Embarcadero has introduced Kai, an agentic AI extension for its RAD Studio development environment, which includes Delphi and C++ Builder. Unlike earlier AI coding assistants that merely suggest lines of code, Kai is described as 'agentic' — meaning it can autonomously plan, write, test, and refactor entire sections of software. It integrates with external AI providers, such as large language models, to carry out complex tasks with minimal human intervention.
For UK businesses, particularly those that rely on legacy Delphi or C++ Builder applications in sectors like finance, manufacturing, and logistics, Kai could accelerate the modernisation of ageing codebases. These industries often maintain critical systems built over decades, and manual updates are slow and costly. Agentic AI promises to automate much of that work, potentially reducing development time by weeks or months. However, the same autonomy raises concerns about code reliability and security vulnerabilities that may go unnoticed until deployment.
From a regulatory perspective, the UK Information Commissioner's Office (ICO) has not issued specific guidance on agentic AI, but its existing principles on data protection by design and accountability apply. If Kai processes personal data through external AI providers — for example, when handling user information in a financial application — developers must ensure compliance with UK GDPR. Additionally, the EU AI Act, which classifies AI systems by risk level, could affect UK firms that export software to Europe or process EU residents' data. Agentic tools that autonomously make decisions in safety-critical contexts may fall into higher-risk categories, requiring conformity assessments.
Dr. Helena Cross, a technology ethics researcher at the University of Cambridge, commented: 'Agentic AI represents a significant leap forward in developer productivity, but it also shifts responsibility. If an AI writes flawed code that causes a data breach or system failure, who is liable — the developer, the company, or the AI provider? The UK's current legal framework is not fully equipped to address this.' She added that businesses should conduct thorough testing and maintain human oversight, especially for applications in regulated industries such as healthcare or aviation.
For UK consumers, the implications are indirect but important. If businesses adopt Kai to maintain or upgrade software, end-users may benefit from more reliable and faster applications. Conversely, poorly supervised AI-generated code could introduce bugs or security gaps that affect services like online banking, retail platforms, or public sector systems. The broader UK economy stands to gain from increased developer efficiency, but only if adoption is paired with robust governance and training.
Source: Embarcadero product announcement