Inside the Agentic AI Playbooks of Bayer, Zoom and a Health Tech Firm

How Bayer, Zoom and a Health Tech Firm Are Deploying AI in Marketing

Marketing teams are moving agentic AI from pilots into day-to-day workflows at very different speeds, and three of the most visible examples were laid out in Adweek's Agentic Marketers series. Bayer CMO Sam Avivi, Zoom CMO Kimberly Storin, and the marketing SVP of an unnamed health technology company described what their teams have shipped over the past month and what the next quarter holds.

Bayer is focused on generative engine optimization (GEO), making six of its U.S. over-the-counter brands show up in AI answer engines: MiraLAX, One A Day, Afrin, Aleve, Bayer Aspirin and Claritin. Avivi says the approach rests on three pillars — visibility, so brand content ranks near the top of AI answers; reliability, where the company finds that credible earned media influences AI answers more than user-generated content; and shoppability, backed by a partnership in which Amazon selected Bayer as its exclusive OTC partner for building AI data and infrastructure, including the symptom-to-product mappings that shape purchase decisions. Bayer plans full U.S. scale in Q4 and a global rollout in key markets starting in Q1.

Zoom, meanwhile, built an internal AI agent that automates product naming for the 2,500-plus features the company launched last year. Storin said the team had to create a rubric, prompts and brand-guideline checks before automation could take over, and the agent was deliberately built inside Zoom's own ecosystem rather than on an external tool such as Claude — both to free up staff time and to demonstrate what Zoom's federated AI can do. That federated model routes each task to the most cost-effective model for the job. Next up: a dedicated AI engineer sitting inside the marketing team and a group focused on agents full-time.

The health tech SVP described a different playbook: using Profound to raise the company's discoverability on large language model platforms, and layering Writer on top as an agentic platform that turns trends and insights into content briefs. The executive, whose team started with Profound in 2025, said the company is also in conversations with OpenAI as it rolls out its ad platform, and next wants to automate the analysis of social comments and direct messages, evaluating Nectar Social for that work.

Why AI Answer Engines, Federated Models and OpenAI Talks Matter for Marketers

For Bayer, AI Answer Engines Are a New Retail Shelf — With Amazon as the Aisle

The three-part framework Avivi describes — visible, reliable, shoppable — effectively treats AI answers as a discovery channel in its own right, alongside search and social. The most consequential detail is the Amazon partnership. Being selected as Amazon's exclusive OTC partner for building AI data and infrastructure gives Bayer unusual influence over how symptoms get mapped to products, which is precisely where OTC purchase decisions start. But it also concentrates a chunk of that emerging channel in one retailer, and the measurement to prove ROI is not finished: Avivi concedes the infrastructure to measure all three areas continuously still has to be built.

The reliability finding deserves attention. If credible earned media shapes AI answers more than user-generated content, GEO rewards editorial authority rather than raw engagement — a shift that puts PR and earned coverage back at the center of performance marketing.

Zoom's Naming Agent Is as Much Product Demo as Cost Saving

Zoom's product-naming agent is a workflow win on the surface: 2,500-plus features, one rubric, no more manual naming. But the design choices reveal a second agenda. Building the agent inside the Zoom ecosystem rather than on an external model turns the marketing team into a live demonstration of Zoom AI's federated architecture, which selects the best and cheapest model for each task. Hiring a dedicated AI engineer for marketing pushes the same logic further — the marketing organization becomes both a user and a reference case for agentic workflows, which is useful when your company sells AI products to the very teams it is trying to impress.

The Health Tech Stack: Profound for Discovery, Writer for Briefs, OpenAI for Early Access

The health tech executive's setup is the most readily replicable. Profound surfaces where demand exists on LLM platforms — the example given is a parent asking about the top five solutions for a child's airway problem — and Writer converts those insights into briefs, so content production compounds inside a knowledge base that improves with each cycle. The OpenAI conversations are the forward-looking piece: as OpenAI builds out its ad platform, marketers who engage early can shape how the offering works and where their brands appear in knowledge-base answers, an echo of the early-adopter advantage brands once had with Google ads.

The Adoption Curve Is Real, but Uneven

Avivi's caveat — that methods working for a CPG company won't automatically work for an OTC manufacturer — is the clearest warning against copying any of these playbooks wholesale. Bayer is optimizing regulated health information, Zoom is demonstrating a product, and the health tech firm is rebuilding a content engine. The common thread is that AI answer engines are becoming measurable marketing real estate, and the teams defining placement rules now will set the benchmarks everyone else measures against.

What Marketing Teams Can Copy From Bayer, Zoom and the Health Tech SVP

• Adopt a named framework before you optimize: Bayer uses visibility, reliability and shoppability as its three GEO dimensions; Avivi says the measurement infrastructure is still being built, so define the metrics before scaling. Bayer targets full U.S. scale in Q4 and key global markets from Q1.
• Treat earned media as AI-discoverability collateral: Bayer reports that credible earned coverage influences AI answer engines more than user-generated content, so PR placements should be briefed with AI answers in mind, not just journalists.
• Write the rubric before the automation: Zoom's naming agent worked only after the team defined prompts, a scoring rubric and brand-guideline checks; for high-volume workflows such as its 2,500-plus feature backlog, the automation pays for the setup.
• Route tasks to the cheapest adequate model: Zoom's federated AI model avoids paying premium-model prices for routine jobs; audit which of your workflows are overspending on expensive models.
• In regulated categories, make accuracy the trust gate: Bayer's symptom-to-product mapping is built alongside Amazon, but OTC and health brands should independently audit what AI engines say about their products, since wrong medical answers carry liability as well as reputational damage.
• Get a seat at the OpenAI table early: the health tech SVP is already in conversations as OpenAI's ad platform rolls out, and early participants can influence how the ad offering and knowledge base function.