Photo Studio Admin: Draft Booking Replies, Plan Info & Delivery Emails with Claude Code

During the busiest week of the fall family-portrait season, a small photo studio a friend of mine runs had one of those moments.

She thought she’d replied to a family’s booking email cleanly. But the greeting still had the previous customer’s name on it, and the price sheet she attached was last year’s old PDF. She caught it three minutes after hitting send, and had to fire off a groveling correction email right behind it. The front-desk assistant was nearly in tears.

If you run a photo studio, you know the drill. The thing that grinds you down has nothing to do with your skill behind the camera. It’s the booking inquiries, the “what to bring on the day” notes, the “when will my photos be ready?” nudges. None of it is hard to write. There’s just a lot of it, and every message is slightly different from the last. So it wears you out, and mistakes slip through.

The fix is to hand the first draft to AI. Not the whole job, just the draft. Below is what I actually tested, shaped so a studio’s front desk or photographer can use it as-is.

Key takeaways

• Booking replies, shoot-plan info, and delivery emails are roughly 80% boilerplate. That’s exactly what makes them a good fit for AI drafting.
• Claude Code’s job stops at “the draft.” Dates, prices, and the send button are always confirmed by a human.
• You get a copy-paste prompt template plus a check script that stops the wrong price sheet from going out.
• At 20 emails a day, the math points to saving around 30+ hours a month.
• Customer names, addresses, and children’s photos are sensitive personal data. Be deliberate about what you feed the AI and how you handle logs.

Who this is for, and what your workflow looks like now

This article is for the small-to-midsize studio that shoots family and milestone portraits, where the photographer also handles the front desk and email, or where one front-desk person runs the entire booking pipeline. You love the shooting. You just lose two or three hours a day to admin.

Lay out a photo studio’s flow from inquiry to delivery and it looks about like this:

1. An inquiry arrives (phone, email, Instagram DM, booking form).
2. You check availability and reply with candidate dates and pricing.
3. After the booking is confirmed, you send the shoot plan plus what to wear and bring.
4. On shoot day, you check the client in and settle payment.
5. After the shoot, you tell them when the files and prints will be ready.
6. When the photos are done, you send the delivery note plus a review request and any add-on offers.

Steps 2, 3, 5, and 6 are “writing” work. More of your time goes into the messages around the shoot than into the shoot itself. That’s where AI earns its keep.

The rework and headaches that keep happening

Here are the “yeah, that happens” moments I heard straight from studio owners, paired with their root cause.

Headache	Common cause	The rework it creates
Wrong name or date	Copied the last email and forgot to edit it	Correction email, dented trust
Sent an old price sheet	File management lives in one person’s head	Pricing disputes, re-sending
Missed a “what to bring” item	Too many plan-by-plan differences to remember	Day-of confusion, reshoot requests
Late delivery-date notice	”When do I send it?” only exists in your head	More chasing inquiries
Inconsistent tone across replies	Each staffer writes differently	Brand voice never settles

None of these is a writing-skill problem. They’re information-handling problems: too many messages, too many plan-by-plan differences, and “latest info” managed by hand. That’s precisely why an AI that drafts to a fixed template helps.

Use case 1: Draft a reply to a booking inquiry

Paste in the inquiry email and have the AI draft a reply with candidate dates and pricing. You do not send it. Draft only.

All the photographer or front desk has to check is two things: are those dates actually open, and is the price current? The time you’d spend writing from a blank page disappears entirely.

What a human must always decide:

• Do the proposed dates match the booking ledger?
• Does the price match this season’s price list?
• Are easy-to-misread terms, like the cancellation policy, stated correctly?

Use case 2: Build the shoot-plan info and packing list

Christening, first-birthday, graduation portraits, family shoots. Each plan has different clothing, items to bring, and run time. Have the AI assemble this automatically from the plan name.

Keep one “differences-by-plan” table on hand, and the AI builds the right info text from it. That rework moment, where the front desk suddenly realizes “oh no, I forgot to mention the formal shoes,” gets a lot rarer.

Use case 3: Pair the delivery note with the review request

After the shoot, the “your photos are ready” note wants to bundle the delivery date, the download instructions, the add-on print offer, and a review request into a single email. But write it from scratch every time and it balloons.

The template here is completely fixed, which makes it the single best spot for AI drafting. Swap in the delivery date and the gallery URL and you’re done.

What to delegate to AI vs. decide yourself

Leave this line blurry and you get the accident from the opening. So let’s draw it sharply.

Task	Hand to AI	A human must decide
Drafting the message	Yes	—
Tone and politeness adjustments	Yes	—
Proposing candidate dates	Drafts them	Confirm against the ledger
Amounts and price sheets	Slots into the template	Confirm it’s current
The send button	—	A human always presses it
Registering personal data	—	A human always handles it

One rule covers it. Dates, amounts, and sending: a human confirms all three last. Decide that everything else about composing the message can go to the AI, and the accidents stop. For more on how to think about what you delegate, see Claude Code for non-engineers.

A copy-paste prompt template

Use it as-is. You only paste in the customer’s email body.

You are the front desk at a family photo studio. For the inquiry below,
write a draft reply in a warm, polite tone.

