Skip to main content

How Teams Work Together

Sailfish is a cross-team operating system for customer issues. For it to work, each team needs to know exactly what they own. Without this clarity, everyone assumes someone else is handling it — and nobody does.

Engineering​

What they do:

  • Jump into deep debugging when Sailfish surfaces an issue with full context
  • Validate that telemetry is flowing correctly from their services
  • Fix issues — either manually or by reviewing Sailfish's auto-fix PRs
  • Close the loop on engineering tickets

What they don't do:

  • Triage every issue manually — Sailfish prioritizes by impact automatically
  • Guess reproduction steps — the full execution trace is already there
  • Search through logs, dashboards, or APM tools — everything is in one place

Product​

What they do:

  • Prioritize issues based on customer impact, revenue exposure, and severity
  • Decide what gets fixed now vs. later vs. never
  • Monitor trends — are issues increasing? Is a release causing regressions?
  • Use impact data to inform roadmap decisions

What they don't do:

  • Investigate root causes — that's Engineering's job, and Sailfish gives them the context
  • Manually count affected users — Sailfish calculates blast radius automatically

Support / Customer Success​

What they do:

  • Surface issues from customer conversations (Slack, Zendesk, email)
  • Link customer reports to issues in Sailfish so the technical context is attached
  • Communicate status back to customers — "We've identified the issue and are working on a fix"
  • Flag high-priority customers who need proactive outreach

What they don't do:

  • Diagnose technical root causes — Sailfish handles that
  • Manually escalate to Engineering with incomplete context — Sailfish creates tickets with everything Engineering needs

The Handoff Flow​

Customer reports issue
        ↓
   Support links it in Sailfish
        ↓
   Sailfish shows impact + affected users
        ↓
   Product prioritizes based on impact
        ↓
   Engineering debugs with full context
        ↓
   Fix shipped → Support closes the loop with customer

When this flow is running, issues go from report to resolution without anyone chasing context, scheduling meetings, or writing reproduction steps.