Gartner identifies six software engineering trends shaping enterprise AI development

Gartner has identified six strategic technology trends expected to influence software engineering in 2026 and beyond. The trends cover two areas: the use of AI across the software development lifecycle (SDLC), and the engineering practices required to build and operate AI-powered applications.

According to Gartner, these developments will influence how enterprises modernise legacy applications, accelerate IT transformation and manage application maintenance costs.

“The rapid advancement of AI is reshaping the software development life cycle (SDLC), forcing software engineering leaders to balance immediate productivity gains with longer-term investments in operating models, skills and governance,” said Manjunath Bhat, Distinguished VP Analyst at Gartner.

AI-native development platforms and harness engineering

One of the trends identified by Gartner is the emergence of AI-native development platforms. These platforms integrate AI into software development processes, ranging from tools that allow users to create applications through natural language prompts to AI coding agents that carry out development tasks under the supervision of professional developers.

Gartner predicts that by 2030, 40% of enterprise application portfolios will consist of custom applications built using AI-native development platforms, compared with 2% in 2025.

As AI-generated code becomes more common, organisations will also need mechanisms to ensure that the output follows internal standards and engineering practices. This is where harness engineering becomes important.

Harness engineering involves providing AI coding agents with the context, instructions, tools and guardrails needed to perform software engineering tasks reliably. These controls can include access to organisational knowledge, coding standards, development workflows and governance requirements.

Gartner predicts that organisations with AI coding harness infrastructure will experience 60% fewer post-deployment defects by 2028 than those without it.

The trend indicates that improving AI-generated code will require more than selecting a capable AI model. Enterprises will also need to establish the systems and processes that guide how coding agents operate.

Agent sandboxing addresses security risks

As organisations give AI coding agents greater autonomy, controlling their access to systems and resources becomes an important consideration.

Agent sandboxing involves running AI agents in isolated execution environments, limiting the extent to which they can interact with production systems, sensitive information or other critical resources.

This can help reduce risks associated with unintended agent actions, untrusted code execution and privilege escalation. For example, an AI coding agent testing a piece of software could be restricted to a controlled environment rather than being given unrestricted access to an organisation’s infrastructure.

Gartner predicts that 70% of enterprises will sandbox agentic development workflows in pre-production environments by 2030, up from 10% in 2026.

“Agent sandboxing provides a secure foundation for this shift, enabling teams to experiment while protecting systems from unintended agent-driven actions, untrusted code execution and privilege escalation,” Bhat said.

The growing use of AI agents in software development therefore requires organisations to consider security controls as part of the development environment, rather than treating them as a separate layer added after deployment.

Local AI and evaluation-driven development

Gartner also identifies local AI as an important deployment strategy for enterprises. This involves running AI models and agents within an organisation’s own infrastructure instead of relying primarily on APIs hosted by external providers.

Local deployment can help organisations address data sovereignty requirements, manage inference costs and support applications where low latency is important.

Gartner predicts that by 2028, 60% of regulated enterprises will migrate AI workloads to local infrastructure after discovering that a provider’s headquarters location, rather than the location of data storage, determines government access.

Another emerging practice is eval-driven development, which involves evaluating AI applications against predefined criteria throughout the development process.

Unlike conventional software testing, which typically checks whether an application behaves as expected for defined inputs, evaluations for generative AI and agentic systems must also consider the quality, accuracy, safety and reliability of their outputs.

Task-specific evaluations can help engineering teams identify weaknesses in AI applications and measure whether changes to models, prompts or workflows improve performance.

Gartner predicts that 70% of software engineering teams will adopt eval-driven development by 2028, compared with fewer than 20% in 2026.

Forward-deployed engineers connect AI with business requirements

The sixth trend identified by Gartner is the growing role of forward-deployed engineers (FDEs). These professionals work directly with business and technology teams to adapt and deploy AI solutions in real-world operating environments.

Unlike conventional software development roles that may work primarily from central engineering teams, FDEs combine technical expertise with an understanding of customer requirements and business processes.

Their role includes translating business problems into technical solutions, integrating AI applications with existing systems and helping organisations move from experimentation to production deployment.

Gartner predicts that more than 70% of AI technology and service providers will use FDEs by 2028, compared with 10% today.

The growing focus on these roles reflects the challenges involved in deploying AI across different enterprise environments, where applications often need to be adapted to existing data, workflows and infrastructure.

As AI becomes more deeply integrated into software development, enterprises will need to rethink not only the tools used by engineering teams but also the controls, skills and operating models supporting them. The six trends identified by Gartner point towards a software engineering environment where AI-assisted development is increasingly accompanied by stronger governance, testing and deployment practices.

Agentic AIAI AGENTSArtificial IntelligenceCybersecurityEnterprise AISoftware Engineering
Comments (0)
Add Comment