Caption Lab

Caption Lab is Fuze's writing workshop, designed for social posts. It combines three modes around one idea: your captions should sound like your studio, not like a robot.

The three modes

Style fingerprint

Analyzes your past captions to extract your style: tone, length, structure, emoji and hashtag usage. Provide at least 3 captions of about twenty characters minimum each (pasted or via screenshots); you can manually correct the detected profile. Your analysis history is kept.

Generation

Writes captions from your style fingerprint and your brand (words to use or avoid, luxury level…). The topic is optional: simply send your photos and Fuze writes from what is genuinely visible in them. You can attach up to 3 photos per generation; they are analyzed on the fly and never stored.

Audit

Paste an existing caption: Fuze grades it, points out what weakens it and proposes a rewritten version.

Generate from your photos alone

The "What is this post about" field is optional: it becomes simple additional context. Without a topic, Fuze analyzes the photo (visible subject, mood, season, session type) and writes the full caption — hook, body, call to action, hashtags — following your style fingerprint. It never invents a price, a promotion or a testimonial: only what is visible and what you declared in your studio is used.

The caption template (advanced mode)

In the generation settings, you can define your own template: free text mixing fixed text (reused verbatim in every caption) and variables filled in by Fuze. Clickable chips insert the variables without typing them; unknown variables are flagged before generation.

{{legende}}

The caption body written by Fuze (photo analysis, optional context, style fingerprint).

{{hashtags}}

The local, layered hashtag block generated for the post.

{{CTA}}

The call to action written by Fuze.

{{site internet}}

Replaced by your website address, as declared in your studio profile.

The template is saved in your fingerprint and applies to all generations (single and batch) while it is active; a switch lets you disable it temporarily without losing it. Your style still applies to the written parts: tone, target length, emojis, banned words.

Batch mode (several posts at once)

1. Create 2 to 5 series: each series is one post, with 1 to 3 photos and optional context.
2. Run "Generate all": series are processed one after another, with a visible status for each (waiting, running, done, error). The estimated total cost is shown before launch.
3. Each completed series produces a full caption ready to post, with its thumbnail, an individual copy button and a "copy all" button. A failed series does not block the others and costs nothing; you can regenerate it on its own.

Getting the most out of it

• Analyze your fingerprint first: generating without a style profile produces more generic captions.
• Paste your real captions, not idealized examples: the fingerprint must reflect your actual writing.
• Without a topic, pick sharp, representative photos: the caption is based on what is genuinely visible.
• An audit before publishing is the fastest way to improve a text that feels "almost right".