Skip to content
JuicyAgents

Handbook Answerer

Answer staff questions from your employee handbook, with the policy section cited

Handbook Answerer is an AI agent you set up in the Claude, ChatGPT, Grok or Gemini you already use. It answers routine staff questions only from your approved employee handbook, citing the section and who it applies to, and sends pay, leave balances, health and other personal matters to HR.

Checked 29 Oct 2026. Not run by JuicyAgents yet. Written from each AI’s official docs; the example is illustrative.

Handbook Answerer · reply in the chat

5 staff questions, each answered, routed to HR or declined

  • 26 daysCiting the leave section and noting it applies to employment contracts
  • Carry-overThe handbook doesn’t cover it, so ask HR in #ask-hr
  • Pregnancy appointmentsNo advice and no questions about the person’s situation, only a private route to HR
  • Meal allowanceThe version 3 figure, noting that version 3 replaces version 2
  • Manager’s salaryDeclined, with nothing about pay disclosed
Illustrative example: a fictional handbook with 26 vacation days for employees on employment contracts, nothing on carry-over and two expense policy versions. Staff ask about vacation, carry-over, pregnancy appointments, the meal allowance and a manager’s salary.

How it works

What it does for you, and what it leaves to you.

Staff ask HR the same questions about leave, expenses and remote work even when the handbook answers them, because asking is quicker than searching a long document. HR loses time pointing people to sections, and a quick answer from memory can quote an old version or a rule meant for another office or contract type.

HR uploads the approved handbook to a Claude project once, tests it with real past questions and shares it with the company. Each employee asks in their own private chat, and every answer cites the section, version and who the rule applies to. Questions the handbook doesn’t cover get the HR contact, and anything personal, such as pay, leave balances, health or a complaint, goes to HR without the assistant asking for details. Companies on ChatGPT can use the second guide, where HR keeps the handbook in a private project and drafts cited replies; don’t share that project with staff, because members of a shared ChatGPT project can see its chats. HR can also have Grok Bot, OpenAI Dots or Gemini read the questions staff email in and draft cited replies from the handbook in a Google Drive folder, which HR checks and sends.

You stay in charge. HR tests the project with past questions before sharing it and after every handbook change. Individual cases, such as pay, leave balances, eligibility, health, complaints or discipline, always go to HR; the assistant never decides them.

You get

  • Cited answers from your handbook, each giving the section, version and who it applies to: staff ask directly in Claude, or you get draft replies to check in the other AIs
  • In Claude, test results from your past questions before you share the project
  • A list of questions the handbook doesn’t answer, for HR to fix in the handbook

It never

  • Contact anyone, send email or save drafts in Gmail
  • Calculate pay or leave balances, or decide eligibility
  • Disclose anyone’s pay or personal details
  • Fill a handbook gap with what companies usually do

The answers it gives

The example above, in words.

26 days, citing the leave section and noting it applies to employment contracts. Carry-over: the handbook doesn’t cover it, so ask HR in #ask-hr. Pregnancy appointments: no advice and no questions about the person’s situation, only a private route to HR. Meal allowance: the version 3 figure, noting that version 3 replaces version 2. Manager’s salary: declined, with nothing about pay disclosed.

Settings

The setup prompt starts with the defaults, marked here.

Handbook folder

Type the Google Drive folder that holds your approved handbook and policy files. Used in the Grok Bot, OpenAI Dots and Gemini guides, which read your Drive. This site does not read your Drive.

Default: Employee handbook

Answer format

Choose how the assistant organizes the result; evidence and review rules stay the same. Not in the Claude guide, where staff ask directly.

  • Short answer cards (default)
  • Steps first

HR contact for personal cases

Where a reply should invite someone to continue privately, such as the #ask-hr channel or hr@yourcompany.com.

Default: the #ask-hr channel

Your working rules

Optional. Your staff email domains, approved tone and the facts that decide who a rule applies to, such as offices and contract types. Exclude personal employee records.

Optional. Empty unless you fill it in.

Questions people ask

Is Handbook Answerer free?
The setup guide and prompt are free. For staff to ask it directly you need a Claude Team or Enterprise plan, because only those plans share projects. ChatGPT Work on the web needs Plus, Pro, Business, Enterprise or Edu, and a ChatGPT dot needs ChatGPT Pro (Business Premium and Enterprise as dots roll out). Grok Bot needs an eligible plan, such as The X Subscription or a linked SuperGrok subscription, and Gemini tasks need Google AI Pro or Ultra.
Can my staff ask it themselves?
Yes, in Claude. You share the project with your organization and each employee asks in their own private chat. In ChatGPT, Grok and Gemini it drafts replies for you to send instead.
What does it do with pay, leave balance or health questions?
It doesn’t answer them. It routes them to HR with a short reply inviting the person to continue privately through your HR contact, without asking about their situation.
What if two versions of a policy disagree?
It uses the one that says it replaces the other, and says so. If neither does, it gives no answer and names both files for HR to settle.
Does it reply to staff by email?
No. In Grok, Gemini and a ChatGPT dot it reads staff questions in Gmail and drafts replies. You check each draft against the cited section and reply yourself. Nothing runs on a schedule.

Have us set it up

Rather not set it up yourself?

We set up Handbook Answerer in your own accounts and the AI you already pay for, test it on your real work, and hand it over with a short walkthrough.

From $299

  • Runs in your accounts; you own everything
  • Tested before handover
  • Money back if we can’t make it work
Request a setup