Managing Agents — The Human-Agent Management Relationship
The practice of managing AI agents as autonomous participants — deciding what work to delegate, how to delegate effectively, evaluating output, and developing agents over time.
Thesis: AI makes talent abundant and cheap. What’s scarce is knowing what to ask for — scoping problems, defining deliverables, recognizing quality, giving feedback. Management skills, not AI literacy, are the competitive advantage in the agentic era.
This is distinct from system design (tool-agnostic patterns for building agents) and harnesses (platform-specific knowledge). Managing agents is about the relationship between human and agent — the same relationship a good manager has with their reports, adapted for AI participants.
Thesis-Level Entries
- Talent-to-Direction Scarcity Shift — AI inverts traditional scarcity: talent is now abundant, knowing what to ask for is the bottleneck
Sub-domains
Task-Agent Fit — When to Delegate
Decision frameworks for what work to give agents — when delegation is worthwhile, risk assessment, the jagged frontier of agent capability.
- Delegation Decision Framework — Three-variable equation (Human Baseline Time × Probability of Success × AI Process Time) for when delegation pays off
Delegation Craft — How to Delegate
How to effectively hand off work to agents — instructions, documentation formats, defining done, checkpoints, iteration strategies.
- Delegation Documentation as Agent Prompts — Every field’s delegation docs (PRDs, shot lists, Five Paragraph Orders) are AI prompts; common elements pattern
Evaluation & Feedback — How to Assess
Assessing agent output, giving feedback, iteration cost management, quality recognition, feedback loops.
No entries yet. Planned area — see taxonomy.
Agent Selection & Onboarding — Who to Hire
Choosing which agents/tools for which roles, matching capability to need, initial setup and calibration.
No entries yet. Planned area — see taxonomy.