What it is
Individual practitioners become remarkably effective with agents, evolving personal workflows — their own prompts, tools, session habits, and steering instincts — that produce excellent local results. But personal cleverness does not automatically compose into a team operating model. A dozen people each running their own private, undocumented workflow is not an organization that operates well with agents; it is a dozen islands whose practices cannot be reviewed, transferred, governed, or improved together. Worse, agents optimize locally by default, and the best local answer can quietly violate the organization's strategy, policy, or shared taste — a workflow that is optimal for one engineer's task may pull against where the whole system needs to go. The fix is to deliberately build organizational topology on top of personal effectiveness: shared artifacts (specs, plans, decisions, logs) that make work portable across people; shared review norms so quality is a team property rather than an individual habit; shared observability so the organization can see what its agents are doing and spending; and a shared pattern library so a useful operating move discovered by one person becomes available, with its failure modes, to everyone. The aim is not to flatten individual ingenuity but to make it compose — to turn private leverage into an aligned operating model.
When to reach for it
- When several people have each become effective with agents in isolation and their practices cannot be reviewed, transferred, or improved collectively.
- When a locally optimal agent workflow keeps producing results that drift from team strategy, policy, or shared taste.
- When valuable operating moves live only in individuals' heads or private setups, capture them as shared patterns with usage guidance and failure modes.
- When onboarding is slow because there is no shared operating model to hand new people, only a scatter of personal workflows.
- When the organization cannot see, across people, what its agents are doing or costing, invest in shared observability and review norms.
When NOT to
- Do not standardize prematurely; in early exploration, let personal workflows diverge and prove themselves before you harden them into shared norms.
- Avoid bureaucratizing individual effectiveness into compliance overhead that kills the ingenuity you were trying to compose.
- This is not a mandate that every practice be identical; the goal is alignment and portability, not uniformity for its own sake.
- Do not impose shared topology where the work is genuinely one person's isolated, low-stakes domain with nothing to compose against.
Exemplars
- Mel Conway, "How Do Committees Invent?" (Datamation, 1968) — http://www.melconway.com/Home/Committees_Paper.html — classic antecedent: demonstrates that systems mirror their organization's communication structure; the "inverse Conway maneuver" — deliberately restructuring team topology to achieve a desired system shape — is a later practitioner extension of Conway's Law (not from the 1968 paper) and grounds this pattern's prescription to build shared artifacts and review norms on top of personal effectiveness