Inquiry Reply Kit

When this helps

You want to reply to inquiries quickly, but rushing means writing discounts or refunds you haven't decided on

What it does

Hand it the text of an inquiry form or email, and it sorts it into questions, bookings, cancellations, complaints, sales pitches or spam, then shows how urgent it is and a target time for replying. Reply drafts are written using only the company's own answers — FAQs, price sheets and similar — and are checked before sending for numbers or promises not found in those answers, or for another customer's details. Matters that should go to a supervisor (refunds, legal language, injuries) are flagged rather than decided. A ledger tracks reply times and what's still open. A person sends the email; the kit never does.

The fixes baked into this kit

1. Replying to a form-notification email sends it back to your own companyThe notification's sender is the company's own no-reply address; the customer's address is in the body. The kit sets the reply-to to whatever is in the body's "email address" field and flags that it did so.
2. Adding "Re:" to the subject duplicates the "様" honorificWhen the notification subject is like "[Inquiry from website] Jane Doe 様", just adding "Re:" duplicates the honorific in the customer-facing subject. For form notifications, the reply subject becomes "Re: Your Inquiry" instead.
3. Paraphrasing a number from the source introduces errorsRewording "14 days" as "two weeks" opens the door to mistakes. Numbers and promises not present in the source answers are flagged to fix, and the kit never calculates totals or refund amounts itself.
4. A quoted earlier message can misclassify the inquiryIn a reply-to-a-reply, quoted text below (e.g. "＞ Cancellations up to 3 days before") can tip classification toward cancellation. Anything below a quoted line is ignored for classification.
5. A product question gets escalated as an injuryQuestions like "can a child with an egg allergy eat this?" were escalated as injuries. Escalation is now limited to actual harm that occurred (vomiting, burns, needing emergency care, etc.).
6. "Cancel without paying for the first shipment" wasn't escalatedCommon subscription-cancellation wording like "without paying" or "cancel the charge" wasn't recognised as a request for money back. This wording was added to the escalation list.
7. Phishing subjects alone are hard to tell apartOf 47 published phishing-email subject lines, only 3 could be told apart by subject alone. Reading past invisible characters and adding look-alike characters and tax/parcel/points-related phrasing as clues raised that to 41.

What's inside

• A system that sorts inquiries and gives an urgency level and reply target time
• A system that drafts replies using only the company's FAQs, price sheets and similar answers
• A check for numbers, promises, or other customers' details not found in those answers
• A system that flags matters to escalate (refunds, legal language, injuries, personal data, etc.)
• A send page with a one-click copy for the recipient, subject and body
• A ledger of cases (open items, totals, questions missing from the FAQ, and so on)
• A sample run with 8 fictional inquiries to a bakery, to try it out

The first three decisions

1. Your company name and the sign-off for replies
2. Your reply target time (e.g. within how many business days)
3. Who to escalate to (refunds, legal language, injuries, etc.)

What you need

• No extra costClaude Code or CodexBoth CLAUDE.md and AGENTS.md are included
• No extra costNode.js 18 or newerNeeded to run the tools. Nothing else to install
• No extra costYour company's own answersFAQs, a price sheet, return/cancellation policy, etc. ★Without these, almost every reply becomes "we'll confirm and get back to you"
• No extra costAn email clientA person sends the email. Copy the recipient, subject and body and send it yourself

How to use

Just paste this into your AI.

I want to use the nice to have kit "Inquiry Reply Kit". Follow https://nicetohave.app/en/g/LCZUUX44-•••••••••• and set it up.

Works the same in Claude Code or Codex.
Read what this tells your AI to do, before you paste it

FAQ

Can it send emails automatically?

No. It only produces the text; a person sends it. Copy the recipient, subject and body from the "send" section of the results page and send it from your own email client.

Can I try it with sample data?

Yes. Say "let me try the sample" and it runs 8 fictional inquiries to a bakery (form notifications, booking changes, cancellations, damaged goods, sales pitches, spam, etc.) all the way through drafting, checking and the ledger.

What happens if there are no company answers (FAQ, etc.) to work from?

Almost every reply becomes "we'll confirm and get back to you." Drafts can't be written without source answers, so provide your FAQ, price sheet and return/cancellation policy first.

Will it tell me whether something is illegal?

No. It only flags what to check (e.g. whether your source answers state a condition under which cancellation isn't allowed); whether it applies is for your company to decide.

Where is personal data stored, and does it leave the company?

The kit's own programs don't connect to the internet or send anything out. The inquiry text and source answers do go to your AI service, though. Storage is a data folder (or a shared folder you choose); since there's no login system, limit who can see it through computer or folder permissions.

Supporters

Once support opens, the people who backed this kit will be listed here.