Endava's AI Engine

Global software firm Endava is redefining its operational model by adopting an "agentic organization" structure, powered by OpenAI's Codex. This strategic shift embeds senior engineering knowledge into AI agents, scaling expertise across its entire client engagement process, according to OpenAI News.

Joe Dunleavy, Endava’s regional CTO for Europe, stated that the firm has moved from primarily producing code to overseeing AI-generated output, resulting in a significant increase in quality. "The quality of output has just gone up exponentially," Dunleavy noted.

Endava now conceptualizes itself as an entity where senior-level judgment is codified into AI agents. These agents work alongside human teams from project intake and ideation through to final delivery.

"What Codex has really helped us do is have small teams of people deliver massive value in a very condensed timeframe," Dunleavy added.

Magnifying Expertise

Mike Krolnik, Endava’s Global SVP of Agentic Architecture, explained how Codex bridges the gap between senior and junior engineers. Senior architects can articulate complex requirements, which Codex then translates into accessible information for junior team members. This allows less experienced developers to produce outputs typically requiring senior oversight.

Codex functions as a real-time guide, reinforcing best practices and architectural decisions. Junior developers can query the AI for clarification, effectively receiving mentorship encoded within the tool.

"I can take my experience, codify it, and have Codex work with the team to teach them better practices in software architecture and development," Krolnik said.

This real-time knowledge transfer represents a core advantage, transforming traditional mentorship methods like pairing and code reviews into an always-on resource.

Krolnik elaborated on the tool's versatility: "Codex has matured as a tool. We use it for requirements analysis, design, specifications, development, and operations; it’s a general desktop agent across our whole lifecycle." Endava's perspective on using Codex as a broad operational tool is detailed in "Endava CTO on Codex: AI as a General Desktop Agent".

Compressing Timelines

The constant availability of senior-level guidance through Codex accelerates every phase of software delivery. Endava no longer treats analysis, design, and build as strictly sequential stages.

"Each of these stages used to take days or weeks of analysis," Krolnik recalled. "Now with Codex packaging together analysis, design, and build, we can do that as a single unified tool."

A recent internal project illustrated this efficiency: Endava's legal team needed thousands of contract pages reviewed against specific criteria. What would normally involve weeks of inter-departmental back-and-forth was condensed. A two-hour meeting transcript with legal stakeholders was fed to Codex, generating a usable requirements specification after just two one-hour review sessions.

Client-facing work has also seen similar acceleration. Endava teams now generate design documents and diagrams live during client sessions, using Codex to visually represent proposed architectures and rapidly iterate on ideas.

"It rapidly accelerates the back-and-forth, and it really opens a lot of doors," Krolnik stated.

Leadership Lessons from Endava

• Codify senior expertise: Capture the judgment of senior architects in AI tools to provide scalable guidance to junior team members.
• Treat AI as a desktop agent: Apply tools like Codex across the entire workflow, including requirements, design, client communication, and operations, not just code generation.
• Embrace experimentation: Actively try new AI tools, particularly starting with non-coding workflows, to quickly realize their full potential.