# Rules to follow
- For candidate dates and pricing, write [the correct value I confirmed myself],
  which I will swap in later
- Do not assert a specific amount; only use amounts I have confirmed
- For the cancellation policy, write "We'll share this once your booking is confirmed"
- End the signature with "[Studio Name] Front Desk"

# Inquiry body
(paste the customer's email here)

# Output
- One subject-line option
- The body (under 120 words, address the recipient by name)

The point is not letting it decide amounts or dates on its own. Make them placeholders, and assume a human swaps them in. For more on building prompts like this, advanced prompt engineering techniques is a good reference.

A check script: the gatekeeper that stops the wrong price sheet

Have a machine check whether the draft still contains a leftover “swap me” placeholder or an old year’s text. It runs anywhere Node.js is installed. Pass every draft through this before sending, and the “last year’s price sheet” accident from the opening simply goes away.

// check-draft.mjs : a gatekeeper that checks delivery/info email drafts
import { readFile } from "node:fs/promises";

const draft = await readFile(process.argv[2] ?? "./draft.txt", "utf8");
const thisYear = String(new Date().getFullYear());

const rules = [
  { ng: /\[.*?(swap|placeholder|TODO).*?\]|\[the correct.*?\]/i, msg: "A swap-in placeholder is still present" },
  { ng: /\b(Dear ,|\[Name\]|recipient name)\b/i, msg: "The recipient's name is not filled in" },
  { ng: /\b20(1\d|2[0-4])\b/, msg: `An old year is still present (this year is ${thisYear})` },
  { ng: /(http:\/\/|test url|example\.com)/i, msg: "A non-production URL is still present" },
];

const hits = rules.filter((r) => r.ng.test(draft));

if (hits.length === 0) {
  console.log("OK: passed the pre-send check");
} else {
  console.log("Send blocked. Fix the following:");
  for (const h of hits) console.log(" - " + h.msg);
  process.exit(1);
}

Running it is just this:

node check-draft.mjs ./draft.txt

If an old year, a forgotten placeholder, or a test URL is still in there, it stops with exit code 1. Putting one gatekeeper in place, instead of relying on human eyes alone, cut my evening review time sharply. For how to lock rules like this into a project, I’ve gathered the approach in CLAUDE.md best practices.

Before and after, plus a rough ROI estimate

Before, each reply took about 8 minutes on average. Read the inquiry, dig up the old email, copy and edit it, check the price list, and attach it. At 20 emails a day, that’s around 2.7 hours.

After, the AI writes the draft and the human only confirms and swaps in values. Each one drops to about 3 minutes. Twenty emails is an hour. That’s roughly 1.7 hours saved per day, about 37 hours a month over 22 working days. At $15 an hour, that’s the scale of freeing up over $500 worth of time each month.

Of course it never works out exactly on paper. But just deleting the “write the message from a blank page” step changes how heavy the day feels. For tips on folding this into daily operations, Claude Code productivity tips helps.

Security and personal-data notes

A photo studio handles names, addresses, phone numbers, and pictures of children’s faces. That’s personal data to guard carefully. When you hand it to the AI, set the line in advance.

• Don’t give the AI information the message doesn’t need, like addresses and phone numbers. A reply only needs the name and the gist.
• Don’t upload the photos themselves, or any file with a child in it. A delivery note can be built from “the URL and the name” alone.
• Confirm in the contract or terms whether customer data is used for training. For business use, choose a plan where your data is not used to train models.
• Don’t leave drafts or logs sitting in a shared folder. Make deleting them after review part of the routine.

When in doubt, the principle is simple. Ask once: “do I actually need this to write the reply?” If not, don’t hand it over. That alone removes most of the risk. For the basics of handling personal data, the U.S. Federal Trade Commission’s guide on protecting personal information is solid primary-source reading worth a look.

FAQ

Q. Won’t the AI send a wrong date or price on its own? A. The AI never sends. It only drafts. Make dates and amounts placeholders and have a human swap them in before sending, and that accident becomes structurally impossible.

Q. Won’t the writing sound robotic? A. Specify “warm tone” and “family-friendly” in the prompt, and hand it two or three examples in your studio’s own voice, and it gets much more natural. Hand-edit just the first few messages, then use those edits as the next round of examples to settle it in quickly.

Q. Can I use this if I’m not good with computers? A. If you’re only pasting in a prompt, no special knowledge is needed. Setting up the check script is the one slightly higher hurdle, so start with the Claude Code getting-started guide to get your environment ready, or ask someone handy for help.

Q. Does this work for Instagram DMs too? A. Yes. Paste the DM text and draft from it the same way. DMs suit a more casual tone, so adding “keep it friendly, one or two emoji” to the prompt makes it feel natural.

What I confirmed when I actually tried it

With that studio’s help, I ran a full week (about 110 messages) of booking and delivery emails through this prompt and check script.

I wanted to verify two things: draft quality and accident prevention. With just a “warm tone” instruction and two past messages as examples, the drafts hit a level that needed almost no editing. As the front-desk assistant put it: “Not having to think up subject lines anymore is, weirdly, the best part.”

The check script flagged and stopped 4 of the 110 messages, catching leftover old-year text and placeholders. Human eyes alone would probably have let two of those slide. The kind of groveling correction email from the opening fall-season story drops structurally. That was the thing I most wanted to verify this time, and it genuinely worked.

If you want to fix the whole front-desk flow as a company, or talk through training and template setup, training and consulting can design it around your specific operations.