Use case · AI content creation

AI packaging concepts and mockups, with the label left to people.

AI is well suited to the early and late ends of packaging work: exploring many concept directions quickly and producing realistic mockups for testing, sell-in and pre-launch imagery. It is not suited to the mandatory label. Ingredients, allergens, net quantity and legal text must be set from verified data and checked by a person, and dielines, barcodes and print separations remain specialist work.

Where generation helps: range and mockups

Concept exploration benefits from volume. A team can review dozens of directions for colour, typography, illustration and pack architecture in the time it once took to brief three, then take the strongest into proper design. Mockups are the second use: a new variant shown on shelf, in a hand, in an online listing or in a retailer presentation before a single pack is printed. Keep mockups honest in consumer research by telling respondents they are concepts, and check each direction for accidental resemblance to competitors' packs and trade marks before it goes further; a generated design can echo a well-known pack without anyone intending it.

Mandatory text is not a design element

In the EU, Regulation (EU) No 1169/2011 on food information to consumers sets what a food label must say: the name of the food, the ingredients list with the 14 regulated allergens emphasised, net quantity, date marking, storage conditions, the business name and address, and a nutrition declaration, among others. Mandatory text must meet a minimum x-height of 1.2 mm, or 0.9 mm on packs whose largest surface is under 80 cm², and the name, net quantity and any alcohol strength must sit in the same field of vision. Image models misspell and invent text. Set every mandatory word from the specification database and have a regulatory specialist sign it off.

Dielines, barcodes and print specs stay specialist

A dieline is an engineering drawing: cut and crease lines, glue flaps, bleed and the areas that fold out of sight. It comes from the converter or packaging engineer and is checked against physical samples. Barcodes must be generated from the product's registered GS1 number by barcode software, at a size and with the quiet zone the standard requires; a barcode drawn by an image model will not scan. Print files carry spot colours, varnish, foil and embossing on separate layers for the printer. Generated artwork can feed this stage as approved design direction, but the production files are built by designers who know the press.

Recycling marks and market versions

Sorting and recycling labels vary by market, and the rules are moving. France already requires the Triman logo with sorting information on most household packaging, and the EU Packaging and Packaging Waste Regulation (EU) 2025/40, which applies from 12 August 2026, will require harmonised sorting labels from August 2028 at the earliest. Environmental claims on pack are also caught by the EU's ban on generic green claims, which applies from 27 September 2026. When the studio produces mockups for several markets, product-lock keeps the real pack pixel-accurate while scenes change, and approved claims and banned words from the brand hub apply to the marketing copy around it; the regulated label still gets a human check per market.

Updated 25 September 2026

Questions

AI packaging concepts and mockups, answered.

Can AI design our final packaging artwork?

It can produce the design direction and realistic mockups. The final artwork needs designers working on the converter's dieline, with mandatory text set from verified data, barcodes generated properly, and colour separations prepared for the press. Treat generated designs as approved direction, then build production files the normal way.

Are AI mockups good enough for retailer presentations?

Usually, yes, as long as they show the pack as it will be produced. Use the final approved artwork where it exists, keep proportions and materials realistic, and say when a mockup is a concept. Retail buyers make ranging decisions on these images, so a pack that looks better than the real one causes problems later.

Who checks allergen information on AI-generated packs?

The same regulatory or technical specialist who checks it on any pack. Allergen and ingredient text should never originate in an image model; it comes from the product specification and is proofread on the final artwork. An error here is a food-safety issue that can mean a recall, not a design fix.

Get started

See your AI studio generating this week.

Thirty minutes. A live studio in your colours, your brand hub loaded, a real campaign brief.

See plans