AI Login Agent for Adaptive Web Authentication Testing

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing security testing tools struggle to address the variability in login implementations across web applications due to differences in login sequences, forms, and multi-factor authentication methods, which cannot be fully captured by heuristics-based rules.

Innovation Solution

A login agent interacts with a generative AI foundation model to automatically and intelligently log in to web applications by capturing interactive elements, constructing prompts to determine how to interact with these elements, and using user emulation tools to perform the interactions, with feedback from the model to assess login success or failure.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If heuristics-based rules are used to test login processes, then the testing process is simple to implement, but it cannot fully address the variability in login implementations across different web applications

Engineering Contradiction:

Improveease of implementing security testingVSAvoidability to handle login variations

Core Design Contradiction:

Ease of manufactureVSAdaptability or versatility

Solution Approach 1:

The patent replaces heuristics-based mechanical rule systems with an AI-powered system that uses machine learning models to analyze and adapt to different login implementations. The AI system processes screenshots and web page content to automatically determine login procedures, eliminating the need for manual rule configuration while achieving high adaptability across diverse web applications.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The security testing system performs self-learning and self-adjustment by analyzing screenshots and web page content to automatically determine login procedures. The AI models continuously improve their understanding of different login implementations without requiring external intervention or manual rule updates, enabling the system to adapt to new login methods autonomously.

Inventive Principle:

Principle #25Self-service

2Reliability

If manual testing of login processes is performed, then detailed and accurate testing can be achieved, but it requires significant manual effort and time

Engineering Contradiction:

Improveaccuracy of security testingVSAvoidtime required for testing

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent automates the manual testing process by using AI-powered systems that automatically capture screenshots, analyze web page content, and execute login procedures. The machine learning models process visual and textual information to determine appropriate login actions, replacing manual tester actions with automated AI-driven processes that maintain accuracy while significantly reducing time consumption.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The security testing system operates continuously without interruption, automatically capturing screenshots, analyzing login processes, and executing tests in a continuous workflow. The AI models process information in real-time as the system interacts with web applications, eliminating the intermittent nature of manual testing and maintaining consistent testing accuracy throughout the entire testing process.

Inventive Principle:

Principle #20Continuity of useful action

3Adaptability or versatility

If AI-powered automated login testing is implemented, then adaptability to different login variations is improved, but the system complexity increases

Engineering Contradiction:

Improveability to handle login variationsVSAvoidcomplexity of testing system

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent introduces AI models as intermediary components that bridge the gap between the testing system and diverse web application login implementations. These intermediary AI models process screenshots and web page content, translate them into actionable login procedures, and execute the tests, simplifying the overall system architecture by using specialized intermediate layers rather than complex direct processing.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The security testing system employs universal AI models that can handle multiple types of login implementations across different web applications through a single unified approach. The machine learning models are designed to be multi-functional, processing various screenshot formats, analyzing different web page structures, and adapting to diverse login procedures without requiring separate specialized systems for each login type.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentEP4703896A1Ai conversation driven login

Publication Date: 2026.03.04 VERACODE INC

Figure 1
Figure 2
Figure 3

AI summary

A login agent interacts with a foundation model(s) until successful login to an application or an assessment of a failed login can be obtained. Initially, a web page corresponding to login for a web application will be indicated to the login agent. The login agent captures interactive elements of the web page. The login agent prompts a foundation model(s) to select which of the captured interactive elements to interact with and how to interact with the selected elements. The login agent determines commands based on the response(s) and, with the commands, uses a tool to automatically interact with the web page via a browser. The login agent captures a web page resulting from the user emulated interaction and prompts the foundation model(s) to determine whether log in was successful or failed. The response from the foundation model(s) guides the login agent to either retry login or report results.