What problem does it solve? Teams and individuals lack a disciplined way to decide which work deserves AI automation and in what form, leading to wasted spend on multi-agent builds that a chat message could handle or automation of tasks that should never be automated. ## Core Features & Use Cases - Four-estimates routing rubric: Scores any task on size, independence, separation, and checkability to assign one of four verdicts — chat, single agent, team, or don't bother — with checkability treated as decisive. - Money-dial veto and build sequencing: Applies recurrence-times-value economics to kill low-value builds, then ranks surviving builds and designs each as a reusable nine-stage rig with a hard human gate. - Job-exposure and org audits: Includes the T/C/L/D audit for role exposure, the six preconditions before AI changes a workflow, the five-property automation test, an interface-type audit for legacy software, and a build/buy/hire/wait checkpoint. - Use Case: A platform lead with a backlog of 15 candidate automations runs the four-estimates screen and money dials on each, downgrades two unjustified multi-agent pitches, and produces a ranked build order in under two hours. ## Quick Start Ask the agent to triage whether a specific task or backlog item should be handled by chat, a single agent, an agent team, or not automated at all.
