Andrew Zakonov, VP of Business at JetBrains
The developer community generally acknowledges the immense promise of AI in tackling a growing array of software development tasks, such as coding, testing, and refactoring. However, opinions diverge when it comes to the speed and scope of this transformation.
Forecasting the future is notoriously challenging, but discussions leading up to GITEX in Dubai reveal widespread excitement surrounding AI agents in development, albeit alongside certain reservations, particularly regarding specific implementation aspects.
Global research indicates that while 93 percent of organizations surveyed are planning to experiment with AI agents for software development, only 27 percent believe these agents will achieve full autonomy, a decline from 43 percent in 2024.
This trend is especially relevant in the UAE, where there is a concerted effort to cultivate a robust development ecosystem, highlighted by the National Program for Coders, which aims to train 100,000 programmers over the next five years. In such an evolving environment, the effective deployment of AI agents is crucial.
To prepare for successful collaboration between human developers and AI, teams must consider how swiftly and comprehensively they will embrace AI integration. It’s beneficial to explore two possible scenarios: the first, where AI acts as a supportive tool for developers, and the second, where AI takes a more central role throughout the development cycle.
Scenario 1: Collaboration Between Developers and AI Agents
In this scenario, AI support expands beyond mere code completion. Agents might become integral team members, tackling issues independently, updating documentation, and organizing codebases, thus enabling human developers to concentrate on more intricate and creative tasks. Additionally, AI roles may extend to areas like Quality Assurance, DevOps, and product management.
To effectively interface with coding agents, the developer toolkit must undergo transformation. Integrated Development Environments (IDEs) will need to evolve, providing user-friendly interfaces that allow developers to assign tasks, monitor progress, and evaluate AI-generated outputs. Trust-building tools will be necessary to empower developers to quickly assess AI actions. Besides coding, future AI advancements will automate processes such as debugging, performance profiling, and environment setup, which will significantly enhance human capabilities.
“Even with today’s LLM models, code can be generated in seconds through completion and prediction. In this first scenario, context switching will be minimized with a unified coding interface. Short prompt tasks will be managed entirely by the agents, resulting in immense time savings. However, this won’t fully address business challenges, as these typically require collaboration among multiple developers,” commented Andrew Zakonov, VP of Business at JetBrains.
As the volume of machine-generated code increases, robust control mechanisms will be essential. This includes implementing Git-level tagging to identify AI-generated code for tracking and developing specific auditing and testing tools to ensure the quality of machine-written code.
Scenario 2: Developers Leading AI Initiatives
The progression towards a fully AI-driven future hinges on establishing collaborative models where AI teams take the lead. This involves overcoming three principal challenges: creating seamless AI workspaces, developing next-generation collaboration tools, and establishing an AI-centric marketplace.
The primary obstacle is defining effective interaction points between human insight and AI efficiency, a domain known as Human-AI eXperience (HAX). New IDEs must be radically reimagined to guide and validate the large volume of AI-generated code effectively. These tools should automatically understand, verify, and summarize AI outputs, allowing developers to focus on overarching architecture and requirements.
For AI developers to function efficiently, they require a unified environment that supplies the necessary tools effortlessly. This necessitates moving beyond complex integrations to ensure AI agents can access all development resources (coding, testing, debugging, documentation) from a single interface. Additionally, the location for the agents’ operations—whether centralized cloud solutions or secure local environments—must be considered.
Moreover, increasing AI productivity will demand the creation of a new hub for code reuse. An AI-first marketplace should incorporate built-in sandboxing for safe testing and automated verification processes to assure quality and compatibility. This setup will enable AI teams to advance more rapidly, share resources securely, and continuously enhance codebases, thereby fostering a scalable ecosystem for autonomous development.
Regardless of the eventual trajectory, it is evident that AI will transition from a mere supportive tool to an active participant in coding, testing, and analysis. This evolution will redefine the roles of developers as facilitators and evaluators, working closely alongside AI agents.
Whether AI agents rapidly assume a leadership role in software development or continue assisting human developers for an extended period, their contribution will play a crucial part in the success of development teams. It is vital for organizations to assess the market trends and devise a strategy tailored to their specific needs.