Diagram Jun 4, 2026 4 min read

The handoff: from our delivery pod to your own team.

A diagram of the deliberate exit — how a Silicon Prime pod transfers ownership to your engineers across four phases, so the program survives our departure.

KT
Kelvin TranFounder & CEO · Silicon Prime

We sell our own departure. The point of a Silicon Prime engagement is not that we stay forever — it is that your team can run the thing after we leave. A good handoff is not a final week of frantic documentation. It is a slope, planned from day one.

The crossover happens across four phases. We start holding the work; you start watching. By the end, you hold it; we watch. Here is the curve.

OWNS 0% SILICON PRIME YOUR TEAM SHADOW PAIR LEAD OBSERVE 01 · SHADOW We build. You read along. 02 · PAIR We build together. 03 · LEAD You build. We review. 04 · OBSERVE You run it. We are on call, briefly. The crossover point is planned, not stumbled into. After it, your team is the default owner.
Responsibility crosses over four phases — shadow, pair, lead, observe. The dot is the deliberate handoff point.

Phase 1 — Shadow.

We build; your engineers read along. They sit in our planning, watch our reviews, and ask the dumb questions while the answers are cheap. Nobody touches production yet on your side. The goal here is context, not output.

Phase 2 — Pair.

We build together. Your engineers write code in the codebase next to ours, with our people on the same tickets. This is the slowest phase on paper and the most valuable in practice — it is where the tacit knowledge moves, the kind that never makes it into a runbook.

The handoff is not a document you write at the end. It is the thing you were doing the whole time.

Phase 3 — Lead.

Your team leads; we review. They own the tickets, run the releases, and make the calls. We are still in the room, but our role has inverted — we are the second pair of eyes now, not the hands. The first time your team ships a release without us touching it is the real milestone.

Phase 4 — Observe.

You run it. We stay on call for a defined, short window — long enough to catch the things that only show up under your real load, short enough that nobody mistakes us for permanent. Then we leave.

Why the slope, not the cliff.

Most handoffs fail because they are a cliff: a vendor works in isolation for a year, dumps a wiki, and disappears. The knowledge never transferred because the transfer was never the work.

  • Ownership moves continuously, not in one terrifying final week.
  • The crossover point is named in advance — both teams know when your engineers become the default owner.
  • We measure success by what survives us. If the program needs us back in three months, we did the handoff wrong.

A pod that cannot leave cleanly was never a pod. It was a dependency.

— Kelvin Tran. Walnut Creek, CA. May 2026.

← All posts Read next: How context engineering actually works →

Comments