System for the authorized generation of complete mobile applications from specification data using AI-supported transformation

Abstract

System for the automated generation of a mobile application, comprising the following steps: a) Reading in specification data in different formats, b) Converting the specification data into a structured model presentation using AI-supported analysis, c) Generating source code for frontend and backend based on the model, d) Generating automated test cases, e) Assembling a working mobile application.

Description

Technical field:

[0001] The invention relates to the field of software development, in particular systems for the automated generation of mobile applications and their backend structures from heterogeneous specification data. State of the art:

[0002] Some approaches to automating software development are known, such as Code4 generation, visual toolkits, or text-based prompt systems. However, these systems are limited to individual subtasks, such as creating user interfaces, generating source code from text, or creating test cases. A comprehensive system that implements the entire process from specification to a working mobile application in an integrated system is not yet known. Purpose of the invention:

[0003] The object of the invention is to provide a system that fully automates the creation of mobile applications, thereby significantly reducing development time and costs. Solution to the problem:

[0004] The task is solved by a system that reads in specification data in any format (text, tables, screenshots, speech output), transforms the data into a structured model representation using AI-supported analysis (NLP, OCR, computer vision, speech-to-text), automatically generates source code for fronted and backend from the model, derives automated test cases, and provides a working mobile application as a result. Example implementation:

[0005] A user provides a mix of Excel spreadsheets, text descriptions, and handwritten sketches. The system analyzes this input using OCR and NLP. A model structure (e.g., JSCN or UML) is created. From this, the system generates a user interface (e.g., React Native or Flutter). Database tables and API endpoints are automatically integrated. Test cases are generated and integrated into a CI / CD pipeline. The result is an installable application package for common platforms.

Claims

[1] System for the automated generation of a mobile application, comprising the steps: a) reading specification data in different formats, b) converting the specification data into a structured model presentation using AI-supported analysis, c) generating source code for frontend and backend based on the model, d) generating automated test cases, e) assembling a working mobile application. [2] System according to claim 1, wherein the analysis systems comprise at least one of the methods NLP, OCR, Computer Vision or Speech-to-Text. [3] System according to claim 1 or 2, wherein the model presentation is in UML, JSON or a comparable formal structure. [4] System according to any of the preceding claims, wherein the mobile application is generated in a cross-platform environment. [5] System according to any of the preceding claims, wherein backend structures including databases and AP endpoints are automatically generated. [6] System according to one of the preceding claims, wherein the generated test cases are automatically integrated into a CI / CD pipeline. [7] Application for implementing the system according to any one of claims 1 to 6, comprising modules for reading, analyzing, modeling, code generation, testing and deployment. [8] System according to any of the preceding claims , characterized by , that the Kl factory automatically reads in technical documents, in particular utility models and patent specifications with claims and specifications, analyzes their contents semantically and generates fully functional software products from them without manual source code input. [9] System according to claim 8, wherein formulas, abbreviations, drawings and reference numbers from the documents are converted into machine-readable structures (graphs, constains, function blocks) from whose architecture, program code, tests and security measures are derived. [10] System according to one of claims 8-9, wherein the generated software products have a reduction of the manually entered source code to at most 10% of the total source code of the application. [11] System according to any one of the preceding claims 8-10, characterized by , that the AI ​​factory reads in digital design data and technical models from mechanical engineering (especially CAD, CAE or PLC data), processes them semantically and automatically generates application software for controlling, monitoring, parameterizing or maintaining the respective machine. [12] System according to claim 11, wherein interfaces to industrial standards (including OPCUA, MQTT, Modbus, Profinet) are automatically detected, implemented and tested. [13] System according to one of claims 11-12, wherein the AI ​​factory additionally integrates a policy and compliance layer that ensures that the generated app complies with the applicable standards and guidelines in mechanical engineering (e.g. CE marking, ISO 13849, GDPR) and automatically generates compliant alternatives in case of violations. [14] System according to one of claims 8-13, wherein the self-healing component analyzes telemetry data from the runtime environment of the machine and / or app, detects error patterns and generates, tests and deploys software patches or optimizations without human intervention. [15] (Computer program product) Computer program product comprising program coding means which, when executed on a computer, perform the steps according to any one of claims 8-14.