PepsiCo AI Marketing Review Forces Agencies to Prove Platform Power

PepsiCo is running a global review focused on AI marketing transformation, Ad Age reported on August 17, 2026, drawing high-profile agency groups and consultancies into a contest that centers on systems rather than a conventional creative or media pitch. The food and beverage company wants partners who can embed generative and agentic AI across content, data, personalization, automation and broader marketing workflows for brands that include Pepsi, Lay’s, Doritos, Gatorade and Quaker.

The assignment arrives after years of internal AI build-out. It tests who can operate the infrastructure PepsiCo already owns, not merely supply campaigns.

PepsiCo Opens a Global AI Marketing Systems Review

The process is not a standard agency of record shoot-out. Reports describe a mix of holding-company networks and consulting firms competing to reshape how marketing functions run once AI handles more production, analysis and optimization work that once sat across multiple partners.

Possible scope includes content creation pipelines, audience analytics, personalization engines, campaign automation and the operating model that ties them together at enterprise scale. PepsiCo has already used data and AI for targeting, creative development and media effectiveness. This review examines how those pieces should fit as one system and what external partners still add once more work becomes AI-enabled.

• Content generation and versioning at brand-safe volume
• Data unification and real-time analytics for 200-plus markets
• Personalization and loyalty triggers tied to retail execution
• Agentic workflows that handle routine optimization and escalate exceptions
• Governance that keeps outputs inside Responsible AI rules

Global advertisers are reassessing traditional relationships for the same reason. Generative tools automate pieces of creative production. Agentic systems start to manage parts of media and measurement. The pitch therefore favors operators of platforms over pure idea shops.

The Company Already Runs Agents and a GenAI Platform

PepsiCo is not starting from zero. Public descriptions of its data and AI program put more than 60 petabytes under management, doubling yearly, with a push toward a single governed lake and golden KPIs across roughly 320,000 employees. The company serves products to more than 1.4 billion consumers daily and reports nearly $92 billion in revenue.

Its PepGenX GenAI platform acts as an abstraction layer so business teams call services while platform teams swap models without rewriting every app. More than 1,500 bots, assistants and agents are already deployed. A proprietary agentic layer can expand 100 million consumer records into synthetic populations for digital twins used in R&D and marketing. PepIris lets sales reps scan shelves across 4,500 SKUs and 800,000 stores to fix placement on the spot.

Key PepsiCo AI operating figures

• 60+ petabytes of data under active management
• 1,500+ bots, assistants and agents in production
• 165% sales uplift on a Gatorade GenAI custom-bottle campaign, plus 200% loyalty lift and 150,000-plus unique designs
• 10 million+ shelf images processed globally for compliance and planograms

Magesh Bagavathi, PepsiCo’s SVP and global head of data, analytics and AI, has described the requirement as vision, top-down mandate and the courage to say no so teams focus on value. Governance boards review GenAI use cases against a formal Responsible AI framework before launch. That internal muscle shapes what any external partner can actually own.

Where the Brief Likely Lands

The review sits on top of live technology relationships. PepsiCo uses the Salesforce platform extensively for retail execution, service, field service and marketing. Data 360 unifies retailer, fleet and household signals. A multi-year roadmap with PwC includes next-generation marketing analytics and personalization alongside trade promotion management and equipment service.

Athina Kanioura, CEO of Latin America and global chief strategy and transformation officer, said embracing an AI-first world means humans and intelligent agents collaborate rather than merely coexist. Ramon Laguarta, chairman and CEO, called the Salesforce work another step toward smarter decision-making and sustainable growth. Those statements frame the pitch: partners must plug into existing agentic plans, not replace them.

PepsiCo’s U.S. beverages team has also deepened an always-on social model with VaynerMedia that pairs internal brand strategy with fast cultural content. In-house capability is already part of the operating system. Any winner of the AI transformation work will sit alongside that model, not erase it.

Holding Companies and Consultancies Face Different Tests

Traditional networks bring creative scale, media buying muscle and brand craft. Consultancies bring enterprise architecture, change management and multi-year platform roadmaps. The brief favors the second skill set more than a classic brand campaign does.

