# Local Services Agent Kit: booking, reminders, quote follow-up and review asks, with the owner's yes on every message

> The layer around an AI receptionist for shops that book by the hour. Five desks (intake, booking, reminders, quote follow-up, reviews) draft every customer message into one outbox, the owner approves them from a twice-daily queue with a one-line reply, and a read-only Python check catches thirteen kinds of mistakes before anything goes out. Includes a seeded test ledger with thirteen planted faults and the fixes.

- URL: https://paperchasewebb.ai/bots/marketplace/items/local-services-agent-kit-booking-reminders-follow-up-review-asks/
- Type: Bot template
- Category: Solo business
- By: PaperChaseWebb
- Works with: Claude Code, Codex, Grok Bot, Hermes (named in the listing); Plain .md plus one Python 3 script
- Price: $39 USD
- Difficulty: Intermediate
- Setup time: 5 hr 30 min
- Time to result: Ledger, templates, tests and a watched go-live in one working day
- Format: Markdown (.md) file, local-services-agent-kit-booking-reminders-follow-up-review-asks.md
- Version: 1.0, updated 2026-10-28
- License: Use on your own business and on installs you do for clients. Do not resell or republish the kit itself.

## What you end up with

Every hold, confirmation, reminder, follow-up and review ask waits for the owner's yes, nobody gets double booked or texted after STOP, and every finished job gets the same review ask.

## What you get

- 23 sections in one .md file
- Five desks with what each may do and what it never does
- The owner queue: approval cards and a one-line reply format (yes, edit, kill)
- Single and standing approval rules, and the kinds that always need their own yes
- A five-file ledger: kit.json, customers.csv, jobs.csv, outbox.csv, approvals.csv
- Contact rules for consent, STOP, quiet hours and a weekly cap
- Hold, confirm, reschedule and cancel steps; one confirmation and one reminder per booking
- A three-touch quote follow-up with stop rules
- A review desk that sends the same ask to every finished job, with no gating and no perks
- 13 message templates and 12 prompts
- outbox_check.py: a read-only check with 13 checks, exit codes for a send gate
- A seeded test ledger with 13 planted faults, the 13 fixes and the expected output
- A rules block and harness notes for Grok Bot, Claude Code, Codex and Hermes
- A one-day install plan, go-live checklist, weekly review and failure table
- A teardown of eight receptionist and home services kits, with sources

## Who it is for

Installers and owners of local businesses that book by the hour, who already answer leads (by hand or with an AI receptionist) and lose jobs, or annoy customers, in the gap between the call and the finished job.

## Tools needed

- Grok Bot, Claude Code, Codex or Hermes: runs the five desks: drafts messages into the outbox and builds the owner's queue; never sends
- Python 3: runs the read-only outbox check before anything goes out; no AI needed
- Any spreadsheet app: lets the owner open the five ledger files
- The shop's texting tool and calendar: the separate send step: sends approved rows only, after a clean check
- You need: An agent harness that can read and write files in one folder
- You need: Python 3.8 or newer for the check
- You need: A single real calendar for the shop
- You need: A texting tool you control, for the separate send step

Cost to run: The kit is the only purchase. Your texting tool, phone number and calendar are billed by those vendors.

## How it works

1. **Test the check.** Save outbox_check.py and make_fixture.py from the file, build the seeded ledger and run the check on both folders. Expected: 13 findings on seeded, then 0 on fixed.
2. **Set up the shop.** Make the ledger folder, set kit.json with the owner, write the services list and fill the templates. Expected: The check prints 0 on the shop's own ledger.
3. **Paste the rules block.** Put the rules block and desk sections into your harness and remove texting, email and calendar write tools from desk sessions. Expected: A desk session can draft but has no way to send.
4. **Run the queue.** Start the 7:30 AM and 1:00 PM owner queue. The owner approves every message one by one for two weeks. Expected: The owner replies with a yes, edit or kill line and approvals are recorded with the reply time.
5. **Wire the send step.** Point the shop's texting tool at approved rows only, behind the check's exit code. Expected: Nothing goes out when the check finds a problem; the seeded ledger proves it.

