The Problem
Pick the version that's yours.
// three modes, one failure pattern
TEAM: You sent the email. “Effective Monday, we're adopting AI-assisted development across the team.” Two weeks later, DORA metrics are flat and half the team has gone underground using AI on the side, hiding it from PRs because nobody wants to be the one who shipped “agent slop.”
SOLO: You read the launch post and thought “I should AI-ify my workflow.” A month later you've sprinkled half-finished agent setups across 8 different tools, your prompt library is 200 untested fragments deep, and you can't remember which one you actually trust.
BOT: You shipped the new agent capability to 100% of production traffic on a Tuesday because “the demo went great.” Friday, three customer support tickets land claiming the agent broke their workflow. You don't know how many silent failures you missed before the loud ones surfaced.
Three modes, one failure pattern: adoption without governance. The mandate (or the binge, or the 100%-on-day-one ramp) substitutes enthusiasm for discipline, and the artifact that should have proven the practice never gets built. Six weeks in — whichever mode you're in — you've spent the attention, the tokens, or the trust, and have nothing measurable to show for it.
The Core Insight
Mandates produce shame regardless of actor. Adoption is a launch — for a team, for yourself, for a bot — not a memo. The governance is the launch discipline.
Most rollouts skip the part where you have to actually sell the new way of working to the people (or systems) who'll do the work. The mandate substitutes authority for desire, and authority doesn't transfer the skill, the trust, or the calibration that makes adoption durable. The team complies in the visible parts of their workflow and routes around it where you can't see. The solo operator collapses into tool-thrash. The bot rollout ships before its kill switch exists.
The pattern that works treats adoption as a staged launch with explicit entry criteria for each transition — the shape is universal across modes, but the mechanism changes. For a TEAM, the stages are champions → opt-in → default. For a SOLO operator, the stages are written criteria → time-boxed trial → keep/kill decision. For a BOT, the stages are feature flag → traffic ramp → default-on. You don't move stages because the calendar says so — you move when the metrics that predict success have crossed their threshold.
// across actors — adoption governance, not just team adoption
TEAM: Champion-led rollout, not top-down mandate.
SOLO: Personal practice pilot — pick one workflow, run a time-boxed trial with written success criteria, then keep or kill.
BOT: Gradual capability rollout — feature flag, blast-radius cap, kill switch, staged traffic ramp.
The three walkthroughs below cover one each. The shape is the same; the mechanism changes per actor.