Quality assurance · Production

AstraZeneca accelerates drug development with GitHub Enterprise, Actions, and Copilot—achieving 40% developer velocity increase

The problem

AstraZeneca's software organization was fragmented across Bitbucket and other satellite systems, making collaboration difficult, deployments inconsistent and error-prone, and onboarding slow for a diverse user base of around 3,000 people.

Workflow diagram · grounded in source

Platform consolidation decision

trigger

“moving approximately 3,000 users and 20,000 repositories”

Repository and permissions migration

integration

“they used GitHub Issues and Actions; and to manage the human side of migration, they built custom software, which automated the complex outreach, data collection, and scheduling for every team”

CI/CD automation library

integration

“the central DevOps team at AstraZeneca created a library of reusable GitHub Actions for common, critical tasks like security scanning and automated testing”

AI governance and ethics review

validation

“the tool also underwent a full AI, cybersecurity, and ethics review”

Copilot rollout with team training

ai_action

“over 1,600 developers and scientists onboarded, and active usage growing by 20% per month”

Agentic AI development tasks

ai_action

“by learning to use Copilot's agent mode in their daily work, pilot squads saw an immediate 40% increase in development velocity”

Human engineering review

human_review

“You still need skilled engineers to understand every line and every nuance of why the system behaves the way it does”

Reported outcome

AstraZeneca consolidated nearly 99% of developers and source code on GitHub, grew CI/CD automation events by 100%, and through GitHub Copilot achieved a 40% developer velocity increase with engineers reporting 9-10 hours of extra output per week.

Reported metrics

developer velocity increase (Copilot pilot)40%

automated CI/CD events growth100%

developers and source code consolidated on GitHubNearly 99%

Copilot users onboarded in 6 monthsover 1,600

Show all 11 reported metrics

developer velocity increase (Copilot pilot)40%

automated CI/CD events growth100%

developers and source code consolidated on GitHubNearly 99%

Copilot users onboarded in 6 monthsover 1,600

Copilot active usage growth per month20%

extra developer output per week (current)9-10 hours

extra developer output per week (previous quarter)four hours

project completion time with agent vs. withouttwo hours

Copilot onboarding timeline vs. goalgoal shattered in just six months

remediation cost reduction from shift-left testingexponentially reduced remediation costs—often tens or hundreds of times more cost-effective

initial developer user basearound 3,000

Reported stack

GitHub EnterpriseGitHub ActionsGitHub CopilotGitHub Issues

Source

https://github.com/customer-stories/astrazeneca

Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

AstraZeneca consolidated nearly 99% of developers and source code on GitHub, grew CI/CD automation events by 100%, and through GitHub Copilot achieved a 40% developer velocity increase with engineers reporting 9-10 ho…

What tools did this team use?

GitHub Enterprise, GitHub Actions, GitHub Copilot, GitHub Issues.

What results were reported?

developer velocity increase (Copilot pilot): 40%; automated CI/CD events growth: 100%; developers and source code consolidated on GitHub: Nearly 99%; Copilot users onboarded in 6 months: over 1,600 (source-reported, not independently verified).

How is this quality assurance AI workflow structured?

Platform consolidation decision → Repository and permissions migration → CI/CD automation library → AI governance and ethics review → Copilot rollout with team training → Agentic AI development tasks → Human engineering review.