Batch Amplification · Primary
Problems are found at release, not before it.
How this principle feeds it
work, dependency and risk accumulate before evidence can change direction, as the library defines it.
Principles Library / Progressive Commitment
Progressive Commitment Reduces Uncertainty Exposure
Limit the size, scope, and reach of a commitment while important uncertainty remains, then expand it as credible evidence emerges.
Commit to an outcome, not to a fixed scope or output, so execution can adapt while intent stays stable.
You find out an assumption is wrong while changing direction still costs little.
Teams can change how they work as evidence comes in without losing the outcome.
You expand only what you understand, so it still fits when it reaches a new context.
Problems are found at release, not before it.
work, dependency and risk accumulate before evidence can change direction, as the library defines it.
The team owns the commitment. Nobody owns the gap.
a large commitment pulls in capabilities owned by other teams, and each one makes the commitment harder to unwind.
Delivery is consistent. Value is not landing.
delivery is steady and on schedule while value stops reaching the customer.
Every quarter starts over. Nothing compounds.
when a large bet is abandoned late, the plan resets around a new one and little carries over.
Everything is in progress. Nothing is finishing.
bets that should have stopped keep holding people and budget, so more work stays in progress.
The team built exactly what was asked. It is no longer what is needed.
execution stays consistent with the specified scope after that scope stops reflecting the need.
Every team hits its targets. The product still misses.
one team's local win is treated as proof for the whole system, and the end-to-end result suffers.
Customers find defects before the team does.
failure modes that were never studied surface as defects and incidents in each new context, after they are cheap to fix.
AI signs are on by default. Use the AI switch to exclude them.
For each consequence: Not seen / Sometimes / Persistently / Not sure.
Based on your answers, recognize at most two conditions: No / Partly / Yes, that is us.
Batch Amplification: Problems are found at release, not before it.
Dependency Density: The team owns the commitment. Nobody owns the gap.
Customer Disconnect: Delivery is consistent. Value is not landing.
Strategic Volatility: Every quarter starts over. Nothing compounds.
Workload Saturation: Everything is in progress. Nothing is finishing.
Intent Drift: The team built exactly what was asked. It is no longer what is needed.
Local Optimization Bias: Every team hits its targets. The product still misses.
Quality Fragility: Customers find defects before the team does.
What else have you observed? Optional. Shared only if you choose to start coaching.
Start a Diagnosis session: review and edit a draft containing your note, conditions you recognized in Step 3 (including whether they match partly), warning signs and consequence ratings before sending it. No scores or principle files are sent.
Showing all 6 practices. Answer the Diagnose questions and the most relevant come first.
Start with the smallest commitment that can produce useful evidence. Test or deliver increments before you lock in the full solution.
Start by asking your team: What's the smallest meaningful commitment that would give us useful evidence?
Commitments start small, and core assumptions get tested before the full solution is locked in.
Commit to the outcome intent and leave scope open. Hold the intent steady while execution adapts, and judge progress by how the outcome moves.
Start by asking your team: Are we committed to an outcome or to a scope? What can change based on evidence, and what has to stay stable?
Goals state the outcome, and progress is reported as outcome movement.
Cut batch size and remove the delays between build, validation and release, so feedback shapes direction while it's still cheap to act on.
Start by asking your team: As AI makes execution cheaper, is our lead time actually shrinking, or just our output?
Feedback arrives while scope and architecture are still open to change.
At each stage, use the evidence to decide whether to continue, change, stop or expand. Don't lock scope, architecture, investment or rollout earlier than you need to.
Start by asking your team: What evidence would justify the next level of commitment, and where have sunk costs made changing direction harder than it should be?
Teams change course on new evidence and aren't penalized for it.
Scale a practice or solution only once you understand how it behaves and fails. Adapt it to local context instead of forcing a uniform rollout.
Start by asking your team: What are we scaling before we fully understand why it works, and does it still fit this context?
Pilots expand on solid evidence, and teams know why an approach worked before they copy it.
Constrain agents around outcome signals. Expand agent autonomy as controls and recovery paths are proven, not on a fixed timeline.
Start by asking your team: Which agent permissions have we expanded before proving we can catch and undo its mistakes?
Agent autonomy grows only as failure modes and recovery paths are understood, and agent reporting tracks outcomes.