Endava如何围绕AI代理重新设计软件交付

Endava是一家拥有超过25年历史的全球技术服务公司，致力于通过技术帮助企业解决复杂的业务问题。如今，这一使命越来越聚焦于人工智能。

但对Endava而言，采用AI不仅仅意味着引入新工具。它需要重新思考工作流程、领导行为以及团队在业务中的协作方式。

我们与首席技术官Matthew Cloke进行了对话，了解Endava如何将AI嵌入整个组织，围绕智能体重新设计软件交付，并创造一种实验被视为理所当然而非可选的氛围。

“过去几年，AI对Endava产生了根本性影响，”Cloke表示。“我们确实必须回答一个问题：在新的AI世界中，如何成为一个有相关性的组织。”

这种思维促使Endava选择OpenAI作为其企业AI平台，让公司全体员工都能使用ChatGPT Enterprise和Codex。目标不仅仅是采用——而是让AI成为日常工作流程的一部分。

“在Endava实现AI原生，意味着首先考虑用AI解决问题，”Cloke解释道。“这是你做的第一件事，而不是最后一件事。”

“如果我没有一个智能体在后台运行，我总觉得是在浪费时间。”

——Matthew Cloke，Endava首席技术官

推广过程

Endava的AI转型始于其软件交付团队。

随着开发者开始尝试AI辅助编码和智能体工作流程，团队很快意识到瓶颈不再是工程产出。需求收集、业务分析、规划和利益相关者协调也需要加速。

“我们开始挑战自己：能以多快速度生成需求，能以多快速度为客户提供正确的业务解决方案，”Cloke说。

如今，OpenAI技术已嵌入整个DavaFlow生命周期——从会议准备和业务规划，到产品发现、软件工程和部署。

“DavaFlow的每个环节都在使用OpenAI技术。”

——Matthew Cloke，Endava首席技术官

重要的是，采用并未止步于开发者。

法务团队开始使用AI来简化研究和文档工作流程。项目经理开始使用Codex生成治理报告并总结工程进展。商业团队用轻量级AI生成应用取代了繁重的电子表格规划工作。

在一次内部定价讨论中，员工完全跳过了电子表格，转而构建了一个团队可以立即交互的单页定价应用。

“这彻底改变了对话方式，”Cloke说。

AI智能体也已融入日常运营。领导团队使用智能体来总结项目、自动化沟通、管理收件箱以及异步协调工作。

成果概览

• 通过将AI智能体集成到工程工作流程中，加速了软件交付
• 将AI采用范围从工程部门扩展到法务、财务和运营团队
• 通过AI辅助工作流程减少了手动报告和协调工作
• 使团队能够在没有专门工程支持的情况下构建内部工具和应用
• 将AI熟练度作为公司招聘和晋升期望的一部分

Endava的经验教训

随着Endava在其11,000名全球员工中推广AI，总结出以下原则：

• 将AI采用视为行为改变，而非软件部署
• 领导者需要积极使用AI来推动全组织的采用
• 为实验创造空间——即使结果不完美
• 尽早让非技术团队参与进来，而非推迟
• 亲身体验是克服怀疑的最快方式
• 让AI成为日常工作流程的一部分，而非独立项目

未来展望

作为OpenAI的长期合作伙伴，Endava认为企业AI的下一阶段将围绕编排展开——将模型、智能体、工作流程和人类专业知识整合为集成系统，从根本上重塑组织的运作方式。

“我们对通过组合这些工具所能创建的工作流程感到非常兴奋，”Cloke说。

从推理模型和Codex智能体，到自动化和企业级协作，Endava相信AI正在超越生产力层的角色。它正在成为运营模式本身。

而对于仍处于转型初期的组织，Cloke的建议直截了当：亲自开始使用这项技术。

“未来已经到来，”他说。“你只需拥抱它。”

Endava is a global technology services company that has spent more than 25 years helping enterprises solve complex business problems through technology. Today, that mission increasingly centers on AI.

But for Endava, adopting AI meant more than introducing new tools. It required rethinking workflows, leadership behaviors, and how teams collaborate across the business.

We sat down with Matthew Cloke, CTO, to hear how Endava is embedding AI across the organization, redesigning software delivery around agents, and creating a culture where experimentation is expected—not optional.

“AI has had a fundamental impact on Endava over the past couple of years,” says Cloke. “We really had to answer the question of how to be a relevant organization in the new AI world.”

That mindset led Endava to make OpenAI its enterprise AI platform, giving employees across the company access to ChatGPT Enterprise and Codex. The goal wasn’t simply adoption—it was making AI part of the flow of everyday work.

“To be AI-native at Endava, it’s about thinking about AI to solve the problem first,” Cloke explains. “It’s the first thing you do rather than the last thing that you do.”

“If I don’t have an agent running in the background, I somehow think I’m wasting my time.”

—Matthew Cloke, CTO, Endava

Inside the rollout

Endava’s AI transformation began inside its software delivery teams.

As developers started experimenting with AI-assisted coding and agentic workflows, teams quickly realized the bottleneck was no longer engineering output. Requirements gathering, business analysis, planning, and stakeholder coordination all needed to move faster too.

“We started to challenge how quickly we could produce requirements and how quickly we could produce the right business solutions for our clients,” says Cloke.

Today, OpenAI technology is embedded throughout the entire DavaFlow lifecycle—from meeting preparation and business planning to product discovery, software engineering, and deployment.

“There isn’t a part of DavaFlow that doesn’t use OpenAI technology.”

—Matthew Cloke, CTO, Endava

Importantly, adoption didn’t stop with developers.

Legal teams began using AI to streamline research and documentation workflows. Project managers started using Codex to generate governance reports and summarize engineering progress. Commercial teams replaced spreadsheet-heavy planning exercises with lightweight AI-generated applications.

In one internal pricing discussion, employees skipped spreadsheets entirely and instead built a single-page pricing app teams could interact with immediately.

“It changed the conversation completely,” Cloke says.

AI agents have also become embedded in day-to-day operations. Leadership teams use agents to summarize projects, automate communications, manage inboxes, and coordinate work asynchronously.

Results at a glance

• Accelerated software delivery by integrating AI agents into engineering workflows
• Expanded AI adoption beyond engineering into legal, finance, and operations teams
• Reduced manual reporting and coordination work through AI-assisted workflows
• Enabled teams to build internal tools and applications without dedicated engineering support
• Established AI fluency as part of hiring and promotion expectations across the company

Lessons learned from Endava

As Endava rolled out AI across its 11,000-person global workforce, several principles emerged:

• Treat AI adoption as a behavior change, not a software rollout
• Leaders need to actively use AI to drive organization-wide adoption
• Create space for experimentation—even when outcomes are imperfect
• Bring non-technical teams into the process early, not later
• Hands-on experience is the fastest way to overcome skepticism
• Make AI part of everyday workflows, not a separate initiative

What’s next

As a long-term OpenAI partner, Endava sees the next phase of enterprise AI centered around orchestration—combining models, agents, workflows, and human expertise into integrated systems that fundamentally reshape how organizations operate.

“We’re really excited about the workflows that can be created by combining these tools,” says Cloke.

From reasoning models and Codex agents to automation and enterprise-scale collaboration, Endava believes AI is becoming more than a productivity layer. It’s becoming the operating model itself.

And for organizations still early in the journey, Cloke’s advice is straightforward: start using the technology personally.

“The future arrived,” he says. “You just have to lean into it.”