How Endava Is Redesigning Software Delivery Around AI Agents

Endava is restructuring its entire software delivery lifecycle to center on AI agents. For a technology services company that has spent more than 25 years solving complex business problems, the shift is not about adding new tools, but about fundamentally rethinking workflows and leadership as detailed in this report from OpenAI.

The company has adopted OpenAI as its enterprise AI platform, providing employees with access to ChatGPT Enterprise and Codex. This move aims to integrate AI into the flow of everyday work, moving it from a final step in a process to the primary way problems are approached. As CTO Matthew Cloke notes, the goal is to make AI the first consideration in problem-solving.

The transformation is currently focused on the DavaFlow lifecycle. While the initial experimentation occurred within software delivery teams through AI-assisted coding and agentic workflows, the company found that engineering output was not the only bottleneck. To maintain momentum, Endava is applying OpenAI technology to requirements gathering, business analysis, planning, and stakeholder coordination. Cloke states that there is no part of the Dava-Flow lifecycle that does not use OpenAI technology.

This shift extends beyond engineering. The company is embedding these systems into diverse business functions:

• Legal: Using AI to streamline research and documentation.
• Project Management: Utilizing Codex to generate governance reports and summarize engineering progress.
• Commercial: Replacing spreadsheet-heavy planning with AI-generated applications.

The implication for the broader services industry is clear: the value of a technology partner is shifting from the ability to manage manual workflows to the ability to orchestrate agentic systems. When the bottleneck moves from code production to requirements and planning, the competitive advantage lies in how effectively a firm can automate the entire lifecycle, not just the development phase.

If you are managing a technical organization, consider where your current bottlenecks reside. If your engineering output is high but your planning and requirements phases are lagging, your infrastructure may be ready for agents, even if your workflows are not.