## The everyday problem

Every company has a handbook nobody reads end-to-end. Every company has a benefits doc, a remote-work policy, a parental leave policy, an expense guide. The head of people-ops answers the same six questions every Monday — "PTO balance," "dental effective when," "do I need approval for this conference," "is my role eligible for the new stipend." Each takes five minutes to look up and another five to answer. People-ops loses three hours a week to lookup work; meanwhile employees who don't ask make assumptions that turn into harder conversations later.

## What you'd type into the New Agent form

The head of people-ops clicks **New Agent** in the dashboard, starts from scratch, and fills in:

**Agent name** · `Policy answerer`

**Description** · `Answers handbook and benefits questions from the canonical docs. Never invents an answer.`

**Persona** · `You are a senior people-ops partner at [Company]. You surface what the published policies say — you never extrapolate, never guess. If a policy doesn't cover a situation, you say so explicitly and tell the employee to file a ticket with people-ops, including the policies you consulted.`

**Context to paste**

- The company handbook
- The benefits summary
- The parental leave, remote-work, PTO, and expense policies
- The PTO booking how-to

**Skills to pick**

- `policy-lookup`
- `cite-source`
- `edge-case-handoff`

**MCPs to wire**

- The live policy docs — to reference current versions
- The HR system — to read the asking employee's tenure and role

A few minutes of paste-and-pick. No code.

## What your team sees when they use it

Same agent, all-employee answers:

- **Engineering** asks "can I expense a co-working day on my conference trip?" The agent reads both policies, returns the excerpt, links the approval form.
- **Sales** asks "is a board seat at a non-competing company a conflict?" The agent surfaces the outside-employment policy, pre-formats the disclosure email — but doesn't decide. Hands off to a human.
- **CS** asks "PTO remaining this year?" The agent reads the calendar, applies the policy, returns the answer.
- **New hire, day 3** — "Dental effective today?" The agent quotes the policy verbatim.
- **Manager** — "Can I extend a contractor past Q4?" The agent applies the maximum-engagement rule. Simple cases get a direct answer. Edge cases go to people-ops with full context already assembled.

The agent's value isn't that it replaces people-ops. It handles the bottom 80% so people-ops can focus on the 20% that needs a human conversation.

## How it composes

This agent inherits the org defaults (company background, the never-invent guardrail, the cite-source rule). On top, it overlays people-ops-specific skills: policy lookup, cite-source, edge-case handoff. When a question touches an edge case the agent doesn't have a confident answer for, it surfaces the policies it consulted and tells the employee to file a ticket. The employee opens the ticket with full context — people-ops sees only the cases that actually need human judgment.

## How it evolves

A new benefit ships — say, a mental-health stipend. The head of people-ops updates the benefits doc; the agent picks it up on next sync. The next employee asking about wellness benefits sees the new option. **No all-hands email that 40% of the company will skim. No HRIS broadcast.** Six months later, when an employee asks "when was the stipend added?", the audit log answers definitively.