Automation 4 min read

AI-driven Low-code: Governance to Prevent Shadow IT

A guide for CIOs/CTOs on managing low-code and AI-driven shadow IT. Learn to implement governance, ensure compliance (NIS2/DORA), and secure your infrastructure.

The Evolution of Citizen Development in 2026

In 2026, IT departments in Ukrainian companies are navigating two powerful trends: the critical need for operational agility due to martial law and the explosive adoption of generative AI. As business users gain access to AI agents capable of generating code or building automation logic, they are bypassing traditional IT queues. While this phenomenon, known as Citizen Development, acts as a lifeline for process automation, it simultaneously creates a significant threat: Shadow IT—unmanaged infrastructure that fails to meet security, compliance, and architectural standards.

For CIOs and CTOs, the challenge is not to ban low-code tools, but to integrate them into a comprehensive IT governance strategy. Given European integration requirements like NIS2 and DORA, any business application created without proper oversight becomes a potential vulnerability. This article explores how to build a governance system that transforms chaotic employee creativity into a structured tool for digital transformation.

The Essence and Principles of Managed Citizen Development

The core principle of low-code governance is shifting from a "control through prohibition" model to a "control through platform" model. We must provide employees with a secure sandbox where they can experiment without compromising the company's security perimeter. Key principles include:

  • Identity Centralization: All applications must use corporate SSO and support integration with QES (Qualified Electronic Signature) and Diia.Signature (a secure digital signature service integrated with the Ukrainian state app) for legally binding actions.
  • Data Governance: Clear separation between data accessible to citizen developers (e.g., public directories) and critical data (personal data, financial reports) that requires strict access controls.
  • Monitoring and Visibility: The IT department must have visibility into all created applications, even those not deployed in production environments.

Architecture and AI Agent Integration

Modern low-code architecture is based on an "API-first" concept. When a business user builds an app, they do not have direct database access. Instead, they use a set of certified API gateways provided by IT. AI agents assisting in code generation must be connected to corporate LLM models that do not train on user data, ensuring confidentiality. An orchestration layer is essential, allowing the IT team to "adopt" user-created apps if they outgrow prototyping and require professional support, scaling, or integration into core business systems like ERP or CRM.

Criteria for Selecting Tools and Platforms

Selecting a low-code platform must be based on the tool's ability to uphold corporate security standards. The table below compares selection criteria for different task complexities.

CriterionBusiness Low-codePro-codeAI-agent Automation
Access LevelRestricted (API)Full (DB/Code)Contextual (LLM)
SecuritySSO/RBACDevSecOpsPolicy-based
ScalabilityLow/MediumHighDynamic
ComplianceBasicNIS2/DORAAudit logs

Implementation Practice: A Step-by-Step Algorithm

Implementing a governance system is a process, not a one-time project. TechCom, a Kyiv-based systems integrator in business since 2003, helps companies navigate this path through the following stages:

  1. Current State Audit: Identifying existing shadow solutions (Excel macros, unauthorized cloud services).
  2. Policy Definition: Creating a "whitelist" of approved platforms and tools.
  3. Center of Excellence (CoE) Formation: Establishing a team to train business users and consult on security.
  4. Monitoring Implementation: Setting up systems that automatically scan new applications for vulnerabilities.
  5. Integration with Corporate Standards: Connecting new apps to the unified identity management system and document signing via QES.

For instance, one industrial enterprise successfully centralized shop-floor data collection by allowing foremen to build data entry forms via an approved low-code platform, while a financial institution used a similar approach for internal approvals, maintaining full audit trails through integration with a corporate log server.

Common Pitfalls and Risks

The biggest mistake is ignoring the application lifecycle. When an amateur developer leaves or changes departments, the app becomes an "orphan." Without documentation and support, it turns into technical debt that can fail at a critical moment. Another risk is the excessive use of AI agents without verification. AI may generate logic that violates personal data processing rules, which is critical under European integration and eIDAS 2.0 requirements.

Economics: How to Evaluate the Impact

Evaluating governance effectiveness should not rely solely on development speed. Key metrics include:

  • Vulnerability Remediation Time: How quickly IT can identify and isolate an insecure application.
  • Support Costs: Comparing the cost of maintaining professional systems versus fixing errors in shadow applications.
  • Compliance Level: The percentage of apps that have passed security audits and adhere to internal policies.
  • Business Productivity: Time saved by employees through the automation of routine operations.

Conclusion

Low-code and AI are not threats to the IT department, but opportunities to transform from a "cost center" into an "innovation hub." By implementing the right governance mechanisms, you do not limit the business but provide it with secure tools for growth. In 2026, the winners will be those who can combine the speed of business initiatives with the reliability of corporate standards, ensuring operational continuity amidst any challenge.