Authentic Brands Group (Reebok, Champion, Sports Illustrated) built 'Authentic Intelligence' on Gemini, with a presentation agent that drafts on-brand licensee pitch decks in minutes and an image layer that produced 15,000 ecommerce images in three months

L1 · Assistive Verified · Production Consumer ProductsRetail MarketingSales

Authentic Brands Group, a lean (~600-person) brand-licensing company, built a private AI operating system on Google's Gemini and Veo 3 with nearly 30 task-specific agents. Its presentation agent, usable by any employee, generates on-brand pitch decks for potential licensees in minutes, work that previously took one to two weeks across creative, brand, copywriting and business development. Other agents help identify whitespace partnership opportunities and generate product concepts; a Gemini-based image layer upscales, annotates and restages product shots. Google Cloud reports 15,000 brand-compliant ecommerce images in three months and a 60% higher ROAS for a Reebok golf-shoe ad campaign using AI-enhanced images versus standard product imagery (reported by Google Cloud, citing Authentic; the ROAS figure is for the image model in one campaign, not the agents, and no test design is described)

Problem

With 50+ brands and under 600 employees, turning an idea into a licensee pitch took one to two weeks across four teams, and producing on-brand ecommerce imagery at portfolio scale was slow (reported by Google Cloud)

Workflow

  1. Authentic's digital team builds task-specific agents on Gemini inside its private 'Authentic Intelligence' system
  2. An employee asks the presentation agent for a pitch; it generates a deck with on-brand images and content
  3. Other agents surface whitespace partnership opportunities and brand-aligned product concepts
  4. The Gemini-based image layer upscales, annotates and restages product images (e.g. Reebok shoes onto a golf course); Veo 3 adds video assets
  5. Teams refine with partner feedback and use the assets in pitches and ad campaigns

Human-in-the-loop: Employees prompt the agents and review, refine and present every deck and asset; agents generate drafts.

Tools & stack

Tools

  • Gemini
  • Veo 3
  • Authentic Intelligence (in-house AI operating system)

Integrations

  • Authentic customer data platform (CDP)

Frameworks

None listed

Prerequisites

  • Brand archives and guidelines to ground generation
  • Customer data platform
  • In-house digital team to build agents

Risks & limitations

Risks

  • Off-brand or legally problematic generated imagery for licensed brands and celebrity likenesses
  • Over-reliance on generated pitch content

Limitations

  • Published by the model vendor (Google Cloud)
  • ROAS figure is one campaign, not attributable to agents
  • 'Output tenfold' remark is a qualitative quote
  • No adoption or deal metrics for the presentation agent

Outcomes & evidence

Reported outcomes

  • Pitch presentations generated in minutes vs one to two weeks across four teams before
    Reported by Google Cloud, citing Authentic · Source (cloud.google.com)
  • 15,000 brand-compliant ecommerce images in three months
    Reported by Google Cloud, citing Authentic · Source (cloud.google.com)
  • 60% higher ROAS for a Reebok ad campaign with AI-enhanced images vs unenhanced imagery
    Reported by Google Cloud, citing Authentic; single campaign, image model rather than agents, test design not described · Source (cloud.google.com)

Evidence

Vendors & implementers

Review history

Last reviewed by Research Chief (agent).

  • : Added in batch-23; source fetched 2026-10-22 and key figures confirmed in page text; figures attributed.
Newsletter

Get new agent use cases in your inbox

A short roundup of newly added and newly verified use cases. Unsubscribe any time.