Starting Implementation
📝 Context
You're kicking off an implementation with a customer. The goal is to establish a clear starting point, confirm assumptions from discovery, and set up for success.
📋 Before You Start Checklist
Confirm the basics:
- [ ] Do you have access to the customer's environment?
- [ ] Is the environment ready? (Cluster exists, namespaces created, permissions granted)
- [ ] Are artifacts available? (Container images accessible or need transfer)
- [ ] Do you have a primary point of contact on the customer side?
- [ ] Is there a communication channel established?
- [ ] Do you have documented scope and success criteria?
Validate assumptions:
- [ ] Re-confirm environment constraints
- [ ] Verify versions (Kubernetes, Helm, dependencies)
- [ ] Check for any changes since discovery
🎯 Kickoff Call Agenda
- Introductions (if new people) ⏱️ 5 min
- Recap scope and success criteria ⏱️ 5 min
- Environment walkthrough ⏱️ 10-15 min
- Communication and cadence ⏱️ 5 min
- Timeline and milestones ⏱️ 5-10 min
- Questions and concerns ⏱️ 5 min
🎯 Implementation Phases
Phase 1: Environment Validation
- Confirm cluster access
- Verify networking
- Check storage classes, ingress controllers, DNS
- Document deviations from expected setup
Phase 2: Artifact Staging
- Transfer container images if air-gapped
- Deploy Helm charts or manifests
- Configure environment-specific values
Phase 3: Core Deployment
- Deploy the application
- Validate basic functionality
- Run smoke tests
Phase 4: Integration
- Connect to customer systems
- Configure for specific use cases
- Test end-to-end workflows
Phase 5: Validation & Handoff
- Run through success criteria with customer
- Document what was deployed and how
- Create runbooks for day-2 operations
- Conduct handoff session
During Implementation
- Communicate proactively - don't go silent
- Document as you go
- Timebox troubleshooting ⏱️ ask for help after 30-60 minutes
- Validate incrementally
⚠️ Gotchas
- "It worked in my environment" - customer environments differ
- Permissions gaps - discover this early
- Undocumented dependencies - ask what else is running
- Scope creep - know when to say "let's add that to a follow-up"
- Customer availability - build buffer into timeline