Skip to content
TriedPrompts

Preview flower arrangement options before cutting a stem with ChatGPT Images

Posted
by Ashley Greer (The Zebra)
Last checked
Method:
How we verify

Evidence: Article or interviewA newsletter, journalist or interview account of what happened.

Outcome

Florist Ashley Greer of Atelier Ashley Flowers tested ChatGPT Images 2.0 with the flowers for a favorite client's arrangements. The first prompt returned three designs in brass footed vases after a little over a minute. Greer then showed it the vases and bases already in use, asked for three versions of each (round and tight, round and bouncy/wild, wild and asymmetrical), and told it to remove the dahlias and phalaenopsis orchids Greer had decided against. Greer called the images 'polished provocations, several possible outcomes made visible before I cut a single stem'. Asked to identify the flowers without a materials list, it did 'a fairly decent job', given some obscure varieties.

AI tool

ChatGPT

Prompt Improved prompt

Our fuller version of the source's prompt, or one written to avoid a problem in the source. Not tested by the poster. It doesn't change what happened in the source.

What the prompt labels mean

Before you start, ask me 1–3 short questions about the details below and wait for my answers. Once I've answered, confirm the plan in one line and run the whole task start to finish.

Goal: Show me design options for an arrangement using my actual flowers and containers, so I can choose a direction before I start.

Questions to ask me first:
- What's the occasion, client or style brief?
- Which containers am I choosing between, and roughly what size should it be?
- Are any flowers off the table, or must-haves?

Inputs (fill in or confirm):
- Photo of my flowers: [FLOWERS PHOTO]
- Photos of vases or bases: [CONTAINER PHOTOS]
- Styles to try: [STYLES, e.g. round and tight, loose and wild, asymmetrical]
- Flowers to leave out: [EXCLUDE]

Rules:
- First list the flowers you can identify, with a confidence note for each, and ask me to correct it.
- Use only the flowers and containers in my photos; leave out anything I exclude.
- Make three versions per container, one per style, and label them.
- Keep the scale realistic for the container size I give.
- Treat these as mock-ups for discussion, not finished designs.

Checkpoint: Show me the flower list and one container's three versions before doing the rest.

If you get stuck: If you can't tell what a flower is, say so and ask rather than substituting something else.

Verify before you finish: Check each image against my exclusions and container choices before showing it.

Output: Labeled images by container and style, plus the corrected flower list.

Done when: I can pick a direction to build, or share options with my client.

Try this in:

This prompt has blanks to fill in, like [FLOWERS PHOTO]. We'll copy it instead of filling it in, so you can edit it before you send.

Opens in a new tab. We don't see what you send.

These links are only a shortcut. We're not affiliated with OpenAI, Google or xAI, and your results may not match the source.

The prompt they actually used

The exact prompt from the source.

Based on these flowers, make me three different designs.

Source

https://thezebra.org/2026/09/09/can-chatgpt-arrange-flowers-florist-tests-ai/

  • Turn key points into parent, student and Classroom messages with GeminiRecreated promptEvidence: Company case studyBeginner-friendly

    Eric V., a high school English teacher in New York City, gives Gemini the key points of a message and gets three tailored versions: one for parents, one for students and one for Google Classroom. Eric V. says that saves around 80% of the usual time on emails, and closer to 20% on unit and lesson planning, where the request might name the book and ask for three lessons on historical context plus a field trip idea. Gemini also helps make assignments clearer and more concise for the 11th graders.

    Gemini · Content · Source date:

  • Alexander Machulin's spouse edited a full talking-head reel with Claude. From 10 minutes of raw footage, Claude cut the video in CapCut, added subtitles and picked photos from the spouse's gallery, and nothing had to be fixed by hand. Machulin only helped with the setup: Whisper had started downloading to the laptop, and the regular chat had been opened instead of Claude Code. Machulin says the video came out good.

    Claude · Content · Source date:

  • Find repurposable clips in years of YouTube videos with Claude CoworkImproved promptEvidence: Screenshots or logsPartial

    Given brand context, timestamped transcripts, example clips and performance data, Claude Cowork analyzed the content well and surfaced topics and hooks, saving hours of watching; but the clips it cut started mid-thought or lacked context or conclusions, even after several re-prompts.

    Claude Cowork · Content · Source date: