invention includes the following steps:

[0052] S101, setting AI assistant information for AEO internal audit; wherein the setting includes setting assistant prompt words, assistant model, assistant knowledge base and preset workflow;

[0053] S102, adding the external data account configuration required for the audit to the assistant knowledge base;

[0054] S103, upload the archived audit materials through dialogue, make integrity judgment through preset workflow and big model, and reply the audit opinions;

[0055] S104, if incomplete, access external data through the configuration, automatically collect data, and automatically archive the data into the data for this review.

[0056] When applying, refer to Figure 3 The settings include setting the AI ​​assistant information, including the name, icon and description; selecting the big model used by the AI ​​assistant, setting the big model prompt words according to the business type, creating the knowledge base required by the AI ​​assistant, uploading the AEO audit-related knowledge data to the knowledge base, setting the permissions and addresses that the AI ​​assistant needs to access the external system, and setting the internal audit workflow, and finally saving the settings.

[0057] The assistant prompts include but are not limited to "Please provide the information that needs to be reviewed" and "Please supplement the relevant information"; the assistant model includes but is not limited to text analysis models, speech recognition models and natural language processing models; the assistant knowledge base includes but is not limited to rule base, policy and regulation base, AEO training course base and internal enterprise knowledge base; the preset workflow includes but is not limited to document review workflow, data collection workflow and problem answering workflow; the knowledge base in the attached figure has the same meaning as the assistant knowledge base.

[0058] In this embodiment, the external data account configuration required for the audit is added to the assistant knowledge base; the configuration includes account number, access rights and address, etc.; external data refers to the corresponding external system, including the single window system, internal enterprise ERP system, financial system and tax system, etc. The address of the system is configured in the library.

[0059] In this embodiment, the large model is a generative large model; the review opinion includes pass or fail;

[0060] Workflow technology is used to coordinate the relationship between different systems and data, so as to effectively organize information and complete the coordination of complex transactions. The workflow also classifies files and establishes a submission review archive. The preset rules in the rule library are first used to detect whether the file is missing. The preset rules include but are not limited to the integrity of the file type, whether the file time is consistent with the review cycle, whether the business data and business documents correspond, etc. If missing, the missing data will be read through the big model to read the policy documents in the knowledge base to check whether the company needs to submit such documents. Finally, the data will be sent to the big model via text. The big model analyzes which type of data is missing, and then calls the external system for data collection through the preset workflow. Finally, the collected data will be classified into the file archive submitted by the user next time.

[0061] Furthermore, in the audit policy query, the vector model in the assistant model is also used to vectorize the data in the assistant knowledge base.

[0062] When applied, the public policies, regulations, and AEO training courses are converted into text, and the data is vectorized through a vector model to facilitate data retrieval and analysis, effectively store and manage data, and improve data utilization. Among them, vector models include but are not limited to Word2Vec, GloVe, FastText, etc., and vector databases include but are not limited to ElasticSearch, Solr, Apache Cassandra, etc.

[0063] In another embodiment, based on the above solution, the method further includes:

[0064] Perform automatic detection again. If automatic collection cannot meet the authentication requirements, the system will actively prompt the missing information and remind the user to upload it manually.

[0065] Specifically, combined Figure 4 First, according to policy requirements, the system will provide an AEO audit template archive based on the annual audit requirements. Users can create an audit archive folder for their company this year based on the template;

[0066] By first receiving the preliminary review file uploaded by the user (i.e., archived review materials), the file is first checked through the rule base to see if it complies with the rules;

[0067] If not, the files are analyzed through LLM to see which are missing, and then the business system data is collected according to the configuration and then automatically archived to the archive;

[0068] The workflow also has a corresponding process for this situation; that is, when it is found that there is no data that can be automatically collected in the rules or the business system is uncertain, the user is prompted to upload manually; the specific process is shown in the attached figure and will not be described in detail here;

[0069] When applied, if automatic collection cannot meet the relevant certification requirements, information will be proactively sent to the person in charge of the enterprise to prompt which data have gaps; through automatic detection again, it can be ensured that the data meets the certification requirements; among them, automatic detection includes but is not limited to data integrity detection, data accuracy detection and data consistency detection.

[0070] At the same time, after the test is passed, the information is downloaded and uploaded for review; by having the user upload additional or modified information, AI will test again to ensure data accuracy. After the test is passed, the information is downloaded and uploaded for review, which can effectively complete the review work.

[0071] If not, LLM is used to analyze which files are missing, and then business system data is collected according to the configuration and automatically archived to the directory library.

[0072] Furthermore, in order to facilitate users to understand and consult the information; refer to Figure 5 , the method further comprises:

[0073] Respond to user inquiries;

[0074] Perform similarity query on the question being consulted in the assistant knowledge base by using RAG technology;

[0075] Return the query results to the big model;

[0076] Combined with the query results, a large model is assembled and presented to the user.

[0077] Furthermore, users can ask questions about the missing departments, and the AI ​​assistant can give answers, thereby effectively solving user problems and improving user experience.

[0078] Furthermore, the method further comprises:

[0079] Receive a single file uploaded by the user, parse the file through the big model, call the big model to perform AEO standard audit analysis on it; feedback the audit results to the user; refer to the specific process Figure 6 .

[0080] The advantage of this is that after the document is compiled or generated, you will know whether it meets the AEO standards and meet the relevant certification requirements in advance.

[0081] The above solution, by building an intelligent trade consultation Q&A application, can automatically conduct preliminary review, respond to review opinions, and interact with users in the form of dialogue, so that users can more easily understand the specific information and improvement suggestions in the review opinions; compared with the existing manual review method, this greatly improves efficiency and reduces manpower and time costs;

[0082] At the same time, the present invention can also access external data through configuration, collect data, and automatically archive the data to the data of this audit. Compared with existing automation tools, this has stronger data processing capabilities and can more effectively identify and solve problems;

[0083] AI can be used to detect whether the data meets the requirements. If the data is incomplete, it can proactively indicate which data has gaps. This helps improve the fairness of trade audits and avoid deviations caused by human factors.

[0084] Vectorizing data through vector models facilitates data retrieval and analysis, and has strong data processing capabilities;

[0085] As well as being helpful to the development of small and medium-sized enterprises, the solution of the present invention can help small and medium-sized enterprises to collect and improve data, thereby increasing their competitiveness and helping to promote the fairness and balanced development of international trade.

[0086] Based on the same inventive concept, the embodiment of the present invention also provides an internal audit AI assistant system based on a large model and workflow, referring to Figure 7 ,include:

[0087] The setting module is used to set the AI ​​assistant information for AEO internal audit; the setting includes setting assistant prompt words, assistant model, assistant knowledge base and preset workflow;

[0088] An association module, used to add the external data account configuration required for the audit to the assistant knowledge base;

[0089] Audit module for:

[0090] Upload archived audit materials through dialogue, make integrity judgments through preset workflows and big models, and respond to audit comments;

[0091] If it is incomplete, the external data is accessed through the configuration, the data is automatically collected, and the data is automatically archived into the data for this review.

[0092] The entire system uses workflow technology to coordinate the relationship between different systems and data, so as to effectively organize information and coordinate complex transactions. The workflow also classifies files and establishes a submission review archive. It first detects whether the file is missing through the preset rules in the rule library. The preset rules include but are not limited to the integrity of the file type, whether the file time is consistent with the review cycle, whether the business data and business documents correspond, etc. If missing, the missing data will be read through the big model to read the policy documents in the knowledge base to check whether the company needs to submit such documents. Finally, it will be sent to the big model via text. The big model analyzes which type of data is missing, and then calls the external system for data collection through the preset workflow. Finally, the collected data will be classified into the file archive submitted by the user next time.

[0093] In this embodiment, the audit module is also used to:

[0094] Perform automatic detection again. If automatic collection cannot meet the authentication requirements, the system will actively prompt the missing information and remind the user to upload it manually.

[0095] Furthermore, the audit module is also used to:

[0096] Receive individual files uploaded by users and review them against AEO standards;

[0097] Feedback the review results to the user.

[0098] In another embodiment, based on the above technical solution, the internal audit AI assistant system based on the big model and workflow further includes a consulting module, which is used to:

[0099] Respond to user inquiries;

[0100] Perform similarity query on the question being consulted in the assistant knowledge base by using RAG technology;

[0101] Return the query results to the big model;

[0102] Combined with the query results, a large model is assembled and presented to the user.

