Summary: Endava’s CTO describes its AI-native shift as a redesign of workflows and behavior rather than a tool rollout. OpenAI technology runs through the DavaFlow lifecycle from requirements and planning through engineering and deployment, and use has spread to legal, project management, and commercial teams. The company says the bottleneck moved from engineering output to requirements gathering and stakeholder coordination.
Inside Endava’s Shift to AI-Native Operations
With more than 25 years in technology services, Endava has put AI at the heart of its mission. Yet bringing AI in was never merely a matter of new tools—it meant rethinking how work flows, how leaders behave, and how teams collaborate. Speaking with OpenAI, Endava CTO Matthew Cloke described how the company wove AI through the whole organization and rebuilt software delivery around agents.
Per Cloke: “AI has had a fundamental impact on Endava over the past couple of years.” The question to confront: “We really had to answer the question of how to be a relevant organization in the new AI world.”
That conviction pushed Endava to adopt OpenAI as its enterprise AI platform, with ChatGPT Enterprise and Codex opened up to staff. Adoption alone wasn’t the point; the point was folding AI into the normal rhythm of daily work.
His definition: “To be AI-native at Endava, it’s about thinking about AI to solve the problem first,” where AI is “the first thing you do rather than the last thing that you do.”
Inside the Delivery Lifecycle: Where AI Sits
The shift started with the software delivery teams themselves. Once developers began experimenting with AI-assisted coding and agentic workflows, it became obvious that engineering output was no longer what held work back. Requirements gathering, business analysis, planning, and stakeholder coordination now had to speed up as well.
“We started to challenge how quickly we could produce requirements and how quickly we could produce the right business solutions for our clients,” Cloke recalls.
OpenAI technology now runs through every stage of the DavaFlow lifecycle, covering meeting preparation, business planning, product discovery, software engineering, and deployment. In Cloke’s words: “There isn’t a part of DavaFlow that doesn’t use OpenAI technology.”
Nor did usage stay confined to engineering. In legal, AI took over the grinding parts of research and documentation. Project managers turned to Codex for governance reports and engineering progress summaries. Commercial teams swapped spreadsheet-heavy planning drills for lightweight applications generated with AI.
During one internal pricing discussion, staff abandoned spreadsheets altogether and put together a single-page pricing app the whole team could interact with immediately. “It changed the conversation completely,” Cloke says.
Agents have likewise settled into routine operations: leadership teams rely on them to summarize projects, automate communications, keep inboxes under control, and coordinate work asynchronously.
Where the Gains Show Up
Endava’s experience translates into concrete wins for product builders:
- Faster software delivery: agents folded into engineering workflows shortened delivery times.
- Adoption across functions: use spread from engineering into legal, finance, and operations teams.
- Less manual toil: AI-assisted workflows trimmed hand-done reporting and coordination.
- Non-engineers empowered: teams created internal tools with no dedicated engineering help.
- AI fluency in hiring: fluency now sits within hiring and promotion expectations.
The lesson for product builders: look for the bottlenecks sitting outside engineering. If requirements gathering or stakeholder coordination is your constraint, think about where agents could lend a hand—drafting requirements from meeting notes, producing status reports, or standing up lightweight internal tools that do away with spreadsheets.
What It Costs and Where It Falls Short
Endava’s results are strong, but the trade-offs are real: the transformation demanded a deliberate cultural shift rather than a tooling exercise. As Cloke stresses, adopting AI is behavior change, not a software rollout—leaders have to use AI themselves for adoption to spread organization-wide, and room to experiment (imperfect outcomes included) has to exist.
AI is no silver bullet either. The bottleneck just relocated from engineering to other functions, so workflow redesign never stops. Non-technical teams have to join early rather than late if alignment is to hold, and while hands-on experience dissolves skepticism faster than anything else, it takes time and investment.
Lessons Product Builders Can Act On
Endava’s journey leaves usable lessons:
- Approach AI adoption as behavior change, not a software rollout; the real work is changing how people think about solving problems.
- Have leaders use AI personally, modeling the behavior that drives adoption.
- Leave room to experiment, imperfect outcomes and all.
- Pull non-technical teams in early so they help shape workflow design.
- Build AI into everyday workflows instead of running it as a separate initiative.
For organizations still near the start, his advice: use the technology personally. “The future arrived,” he says. “You just have to lean into it.”
The Road Ahead: Orchestration as the Operating Model
What comes next, in Endava’s view, revolves around orchestration: models, agents, workflows, and human expertise joined into integrated systems that change how organizations run. “We’re really excited about the workflows that can be created by combining these tools,” Cloke says.
Reasoning models, Codex agents, automation, enterprise-scale collaboration—taken together, AI is turning into something bigger than a productivity layer: the operating model itself. That asks product builders to design products able to orchestrate AI agents and human workflows seamlessly.
The question worth asking is no longer “should we use AI?” but “how do we redesign workflows so AI is the first thought?” Endava’s case suggests the answer: put AI into every stretch of the delivery lifecycle, requirements through deployment, and treat it as a core strand of company culture.