The dilemma
SoulCode could sprint inside teams but stumbled between them. Dependencies hid in chats. SLAs were soft. Changes landed without notice. Escalations bounced across org lines. Work looked busy. Throughput lagged.
What was fractured
Hidden dependencies
Interfaces lived in people’s heads. Surprises were normal.
Soft SLAs
“ASAP” replaced response targets and acceptance tests.
Orphaned seams
No single owner for a handoff. Both sides assumed the other had it.
No change discipline
Updates shipped whenever. Breakage followed.
Noisy escalation
Two paths. Two stories. Slow decisions.
Rework loops
Handoffs failed first pass, then cycled back without learning.
Quinn’s playbook: Make interfaces first-class
1) Interface contracts
Write the seam. Inputs, outputs, SLAs, acceptance tests, rollback steps. Public doc.
2) Driver–steward pairing
Name one driver per interface and a steward on the partner team. They meet weekly.
3) Request–commit protocol
Ask with context and target. Get an estimate. Agree. Deliver. Verify against the contract.
4) Change windows
Publish windows, notice periods, and blackout dates. Link to test plans.
5) Integration tests
Pre-merge checks and synthetic probes on the seam. Alert on failures.
6) Single-path escalation
Escalate together with a one-pager: issue, options, evidence, attempts. One room. One call.
7) Handoff checklist
Before you pass: owner named, evidence linked, test status green, rollback ready.
8) Interface scoreboard
Track handoff time, first-pass acceptance rate, blocked hours, and rework. Review monthly.
Why it matters now
Autonomy moves fast only if seams are clean. Contracts, owners, and windows convert coordination into flow. Decisions travel a short path. Surprises shrink. Speed becomes a habit.
The shift at SoulCode
Within one quarter, handoff time fell, first-pass acceptance rose, and “surprise” escalations dropped. Teams reported fewer resets and clearer shared expectations. Delivery felt calmer and faster.
Leadership takeaways
Treat seams like products.
Name owners on both sides.
Publish windows, not wishes.
Decisions travel one path.
Measure the interface, not just the team.
Teaser for Episode 27: The Capacity Ledger
Quinn exposes hidden work, sets hard caps on ad hoc, and rebuilds plans around real capacity so teams stop slipping quietly.
