LLM Access Review Proposals for Faster Permission Audits

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Large organizations face inefficiencies and errors in manual access permission reviews due to the accumulation of vast data and frequent changes in user roles and positions, leading to time-consuming, effort-consuming, and error-prone processes.

Innovation Solution

An LLM-based system automatically collects organizational data, constructs triggers, and generates proposals for access permission modifications, providing reasoning and evidence to assist resource owners in timely and accurate reviews.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If manual access permission reviews are performed, then resource owners can make informed decisions about access permissions, but the process becomes time-consuming and effort-consuming

Engineering Contradiction:

Improveaccuracy of access permission reviewVSAvoidtime required for access permission review

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

An automated access review system acts as an intermediary between resource owners and access permission management. The system collects organizational data, analyzes access patterns, generates review proposals with reasoning, and presents them to resource owners for approval. This intermediary automation handles the time-consuming data collection and analysis tasks while preserving human decision-making authority, thereby reducing time loss without compromising review reliability.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If manual access permission reviews are performed, then access permissions can be reviewed with contextual understanding, but the process is error-prone due to human fatigue

Engineering Contradiction:

Improveaccuracy of access permission reviewVSAvoidoperational simplicity of access permission review

Core Design Contradiction:

ReliabilityVSEase of operation

Solution Approach 1:

The access review system performs self-service by automatically collecting organizational data, analyzing access patterns, identifying potential issues, and generating review proposals with supporting reasoning. This automation eliminates manual data collection and analysis tasks that are error-prone due to human fatigue, while the system maintains operational simplicity by presenting structured proposals to resource owners who only need to review and approve or reject the generated recommendations.

Inventive Principle:

Principle #25Self-service

3Productivity

If automated access review systems are implemented, then time and effort are saved, but the system complexity increases

Engineering Contradiction:

Improveefficiency of access permission reviewVSAvoidsystem complexity for access permission review

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The automated access review system is segmented into distinct functional modules: data collection module that gathers organizational data, analysis module that processes access patterns, proposal generation module that creates review recommendations with reasoning, and approval module that presents options to resource owners. This segmentation manages system complexity by organizing functions into separate, manageable components while maintaining high productivity through automated end-to-end processing.

Inventive Principle:

Principle #1Segmentation

4Measurement precision

If comprehensive organizational data is collected for access review, then review accuracy is improved, but data collection and processing time increases

Engineering Contradiction:

Improveprecision of access permission assessmentVSAvoidtime for data collection and processing

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system performs preliminary actions by continuously collecting and pre-processing organizational data in the background before access reviews are needed. Data collection, cleaning, and initial analysis are conducted in advance, so when an access review is triggered, the system can quickly retrieve pre-processed data and generate proposals without time-consuming data collection during the actual review process, thereby achieving both high precision and fast processing.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS20260073028A1Automatic Performance of Access Review Utilizing a Large Language Model (LLM)

Publication Date: 2026.03.12 VARONIS SYSTEMS INC

AI summary

Computerized systems and methods for automated Access Permissions Review in an organization. A method includes: (a) receiving a textual description of an Access Review Policy; (b) constructing a trigger for automatically initiating an Access Review process; (c) automatically collecting organizational data that pertains to: (i) roles and positions of users, (ii) user-specific characteristics, (iii) email messages sent and received by users, (iv) description of events in which users had accessed elements of the Organizational Resource. Upon a triggering of the trigger, the method provides to a fine-tuned Large Language Model (LLM), the Access Review Policy and augmentation data from the organizational data, and a prompt instructing the LLM to generate a proposal to revoke, modify, add or maintain the Access Permissions of a particular user, and further instructing the LLM to generate output that describes reasoning and supporting evidence for the proposal.