What does it mean that portfolio rebalancing?

Periodically review and rebalance your agent portfolio — retire underperformers, invest in high-value agents.

Example: A product manager runs six cognitive agents: a daily prioritization agent, a stakeholder communication agent, a risk assessment agent, a sprint retrospective agent, a competitive analysis agent, and a personal energy tracking agent. During a quarterly rebalance, she notices the competitive analysis agent hasn't been triggered in eight weeks — her market position stabilized and the agent's output stopped changing her decisions. Meanwhile, her risk assessment agent fires every day but only covers technical risks, missing regulatory exposure that nearly blindsided the team. She retires the competitive analysis agent, frees up the cognitive overhead it consumed, and reinvests that capacity into splitting the risk assessment agent into technical and regulatory variants. Same total number of agents. Radically different portfolio fitness.

Try this: List every active cognitive agent you maintain — every recurring process, checklist, decision framework, or structured routine you run regularly. For each one, score two things on a 1-to-5 scale: (1) how often it actually fires in a typical week, and (2) how much its output changes your behavior when it does fire. Multiply the scores. Anything scoring below 6 is a rebalancing candidate: either retire it, merge it with another agent, or redesign it to increase behavioral impact. Anything scoring above 15 is a candidate for investment — consider splitting it into more specialized variants or increasing its trigger frequency.