Skip to content

Agile Development

In this documentation we refer to the cross-functional delivery team as a "Crew" (i.e., the team responsible for delivery on a project). This includes the dev team, dev lead, product manager (PM), data scientists, etc.

Why Agile

  • We want to be quick to respond to change
  • We want to get to a state of working software fast, and iterate on it to improve it
  • We want to keep the customer/end users involved all the way through
  • We care about individuals and interactions over documents and processes

The Fundamentals

We care about the goal for each activity, but not necessarily about how they are accomplished. The suggestions in parenthesis are common ways to accomplish the goals.

  • We keep a shared backlog of work that everyone on the team can access (e.g., Azure DevOps, GitHub, Jira or other backlog tools)
  • We plan our work in iterations with clear goals (ex. sprints)
  • We have a clear idea of when work items are ready to implement (ex. definition of ready)
  • We have a clear idea of when work items are completed (ex. definition of done)
  • We communicate the progress in one place that everyone can access, and keep the progress up to date (ex. sprint board and daily standups)
  • We reflect on our work regularly to make improvements (ex. retrospectives)
  • The team has a clear idea of the roles and responsibilities in the project (e.g., dev lead, Technical Program Manager (TPM), Process Lead, etc.)
  • The team has a clear idea of how we work together (ex. team agreement)
  • We value and respect the opinions and work of all team members.

AI tooling considerations

Teams are increasingly adapting traditional Scrum practices to take advantage of AI tools and workflows. The tools emphasize rapid iteration, short discovery/validation cycles, and AI-assisted asset generation. Consider these principles as you tailor your process:

  • Favor frequent, lightweight planning and validation over rigid, fixed-length cadences when it helps accelerate feedback loops and reduce cycle time.
  • Use AI-assisted tooling (for example, GitHub Copilot or planning assistants) to accelerate planning, generate drafts for reviews, and improve productivity—while maintaining human oversight and review.
  • Embrace shorter spikes and early validation (especially for UX and data science work) to de-risk solutions earlier.

These are patterns, not prescriptions — adapt them to fit your engagement and stakeholder needs.

References

Last update: September 26, 2025