Industry tracking shows the largest holding companies have committed hundreds of millions to AI development. WPP has pursued content engines and internal platforms. Publicis has long-running AI tools and system architecture claims. Omnicom’s combination with IPG put enterprise generative AI at the center of its value proposition. Scale helps amortize the cost of building those layers. Smaller pure-play creative shops face a steeper climb.

Consultancies already appear in PepsiCo’s public AI narrative. IBM Consulting helped map GenAI use cases to standard patterns inside the North Star architecture. PwC co-designed the five-year Salesforce roadmap that explicitly includes marketing analytics. Those relationships give them running starts.

Earlier PepsiCo AI Work Set the Bar

The company has public case studies that any bidder must clear. A 2021 overview detailed AI for perfecting Cheetos shape and Cheetle coverage on the line, predicting plant failures so mechanics shifted to planned work, and shortening innovation cycles for snacks and beverages after social and menu signal analysis.

1. 2021 and prior: AI for crop data with growers, shelf and route optimization concepts, Messi face-mapping for millions of personalized Lay’s messages from minutes of footage.
2. Mid-decade platform build: PepGenX, PepVigil monitoring, Responsible AI board, data lake consolidation, persona-first tools for 120,000-plus frontline staff.
3. 2025 Salesforce Agentforce path: plans for contact centers first, then roughly 20 further use cases including personalized recommendations and AI-generated planograms; marketing analytics and personalization on the PwC roadmap.
4. 2025-2026 social model: deeper VaynerMedia integration for always-on cultural content with the in-house team.
5. August 2026: global AI marketing transformation review opened to agencies and consultancies.

Those milestones show marketing is one more domain being pulled onto the same governed platform that already serves supply chain, sales and R&D. A partner that only delivers ads will not match the integration bar.

Production Costs Keep Falling Outside the Pitch

While the formal review runs, cheaper generative tools continue to undercut the cost of certain creative production. Public examples and industry chatter show brands generating product video and social variants for fractions of traditional agency fees. One widely shared case claimed a Pepsi-style product-to-video workflow that collapsed seven-figure campaign economics into low four figures. Similar threads tracked near-copies of high-profile AI ads made with consumer-grade stacks.

That pressure does not eliminate the need for brand strategy, legal review or cultural judgment. It does change the volume economics that once justified large retained creative teams. Agencies that still price primarily on human hours face margin compression even if they win pieces of the PepsiCo work. The second-order effect is clear: the pitch rewards firms that treat AI as production infrastructure and measurement system, not a novelty filter.

Embracing an AI-first world means reimagining an enterprise where humans and intelligent agents don’t just coexist, they collaborate.

Athina Kanioura made that point in the context of the company’s agentic plans. The same logic now applies to external marketing partners.

The Partners Who Already Sit Inside the Stack

Any outcome will be shaped by relationships that predate the review. Salesforce and PwC already hold multi-year roadmaps that touch marketing analytics and personalization through Agentforce for marketing analytics and personalization. IBM patterns sit inside the GenAI architecture. The PepsiCo digital value transformation work with PwC supplies a recent template for measurable outcomes across functions.

Historical context lives in the company’s own archive of early PepsiCo AI applications in snacks and sales. Broader industry capital flows and platform bets are tracked in analyses of holding company AI capital commitments and platforms.

Enterprise AI still hits physical limits. Capacity and energy questions that affect every large deployment, including those at PepsiCo scale, appear in reporting on power constraints that still brake enterprise AI. Governance failures remain a live risk for any GenAI marketing system; recent cases of uncontrolled agent behavior underscore the governance risks that sit beside every GenAI rollout.

The review will not invent PepsiCo’s AI stack. It will decide which external parties get to extend it into marketing operations, on what commercial terms, and with how much ownership of the resulting data and agent layers. Traditional agencies that treat the pitch as another creative bake-off will miss the infrastructure test. Consultancies that cannot still make brands culturally sharp will face the opposite gap. The companies already inside the data and agent fabric hold the clearest path to influence the final shape.