Endava and AI agents: how software delivery is changing

Endava: Timeline of a Quiet Transformation

Endava isn't a name that comes up often in conversations among Italian SMEs. Yet it is one of the most significant digital engineering companies worldwide. Headquartered in London, with operations spread across Europe, Latin America and North America, the company manages complex projects for clients in finance, healthcare and retail.

In 2025, Endava launched a structured program to adopt AI agents. The stated goal was precise: cut software time-to-delivery without sacrificing quality. The program rolled out in progressive phases, starting with high-volume use cases that required little complex decision-making.

So the starting point wasn't the most advanced technology available. It was identifying the real bottlenecks in the development process. This methodological choice is, perhaps, the most useful lesson for anyone watching the case from the outside.

The Core Tools: ChatGPT Enterprise and Codex

The case study published by OpenAI describes the technology architecture Endava adopted in detail. At its core are two tools: ChatGPT Enterprise and Codex.

ChatGPT Enterprise was integrated into developers' daily workflows. In particular, it was used to generate technical documentation, write automated tests and review existing code. Codex, on the other hand, was used to speed up code production in specific languages, cutting the time spent on repetitive tasks.

Beyond that, Endava built specialized agents for vertical tasks. For example, one agent is dedicated to migrating legacy codebases, another to automatically generating technical specifications from business requirements. In other words, this isn't generic use of AI: it's targeted orchestration of specific capabilities.

According to the estimates reported in the case study, some stages of the development cycle saw time reductions of around 30-40%. These figures are consistent with what McKinsey has observed in its research on developer productivity with generative AI.

Who Gained and Who Had to Adapt

In every transformation like this, some people come out ahead and others have to redefine their role. In the Endava case, the distinction is clear-cut.

Senior developers benefited the most from automation. Freed from low-value tasks, they could focus on architecture, critical review and design decisions. The quality of their output rose in measurable ways.

Junior profiles, by contrast, went through a more complicated transition. The work traditionally given to early-career developers — writing boilerplate code, basic documentation, manual testing — has become the domain of AI agents. As a result, Endava had to redesign onboarding and training paths for this group of professionals.

Finally, project managers saw the nature of their work change radically. Managing dependencies between tasks has become partially automated. As a result, their value has shifted toward governing AI processes and handling exceptions.

The Cultural Side: What the Numbers Don't Capture

At SHM Studio, we believe the most original contribution of the Endava case isn't technological. It's organizational.

The company invested heavily in building an AI-native culture. That meant widespread training programs, internal communities of practice and — an aspect often overlooked — a governance system that defines when agents can act autonomously and when human oversight is required.

As in other enterprise contexts, the main risk wasn't resistance to change. It was superficial adoption: using AI tools without changing the underlying processes. Endava chose the longer but sturdier path. The result: outcomes that last.

This approach is backed by Harvard Business Review research on building an AI-ready culture, which identifies governance and training as the critical success factors for enterprise adoption.

SHM Studio's take: what it means for Italian SMEs

The Endava case is enterprise by definition. Still, the dynamics it describes matter for smaller businesses too. Italian SMEs working in B2B or digital retail face similar choices today, just on a different scale.

First, the logic of AI agents doesn't require big-corporation infrastructure. Tools like the AI solutions already available on the market can automate specific workflows — from content management to lead qualification — with affordable investments.

Second, the lesson about culture transfers directly. A 50-person company that adopts AI tools without redefining roles and responsibilities will see only marginal results. A company that integrates AI into its existing processes with a clear method and proper training, on the other hand, can gain a significant competitive edge.

So for an SME, the starting point shouldn't be choosing a tool. It should be mapping the processes that are high-volume and low in decision complexity — exactly what Endava did.

Real-World Applications: From Development Workflows to Marketing Processes

The AI agent logic Endava applied to software development has a direct counterpart in marketing and digital communication. Here too, there are high-volume, structured, repetitive tasks that lend themselves to intelligent automation.

For example, producing SEO-driven content, managing digital campaigns and running periodic reporting are areas where AI agents can work with reduced human supervision. Our SEO and digital marketing services already build these approaches into operational workflows.

Content creation for LinkedIn campaigns and Google Ads also benefits directly from assisted automation. The point isn't to replace human strategic judgment, but to speed up the operational steps that follow the decisions.

Likewise, professional copywriting is evolving toward a hybrid model: the professional sets strategy, tone and goals; the AI agent produces drafts and variations; the professional validates and refines. This model is already in place at many international agencies.

Next Moves: What We Expect in the Next 18 Months

The Endava case points to a direction that will go mainstream by 2027-2028. Some trends are already clear enough to spot.

First of all, agents will become more specialized. Today, AI agents are relatively generalist. Next, we'll see vertical agents for specific sectors — legal tech, healthcare, manufacturing — with reasoning capabilities tailored to their domains.

Also, agent governance will become a critical skill. Companies that invest today in defining policies and oversight processes will have a structural advantage as system complexity grows.

Finally, AI agents will become a standard part of web development platforms and CMSs. Companies that still manage their digital presence by hand will be at a competitive disadvantage against those that have automated their recurring operational processes.

To dig deeper into these topics or explore how to bring AI into your organization's processes, you can contact the SHM Studio team or browse the related articles on our blog.