AI Moves Into Workflows
Business of Fashion reports AI is shifting from novelty experiments into an operational layer inside marketing and campaign workflows, used for approvals, asset adaptation, planning and copy generation. The case study highlights governance questions — who defines prompts, red lines and approval chains — as teams integrate AI into daily production. (businessoffashion.com)
Generative artificial intelligence is moving from brainstorm sessions into the production line of marketing teams, where it now helps draft copy, resize assets and route work through approvals. (businessoffashion.com) Business of Fashion reported in April 2026 that marketers are using artificial intelligence as a “structural part of daily operations,” not just for one-off tests, inside campaign planning and content workflows. Its case study focused on fashion brands trying to speed production without losing brand control or human review. (businessoffashion.com) In practice, that means software is being used for repetitive jobs that used to sit between creative and media teams: generating first-draft text, adapting one campaign into many formats, and preparing assets for sign-off. Adobe said in March 2025 that Firefly Services and Custom Models were built to help businesses produce on-brand content across social media, e-commerce and mobile at scale. (news.adobe.com) The workflow shift is showing up in the vendor stack around marketers. Salesforce’s Tenth Edition State of Marketing report says it surveyed nearly 4,500 marketing leaders worldwide on artificial intelligence, data and personalization as brands adjust to what it calls “agentic marketing.” (salesforce.com) The operational question is no longer whether a team can generate an image or a caption in seconds. It is who sets the prompt rules, which claims or visuals are off-limits, and which manager, lawyer or brand lead has to approve the output before it goes live. (businessoffashion.com) That governance layer is becoming part of the product itself. Adobe said Firefly Custom Models can be trained on a company’s own assets for brand consistency, while its Creative Production tools are designed to handle repetitive production tasks through a no-code interface. (news.adobe.com) The same pattern is appearing outside design software. OpenAI’s Agent Builder, documented in 2026, is a visual system for assembling multi-step workflows with typed inputs, outputs and live previews, which is the kind of structure companies use when they want artificial intelligence to follow a process instead of answer a single prompt. (developers.openai.com) Fashion is a visible test case because campaign work produces huge volumes of near-identical variations across regions, channels and devices. Business of Fashion’s wider coverage has also tracked companies such as Pixel Moda, which said it supports more than 900 brands and produces 14 million pictures and videos a year with artificial intelligence-assisted and human-led processes. (businessoffashion.com) Marketers are also being pushed by volume. Salesforce said its latest report centers on artificial intelligence, data and personalization, a combination that forces teams to make more versions of the same message while keeping legal, brand and customer-data controls intact. (salesforce.com) So the new bottleneck is shifting from making the first draft to deciding the rules of use. As artificial intelligence settles into approvals, adaptation and planning, the teams that move fastest are likely to be the ones that can define those rules before the next campaign starts. (businessoffashion.com)
