MidPilot

The main goal of the midPilot is to accelerate application onboarding for midPoint and thus improve the overall visibility into the environment. By enabling rapid integration capabilities we will achieve reducing shadow IT, lowering operational costs and the overal security will be improved.

As part of the project, we aim to address the key technical challenges involved in connecting systems to an identity governance and administration (IGA) platform, midPoint. This includes the development of integration code (connectors) for different operations, such as authentication, authorization, object discovery, and identity-related operations. Given the diversity of integration protocols, evolving system interfaces, and the need to verify behavior in live environments, building these connectors can be highly time-consuming. Moreover, modern organizations typically use hundreds of systems making manual development impossible to scale. Our goal is to leverage generative AI (GenAI) for connector development and maintenance and thus reduce the overall effort and time related to onboarding new applications.

In parallel, we plan to implement AI-powered tools to streamline data model mapping between systems. Attribute names often differ due to language, format, or abbreviation inconsistencies, making manual mapping slow and error-prone. The proposed AI-powered recommendation system will identify likely attribute matches. In complex or non-obvious cases, the processes will be further enhanced with analyzing sample mappings, live data, or applying script-based transformations. The process of using mapping recommender will be interactive and iterative, proposing suitable solutions that will always need a confirmation from a user. In cases where no direct mapping is possible, for example because a native midPoint attribute is missing, the AI will propose schema extensions to support organization-specific attributes. By combining GenAI with additional AI/ML techniques, heuristics and midPoint simulations mechanism, we aim to reduce manual effort and increase the speed and accuracy of identity data integration.

Finally, we will focus on automating correlation process. This involves object types correlation followed by identity correlations, aligning records such as user accounts, roles, and memberships between the source system and midPoint. Our approach will use AI to suggest delineation and correlation rules based on data analysis, common approaches, statistical and heuristic methods. For straightforward cases, automatic correlation of the majority of records will be recommended, with manual review reserved for outliers. In more complex scenarios, the system will support deeper analysis and allow step-by-step confirmation of correlation proposals, including simulations of the post-correlation state to support decision-making. If no initial rules are identifiable, the process can start manually, enabling the AI system to learn from user-confirmed matches.

Throughout, the system will be designed to handle real-world data quality issues, such as typos, missing values, and inconsistencies, ensuring reliable correlation and mappings even in less-than-ideal conditions.

Through this, we aim to significantly accelerate application onboarding into midPoint, reduce reliance on manual effort, and improve overall governance and security posture.

Note: A detailed timeline of midPoint releases can be found in the roadmap.