Enterprise Edge Model Rollout Without User Data Exhaust
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems struggle to optimize application and service experiences while maintaining user privacy compliance, as they often require sending user data to centralized services for analysis, which can lead to data mishandling and violate privacy policies.
Innovation Solution
A system that optimizes application and service experiences within an enterprise by using a centralized software service to generate models, testing them on a first cohort of users, and applying the optimized models to a second cohort without sending user data back to the cloud, thereby maintaining privacy compliance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If user data is sent to centralized cloud services for analysis and optimization, then application and service experiences can be optimized through learned models, but user privacy compliance is violated and data security risks increase
Solution Approach 1:
The patent introduces an enterprise metrics database as an intermediary that receives and stores metrics from applications without transmitting user data to cloud services. This intermediary layer enables optimization analysis to be performed locally within the enterprise boundary, maintaining privacy compliance while still allowing learned models to be generated and applied for improving application experiences.
Solution Approach 2:
The patent shifts the optimization paradigm from centralized cloud-based analysis to distributed edge-based analysis within enterprises. By moving the learned model generation and application optimization process from the cloud dimension to the enterprise edge dimension, the system maintains optimization effectiveness while ensuring user data remains within secure enterprise boundaries.
2Productivity
If metrics are continuously monitored and sent to cloud services for analysis, then application experiences can be improved through feedback loops, but network bandwidth consumption increases and power usage rises
Solution Approach 1:
The patent enables enterprises to perform self-service optimization by generating and applying learned models locally using their own metrics data stored in the enterprise metrics database. This eliminates the need for continuous cloud communication for model generation and optimization, allowing enterprises to maintain improved application experiences while significantly reducing network bandwidth consumption and power usage associated with constant cloud connectivity.
3Device complexity
If rules are hardcoded within applications with limited intelligence, then application simplicity is maintained, but optimization capability and adaptability are insufficient
Solution Approach 1:
The patent pre-generates learned models based on historical metrics and user behavior patterns stored in the enterprise metrics database, and stores these models for later application to multiple applications. This preliminary action allows the system to maintain simple hardcoded rule structures in applications while incorporating sophisticated optimization capabilities through pre-computed learned models that can be applied without increasing application complexity.
Solution Approach 2:
The patent creates copied versions of learned models that can be applied across multiple applications and devices within an enterprise. Instead of embedding complex optimization logic in each application, the system generates a learned model once and copies it to relevant applications, maintaining simplicity while providing advanced optimization capabilities through the copied model data.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Systems and methods for optimizing application experiences on devices in an enterprise while maintaining privacy compliance are provided. An enterprise model management service of the enterprise accesses, from a centralized software service, a model comprising one or more rules to be applied to an application. The model is provided to devices of a first cohort, whereby a rule of the model causes an action associated with the application to occur at each device. User metrics associated with the action at each device of the first cohort is aggregated. The user metrics indicate a result of the action at each device. The aggregated user metrics are analyzed, whereby the analyzing includes determining a second cohort at the enterprise. The model is then provided to devices of at least a portion of the second cohort. No user metrics are returned to the centralized software service thus maintaining privacy compliance.