Subagents — simulate the stakeholders who would review your work
Subagents are a Claude feature. Each one is a focused assistant that runs in its own conversation with its own instructions, so it can review your work without cluttering your main session. In Personal PM OS, each subagent simulates a stakeholder you'd normally hand work to for review — a CTO, a UX lead, a sales lead, and more.
When to use this
- You want stakeholder review on a document before it goes wider.
- You're missing a perspective on your team — no CTO, no UX lead, no sales counterpart.
- You want to pressure-test a deliverable before sharing it.
Using subagents
- Pick the document you want reviewed (a PRD, roadmap, strategy memo).
- Invoke the subagent in Claude Code and point it at the file.
- Run multiple subagents in parallel on the same document for a 360° read across perspectives.
How it works
Each subagent reads your context files and the document you point it at, then returns a critique from that perspective — what's missing, what's weak, what would get pushed back on. Run one subagent for a single lens, or several in parallel for a 360° view before you ship.
The 6 subagents included with Personal PM OS
Personal PM OS comes with these six subagents out of the box. You can also customize each one to better match your team, or create new subagents for the stakeholder lenses you rely on most.
| Subagent | Asks |
|---|---|
| CTO / Tech Lead | "Can we actually build this? What are the risks?" |
| UX / Design Lead | "Will users understand this? Where will they get confused?" |
| Sales / GTM Lead | "Can we sell this? How does it compare to competitors?" |
| Executive | "Does this move the needle? Is this the best use of our time?" |
| User Advocate | "Does this actually solve the problem? What about edge cases?" |
| Devil's Advocate | "What are we missing? What's the worst-case scenario?" |