[0103] It should be noted that for a more specific description of the workflow of the system embodiment, please refer to the aforementioned method embodiment part, which will not be repeated here.

[0104] The above technical solution solves the problem of lack of interaction in policy information by introducing generative AI technology. AEO certification data is input into the big model through the knowledge base. When users encounter problems understanding traditional information, they can communicate with the big model through the chat mode, allowing users to understand the AEO certification policy in a more interactive way.

[0105] Solve the problem of low data review efficiency: By allowing AI to read the rules and compare the differences between the documents to be submitted and the data required by the policy, the review can be completed quickly, and AI can be used to explain the policy basis for which areas need to be modified;

[0106] Solve the problem of data being too scattered and difficult to collect; by introducing workflow technology, let AI automatically collect and complete data from various scattered business systems based on missing data.

[0107] In the several embodiments provided in the present application, it should be understood that the disclosed internal audit AI assistant implementation method and system based on large models and workflows can be implemented in other ways. For example, the embodiments described above are merely schematic. For example, the division of the modules may be divided in other ways in actual implementation, such as multiple units or components may be combined or integrated into another system or device, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interfaces, devices or units, or may be electrical, mechanical or other forms of connection.

[0108] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for implementing an internal audit AI assistant based on a large model and workflow, characterized in that: The following steps are involved: Set up AI assistant information for AEO internal audits; the settings include assistant prompt words, assistant models, assistant knowledge base, and preset workflows; Add the external data account configuration required for the audit to the assistant knowledge base; Upload archived audit materials through dialogue, make integrity judgments through preset workflows and big models, and respond to audit comments; If it is incomplete, the external data is accessed through the configuration, the data is automatically collected, and the data is automatically archived into the data for this review.

2. According to claim 1, a method for implementing an internal audit AI assistant based on a large model and workflow is characterized in that: The method further comprises: Perform automatic detection again. If automatic collection cannot meet the authentication requirements, the system will actively prompt the missing information and remind the user to upload it manually.

3. According to claim 2, a method for implementing an internal audit AI assistant based on a large model and workflow is characterized in that: The method further comprises: Respond to user inquiries; Perform similarity query on the question being consulted in the assistant knowledge base by using RAG technology; Return the query results to the big model; Combined with the query results, a large model is assembled and presented to the user.

4. According to the method for implementing an internal audit AI assistant based on a large model and workflow according to claim 2 or 3, it is characterized in that: The method further comprises: Receive individual files uploaded by users and review them against AEO standards; Feedback the review results to the user.

5. According to claim 4, a method for implementing an internal audit AI assistant based on a large model and workflow is characterized in that: The method further comprises: The data in the assistant knowledge base is vectorized using the vector model in the assistant model.

6. An internal audit AI assistant system based on a large model and workflow, characterized in that: include: The setting module is used to set the AI ​​assistant information for AEO internal audit; the setting includes setting assistant prompt words, assistant model, assistant knowledge base and preset workflow; An association module, used to add the external data account configuration required for the audit to the assistant knowledge base; Audit module for: Upload archived audit materials through dialogue, make integrity judgments through preset workflows and big models, and respond to audit comments; If it is incomplete, the external data is accessed through the configuration, the data is automatically collected, and the data is automatically archived into the data for this review.

7. According to claim 6, an internal audit AI assistant system based on a large model and workflow is characterized in that: The audit module is also used to: Perform automatic detection again. If automatic collection cannot meet the authentication requirements, the system will actively prompt the missing information and remind the user to upload it manually.

8. According to claim 7, an internal audit AI assistant system based on a large model and workflow is characterized in that: Also included is a consulting module, the consulting module being used to: Respond to user inquiries; Perform similarity query on the question being consulted in the assistant knowledge base by using RAG technology; Return the query results to the big model; Combined with the query results, a large model is assembled and presented to the user.

9. According to claim 7 or 8, an internal audit AI assistant system based on a large model and workflow is characterized in that: The audit module is also used to: Receive individual files uploaded by users and review them against AEO standards; Feedback the review results to the user.

Images