Skip to main content

Effort

The AI Build Phase's effort depends on the solution's complexity defined in Phase 3 and the construction structure (sequential or parallel). Parallel construction doesn't necessarily reduce total effort — what it reduces is per-Story File effort, at the cost of increased coordination and synchronization effort.

Effort by complexity

ScenarioWork volumeConstruction structureWhat it includes
Scoped solution (Q1-2)10-20 Story Files1-2 people. Primarily sequential construction. Minimal parallelization.Simple Decision Log. 1 Fidelity Review. Build Validation Report signed by 2-3 stakeholders.
Medium solution (Q2-3)20-50 Story Files3-6 people. 2-3 parallel tracks with 1-2 integration points.Distributed Decision Log. 2-4 integration synchronizations. Partial Fidelity Reviews. Final Review with 5-8 stakeholders.
Complex solution (Q3-4)50+ Story Files6-15+ people. 3-6 parallel tracks with multiple integration points and cross-dependencies.Distributed Decision Log with impact classification. Frequent synchronizations. Active QA Agent on each Story and each integration. Possible partial returns to Phase 3.

The Phase 3 / Phase 4 effort proportion

The proportion that matters

If Phase 4 effort doubles or exceeds Phase 3 effort, the Context Document was insufficient or the parallel construction management was inadequate. Each rework due to track collisions is an investment that should have been made in the dependency graph or schema synchronizations.

If Phase 4 effort is less than half of Phase 3 effort, either the project was very simple, or the team is taking shortcuts in supervision and coordination that will manifest in the Fidelity Review.

Optimal Phase 3 / Phase 4 effort proportion: 1:1 to 1:1.5. If Context Phase was exhaustive and the dependency graph was rigorous, construction — even parallel — is predictable and fluid.

Effort distribution

ActivityProportion in sequentialProportion in parallelWhat determines intensity
Setup (Step 1)5-10%10-15%In parallel: graph verification + collective briefing + distributed Decision Log setup.
Construction (Step 2)55-65%40-50%Individual effort is reduced by parallelization but coordination overhead is added.
Gap management (Step 3)10-15%15-25%In parallel: cross-track gaps require more resolution effort. The escalation protocol is stricter.
Fidelity review (Step 4)10-15%15-20%In parallel: Block 0 (system coherence) is a significant additional step.
Closure (Step 5)5-10%10-15%In parallel: consolidation of Decision Logs from multiple tracks.

Signals of insufficient effort

  • The Decision Log has few entries — either there are no decisions (unlikely) or they're not being recorded.
  • The Fidelity Review produces surprises — stakeholders don't recognize what was built.
  • Agents make decisions that should have been Rules — systematic gaps in the context.
  • Integration between tracks fails repeatedly — the dependency graph was insufficient.

Signals of excessive effort

  • Each Story File requires more supervision effort than execution effort — excessive supervision.
  • The Decision Log has trivial entries that don't require classification — bureaucratic overhead.
  • Integration synchronizations produce no findings — the Context Document was so detailed there's no possible ambiguity (may be correct, but verify if there's over-specification).
  • The Context Engineer stops tracks out of precaution without evidence of a real gap.

Team composition

RoleDedication in sequentialDedication in parallel
Context EngineerPartial supervisionActive cross-track coordination (high-full dedication)
Tech LeadTechnical review (high dedication)Cross-track coherence + synchronizations (full dedication)
Track Dev LeadsN/AFull dedication (execution + coordination)
DevelopersFull dedication (execution)Full dedication (execution)
QA AgentOn demand per Story FileContinuous (validation + cross-track incoherence detection)
Phase 1-2 StakeholdersFinal Fidelity ReviewPartial Reviews + Final Review
Effort comparison across phases

Phase 1 (Problem Phase) concentrates its effort on research and synthesis. Phase 2 (Solution Phase) on alignment and consensus. Phase 3 (Context Phase) on systems design. Phase 4 (AI Build Phase) on supervised execution and coordination. The four phases of discovery, design, and construction together concentrate the majority of project effort. Phase 5 (Market Phase) is continuous and cumulative.