Mume
Run of Show made easy

Overview
Designing a photoshoot planning tool with AI-assisted building — one sentence in, a pre-production packet out.
Mume is a web tool that turns a one-sentence photoshoot brief into a timed, drag-and-drop run of show, complete with crew assignments, a mood board, a Letter of Request (pull letter), and exportable PDFs. It's for photographers, stylists, and small creative teams who currently stitch together call sheets across spreadsheets, docs, and group chats.
The entire product was designed and built conversationally in Lovable — every screen, interaction, and document export iterated in a live preview. This case study documents the problem, the decisions, and the final product.
Problem
Every shoot answers the same questions from scratch
When does glam start, who changes when, how do you stagger two talents without dead time, how does everyone get the same plan? The incumbent tools fail small teams in specific ways:
StudioBinder is built for film crews — setup-heavy forms, per-seat pricing, and photo shoots feel like an afterthought.
Spreadsheets are flexible but fragile: no live re-timing, no crew view, ugly to send to a client.
Generic project tools don't understand shoot anatomy — style edits, glam, changes, looks, buffers, wrap.
Insight
Photographers already say it in one sentence
"I'm doing a photoshoot with two models, three looks, edgy streetwear, call at 9am." If a tool can read that sentence, confirm what it understood, and draft a schedule, the heavy part of planning is done. Everything after that is editing — fast, familiar work.
Brief to schedule in under a minute — natural language in, editable plan out.
Confidence before commitment — always show an editable interpretation before generating anything.
The calendar is the product — drag, resize, tap-to-edit, not a form with a preview.
Leave with paperwork — the plan isn't real until it's a PDF someone can hold on set.
Feel like a creative tool, not enterprise software — editorial design, calm neutral palette, one confident accent.
Journey
Five stages that every photographer goes through
Before Mume, the same five stages just added friction at every step: the brief was scattered across texts and DMs, with nobody fully sure which version everyone had actually agreed to; the schedule got built by hand in a spreadsheet, with glam and change buffers computed from memory, so one blank cell could break the whole day; the pull letter started from a blank doc every time, retyped per talent, per shoot; the printed sheet stopped being true the moment glam ran long, since nothing on set re-timed itself; and handoff meant screenshotting, texting, and re-typing for the crew until everyone was holding a slightly different version.

Anatomy
Designing for the real shoot in mind
The generator encodes how shoots actually run: crew call and load-in, style edit on the rack (20 min), glam (60–75 min), 15-minute changes, 20-minute look shoots, an optional lunch, buffers between looks, and wrap/load-out.

Constrains
When the read is wrong
The confirmation step catches most misreads, but not all of them — and the clearest example is one I hit demoing the tool on a real Timid brief. I wrote: "call time is 9:00 AM… wrap by 4 PM." Mume built a schedule that actually ran to 8:48 PM. It parsed the talent count, the looks, the call time — all correct — but the generator builds forward from shoot-anatomy defaults (glam, changes, buffers per look) rather than weighting them against a stated wrap time. The result reads as a normal, well-formed schedule; nothing about it looks wrong until you check the end time against what was asked for. Editing a wrong plan is still faster than building one from nothing, even when the first read is off, but that's when confirmation comes to play to verify shoot time.

Responsive
Shipped and in daily use
Mume went from an empty project to a working, installable planning tool with document generation — built entirely through iterative conversation.
Natural-language brief parsing with an animated, self-demonstrating input and PDF/image attachment
An editable details step — shoot type, up to 8 named talents, fashion looks, date & time, searchable location with access notes
A drag-and-drop run-of-show calendar with typed colored blocks, multi-track spanning, crew assignment, and instant re-timing
A Letter of Request generator with per-talent tabs, sizing charts, and a scaled US Letter preview
Export as combined or separate PDFs (crew sheet, mood board, run of show, letters), CSV for spreadsheets, and read-only share links
Local project caching so a session survives leaving and coming back

Learnings
Iterating through prompts
Building in Lovable changed the process itself. Instead of speccing a screen and waiting, I could say "make the sizing labels hug their boxes and align into columns" and see it live in seconds. That speed shifted my role upward: less time producing artifacts, more time judging them.
Iteration loops collapsed from days to minutes — spacing, copy, and layout decisions were tested immediately.
Risky ideas got cheap: the calendar span feature went from question to working interaction in one prompt.
The discipline moved to taste. When anything is buildable, the job is deciding what deserves to exist — I cut a mood/tone field, a gallery page, and a crew tab when none of them earned their place.
If I had more time:
Real accounts and cloud sync, replacing the local-only cache
Crew-facing live links with read receipts and per-person call times
Mood-board intelligence — generated or curated references when a brief names a direction like "edgy, colorful streetwear"
Type-driven schedule templates, so "commercial" and "editorial" shoots generate genuinely different formats