Endava rewires software delivery with OpenAI’s AI agents

PUBLISHED: Thu, Jun 4, 2026, 9:16 AM UTC | UPDATED: Thu, Jun 4, 2026, 6:47 PM UTC

Endava, the London-based IT services provider, is overhauling its entire software delivery model around AI agents powered by OpenAI’s enterprise stack. The move marks one of the first large-scale implementations of autonomous AI agents in professional services, signaling how quickly the technology is jumping from experimental to mission-critical across the $600 billion global IT outsourcing industry.

Endava isn’t just experimenting with AI — it’s rebuilding its core business around it. The IT consultancy revealed it is using OpenAI’s ChatGPT Enterprise and Codex to deploy AI agents that handle everything from code generation to workflow automation, according to a case study published by OpenAI.

Scale and ambition

Rather than piloting AI tools in a single department, Endava is pushing for an “AI-native culture” across its entire organization. Thousands of developers, project managers, and consultants are learning to work alongside autonomous agents that can write code, debug software, and orchestrate complex delivery pipelines.

The approach centers on OpenAI’s Codex — the model behind GitHub Copilot — paired with ChatGPT Enterprise for knowledge work and workflow orchestration, creating a hybrid human-AI workforce.

Beyond code completion

Endava is automating entire workflows:

• Automated code reviews
• Intelligent task routing
• AI-generated documentation keeping pace with rapid development cycles

These are autonomous systems making decisions about how work flows through the organization, not just assistive tools.

Competitive context

Professional services firms have historically been slow to automate (billable-hours economics). Client pressure is changing that calculus. Accenture announced plans to retrain 25,000 employees on generative AI; Cognizant launched an AI practice targeting $2 billion annual revenue.

Endava’s transformation could help attract engineers who want to learn effective AI-agent collaboration — skills expected to define careers for the next decade.

Risks

Deploying autonomous agents at scale means trusting AI with consequential decisions about client codebases worth millions. Hallucinated functions or misunderstood requirements could cascade into serious problems. Endava is betting quality controls and human oversight can catch errors before they matter.

The experiment represents professional services’ potential “iPhone moment” — a technology shift demanding a new operating model from pricing to training, with AI agents as co-workers rather than tools.