# Agile Governance Approach with the Use of Artificial Intelligence

Agile methods play a crucial role in modern governance by reducing work effort, ensuring compliance, and fostering long-term maturity. Agile Governance aims to establish efficient and reliable practices that can adapt to organizational changes in real-time.

In an AI-powered governance solution, one of the key capabilities is the ability to discover and identify active components within the organization's ecosystem. It is essential for AI to analyze the internal content of these components, as it forms the foundation for subsequent analysis and classification using machine learning (ML) models (DAFOE, 2018).

In the realm of governance, several other critical capabilities, characteristics, and classifications have emerged. These must be considered during the governance and solution analysis stages (SANDEEP et al., 2020):

* Ethical Review: Regularly reviewing ML models to identify and mitigate bias or distortion in analysis results.

* Data Visibility: Providing consistent visibility into models and their results to data scientists.

* Functional Tests: Ensuring the reliability and consistency of the solution's execution.

* Decision Control and Action Promotion: Allowing the solution to autonomously perform actions based on the maturity of the ML model, without requiring human intervention.

* Pre-production Phase: Stabilizing and reviewing ML models before full-scale deployment to ensure accuracy and effectiveness.

* Production Monitoring: Continuously monitoring AI performance, identifying discrepancies between expected and actual results, and detecting biased decision-making.

* Integration: Enabling access to solution functionalities through analytical tools and other automation solutions.

These capabilities are crucial for maintaining an up-to-date catalog of enterprise ecosystem components. They ensure that Agile AI-assisted governance solutions align with the activities outlined in the Agile Governance process, whether through RPA or other ML-based predictive component classification mechanisms.

In addition to these capabilities, a strategic evaluation of component lifecycles is essential. The TIME technique (Tolerance, Investment, Migration, and Elimination) can be used to standardize decisions and evaluations related to governed components within the organization (GARTNER, 2014). This evaluation methodology, when structured and qualified using ML, classifies components across different dimensions (Figure 1).


Figure 1: TIME analysis for application catalog governance. Source: Gartner, 2014 (adapted)

AI plays a fundamental role in maintaining strategic development standards by aligning technology with business value and stabilizing component evaluation criteria. This evaluation should consider various aspects, including infrastructure cost, alignment with target system architecture, and value delivery to the business. Other criteria specific to component lifecycles in the organization's ecosystem should also be incorporated.

It's important to note that variations in patterns may occur within a company's ecosystem, which should be objectively evaluated. Proof of concept (POC) scenarios may exhibit lower accuracy in ML analysis results, with evaluation methodologies based on common development scenarios. However, AI's role extends beyond technical evaluations to encompass ethical, regulatory, and information security considerations related to components (SANDEEP et al., 2020).

The Agile Governance process, supported by artificial intelligence, becomes crucial in addressing questions such as cataloging services or systems, determining the correct service version, selecting the appropriate technological evolution approach, defining access levels and permissions, collecting relevant service metrics, identifying incident response stakeholders, and conducting incident review procedures. These questions provide a reference for discussing the boundaries of ML models within governance processes, and data scientists play a vital role in shaping these boundaries.

Data scientists actively participate in governance routines, defining criteria, metrics, and visualization characteristics for the organization's ecosystem. Their primary role is to provide ML models for production and assess whether real-world information significantly differs from controlled environment data. These Agile Governance tasks reinforce confidence in the model and enhance the effectiveness of AI application within the organization (DAFOE, 2018).

## Conclusion

In conclusion, AI serves as a powerful tool to support Agile Governance strategies. It not only ensures compliance with governance aspects but also promotes the healthy evolution of components. By automating processes and enabling learning, AI allows human resources to focus on building intelligent systems and improving governance itself (SANDEEP et al., 2020).