Endava is rebuilding its software delivery operation around AI agents, deploying OpenAI's Codex and ChatGPT Enterprise across engineering teams to automate workflows and shift the firm toward what it calls an AI-native culture. The London-listed IT services company disclosed the rollout in a joint case study with OpenAI, framing the work as a structural redesign of how code gets shipped rather than a tooling upgrade.
The stakes for a firm like Endava are not subtle. IT services has been built for two decades on billable engineering hours, with revenue scaling roughly linearly with headcount. Agents that write, review, and ship code threaten that math directly — which is why incumbents are racing to redesign delivery before clients start asking why they are paying for human keystrokes.
Endava is using Codex, OpenAI's coding agent, as the execution layer inside its delivery pipeline. ChatGPT Enterprise sits across the broader workforce as the knowledge and reasoning layer, giving non-engineering functions access to the same models under enterprise data controls. The company describes the combination as the foundation for an agent-augmented delivery model.
Key facts
- 01Endava is deploying OpenAI's Codex and ChatGPT Enterprise across its software delivery organization.
- 02The rollout targets workflow automation and what Endava calls an AI-native delivery culture.
- 03The move signals a shift in IT services from billable hours toward agent-augmented output.
The specific workflows Endava is automating include code generation, documentation, ticket triage, and test scaffolding — the unglamorous middle of a software project where services firms historically bill the most hours. Moving those tasks to agents compresses delivery timelines and, in theory, lets the same engineering pod take on more concurrent work.
The strategic question is what happens to pricing. If a fixed-price engagement that previously consumed 10,000 engineering hours now consumes 4,000 hours plus agent compute, the services firm either keeps the margin, passes the savings to the client, or — more likely — does a mix while repositioning the engagement around outcomes rather than hours. Endava has not publicly disclosed how it is repricing AI-accelerated work.
Endava is one of a handful of mid-tier services firms — alongside larger players like Accenture, Capgemini, Infosys, and TCS — moving publicly on agent-based delivery. The larger consultancies have announced multi-billion-dollar AI practice investments over the past 18 months, but the integration question is the same at every scale: how do you redesign a partner-and-pyramid org chart when the pyramid's base is shrinking.
OpenAI's interest in publishing the case study is its own signal. Codex has been positioned as a coding tool aimed at individual developers, but the enterprise-services channel is where OpenAI captures durable seat-based revenue and gets distribution into Fortune 500 engineering organizations it cannot reach directly. Each services-firm reference customer compounds into pitches at that firm's downstream clients.
The skeptic's read on agent-accelerated services is that productivity gains in practice trail the demo. Internal benchmarks at large enterprises have consistently shown that coding agents help most on greenfield work and well-scoped tasks, and help less on the legacy-system integration work that makes up the bulk of services revenue. Endava has not published throughput or defect-rate data from its rollout, which would be the numbers needed to validate the shift.
There is also the question of what AI-native culture means inside a 12,000-person organization built on traditional delivery methodologies. Reorganizing teams around agents requires retraining, new review processes, and rewriting the contracts that govern how work is delivered to clients. None of that happens in a quarter, and most services firms are still in pilot mode on the contract side even as engineering teams adopt the tools.
The Endava-OpenAI partnership matters less as a single deal and more as a template for how services firms intend to convert AI from a margin threat into a margin opportunity. The firms that move first — restructuring delivery, repricing engagements, and locking in agent platforms — get to define the next decade of how enterprises buy software work. The ones that wait will find their client conversations dominated by procurement teams asking why human hours still appear on the invoice.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