## Sample prompt: Review asks for every finished job (Grok Bot, Claude Code, Codex, Hermes)

```
List jobs marked done in the last three days with consent on record and no review ask. Draft review-v1 for each, with no changes to the text.
Do not skip anyone for any reason; if a job had a complaint, flag it as NEEDS OWNER and still draft the ask.
```

Prompt 8 of 12. Runs after the rules block in section 14.

## Sample prompt: Record the owner's yes (Grok Bot, Claude Code, Codex, Hermes)

```
Here is the owner's reply: '<reply>'. For each number, write the matching approval row with the reply's time, set yes items to approved, send edit items back to their desk with the note, cancel kill items.
Ask about anything that does not name a number. Then run the check.
```

Prompt 11 of 12.

## Free preview of local-services-agent-kit-booking-reminders-follow-up-review-asks.md

````markdown
# Local Services Agent Kit: booking, reminders, quote follow-up and review asks, with the owner's yes on every message

Version 1.0, 2026-10-28. By PaperChaseWebb. Works with Grok Bot, Claude Code, Codex and Hermes. Plain Markdown plus one Python 3 script.

This kit is the layer that sits around an AI receptionist. The receptionist answers. This kit handles what happens after the call: the hold, the confirmation, the one reminder, the quote that went quiet, and the review ask after the job is done. Every message a customer could receive goes into one outbox file, waits for the owner's yes, and passes a read-only check before anything goes out.

It does not repeat the call script, the call map or the twelve test calls. Those live in the AI Receptionist and Booking Agent Template (pp-ai-receptionist, https://paperchasewebb.ai/bots/marketplace/item/ai-receptionist-booking-agent-template-script-capture-escalation-tests/). You can use this kit without it, with any phone setup, web form or text line that hands you a lead.

## Contents

1. What this kit does, and how it fits with the receptionist
2. Who it is for, and when not to use it
3. The five desks and the owner queue
4. The ledger: five small files
5. Approval rules: single yes, standing yes, and what can never be standing
6. Contact rules: consent, STOP, quiet hours, the weekly cap, the right person
7. Booking desk: hold, confirm, reschedule, cancel
8. Reminder desk: one confirmation, one reminder
9. Quote follow-up desk: th
````

## Worth knowing

- It does not cover the call itself; that is the AI Receptionist and Booking Agent Template. It works with any phone setup, form or text line that hands over a lead.
- It never sends, books or charges. A separate send step you control sends approved rows only.
- Not for medical, dental, legal or financial intake, payments or cold outreach.
- Texting consent, AI disclosure and business texting registration rules differ by place. It is a method, not legal advice.
- On 2026-10-28 outbox_check.py reported the thirteen expected findings on the seeded ledger, then 0 after the listed fixes, and left every file unchanged. The desks and prompts were walked through against it as a design check. Run the seeded ledger in your own harness first.

## Questions

**How is this different from the AI Receptionist template?** The receptionist answers the call and finds a slot. This kit handles everything after: the hold, the owner's confirmation, one reminder, quote follow-up and the review ask, all through one outbox and the owner's yes. They share service ids and fit together, and neither repeats the other.

**Does it send texts automatically?** No. Desks only draft. The owner approves from a queue, and the shop's own texting tool sends approved rows only after the check comes back clean. After two clean weeks the owner may sign a standing yes for fixed reminder, missed-call and review templates.

**Why can't I ask only happy customers for reviews?** Google's Maps policy does not allow merchants to selectively ask for positive reviews, and the FTC says doing so could violate the FTC Act. The kit sends the same ask to every finished job and the check flags any job that was skipped.

**Which tools does it work with?** Grok Bot, Claude Code, Codex and Hermes, with harness notes for each. The ledger is plain CSV and the check is standard-library Python.

## How to buy

$39 one time at https://paperchasewebb.ai/bots/marketplace/items/local-services-agent-kit-booking-reminders-follow-up-review-asks/. Stripe checkout, then an instant download page; the download link is also in your emailed receipt.

Seller: PaperChaseWebb Inc., mgmt@paperchasewebb.com