AI is quickly becoming ubiquitous in software development and is changing the way developers build software. However, many companies have not invested in the key building blocks to sufficiently utilize this new technology. Software engineering leaders who fail to focus on the fundamentals of the AI era risk dooming their companies to irrelevance as faster AI-powered rivals seize innovation, revenue and market dominance.
According to a recent Gartner study, software engineering leaders who equip their teams with the right AI technologies can see a productivity improvement of more than 25%. They must build a new foundation that enables their teams to effectively co-create software with AI.
To achieve this, software engineering leaders must invest in five foundational practices to set up their teams for AI-native engineering success.
Exercise 1: Platform technique
Software engineering leaders must build platform engineering teams to deliver AI tools for software engineering, providing the platforms needed to enhance applications and software with AI capabilities
To achieve this, they must first build platforms that support AI software development tools on paved roads. Paved Roads enable the use of a range of common AI tools throughout the entire software development lifecycle (SDLC). This makes it easier for developers by not only removing the complexity of using the AI capabilities, but also building in guardrails to improve quality, cost, reliability, and security.
In addition, they must support the development of Model Operationalization (ModelOps) and Agent Engineering and Operations (AgentOps). A key part of the platform is to facilitate the full lifecycle of ML models, by offering the implementation, management and operation of large language models (LLMs). These should be compiled and maintained according to the enterprise’s security requirements and provided with several quick injections to tailor the results to the business context.
Finally, software engineering leaders must build platforms that enable AI capabilities. Adding AI capabilities to existing and new business software is necessary to remain viable. Leaders must also deliver internal developer platforms that help developers securely and seamlessly integrate AI capabilities such as chatbots and AI agents into their software. Providing templates, Application Programming Interfaces (APIs), guidance and training will ensure rapid innovation and risk-controlled rollout of AI capabilities.
Exercise 2: Integration and composability
As developers start to compose software instead of coding line by line, they will need API-compatible components and services to put together. Software engineering leaders must start by defining a goal to achieve a composable architecture based on modern composable applications, APIs, and loosely coupled API-first services.
They should also build an integration strategy and tool that implements well-defined API interfaces and creates rich metadata for APIs. Strong integration allows for easy assembly when components follow commonly agreed upon patterns. Gartner predicts that APIs will become an integral part of the functionality of AI agents, providing these agents with the necessary interfaces to consume, analyze, and act on data.
Exercise 3: AI-ready data
The future of software building depends on AI-ready data. Data is everywhere and it is very messy.
Software engineering leaders must support AI-ready data by organizing enterprise data assets for AI use. Generative AI is most useful when the LLM is combined with context-specific data. Platform engineering and internal developer portals provide the means by which this data can be packaged, discovered, and integrated by developers.
The urgent demand for AI-ready data to support AI requires evolutionary changes in data management and upgrades to architecture, platforms, skills and processes. Crucially, Model Context Protocol (MCP) must be considered. This emerging standard is designed to enable seamless integration between AI models, especially LLMs, and external data sources, APIs and tools.
Software engineering leaders must also build out both data mesh and data fabric. They must work with data management leaders to combine these two approaches into a modern data architecture. Fabric acts as the fundamental design pattern for data management and mesh for optimal data delivery using a federated model.
Exercise 4: Rapid Software Development Practices
With the accelerating advancements in AI technology, software engineering leaders must adopt newer, adaptive and iterative software development practices such as agile, DevSecOps and the product-centric model. To realize productivity gains from AI, leaders must focus teams on optimizing the entire SDLC with AI components.
To enable rapid development, software engineering leaders must reinvigorate agile and product-centric practices to respond to rapid code generation and provide reliable, fast routes to production. They must also accelerate the shift to a product-centric business model to strengthen product ownership and customer focus in engineering teams.
Additionally, software engineering leaders should challenge their teams to measure and improve idea cycle time. This is the time between ideation and production work code, and therefore to customer impressions and feedback.
Practice 5: Culture of innovation
Software engineers can become risk-averse unless they are given the freedom, psychological safety, and environment to take risks and experiment. Leaders must create a culture of innovation where their teams are eager to experiment with AI technologies. This also applies to software product ownership, where experimentation and innovation lead to greater optimization of the value delivered to customers.
To foster a cultural mindset that supports innovation, software engineering leaders must create a vision that inspires change and provides ownership of the changes AI requires.
They should also foster an environment of psychological safety, where challenges are viewed as opportunities for learning, and team members can express ideas, raise concerns, ask questions, and admit mistakes without fear of negative consequences.
To drive behavior change, software engineering leaders should build exploration teams to drive rapid innovation in key business areas using lean startup methodology and AI tools. They should also provide teams with dedicated innovation time and reward behaviors that drive innovation. Software engineers will only spend time exploring innovation if leadership emphasizes leadership as a core objective.
