For two decades, values-based marketing ran on a simple mechanic. A brand declared what it stood for, a person felt something about that declaration, and the feeling nudged the sale. AI agents break the mechanic. When a shopping agent researches a purchase on a customer’s behalf, it does not feel anything. It reads. It compares structured facts, enforces the preferences its user stated, and surfaces only what survives comparison.

Values still matter in that pipeline, arguably more than ever. But they now have to be written in a form an agent can parse, check, and pass along. Brand trust is becoming something a company compiles, not just something it claims.

Your Next First Impression Is Not Human

The Nielsen Norman Group put it plainly in its 2026 research on agents as users: “user” is no longer synonymous with human. Agents arrive with goals, encounter interfaces, and either succeed or fail through them, much like people do. The difference is how they perceive. An agent may interpret screenshots (slow, expensive, error-prone), parse a page’s accessibility tree (cheaper and more reliable), or skip the interface entirely and call a structured API.

That has an uncomfortable implication for values-driven brands. A sustainability story told through a cinematic brand film registers with humans. To an agent reading the accessibility tree, the film is a black box: heavy media, little parseable text, no product-level facts. Semantic HTML, descriptive labels, and clean structure are what make a brand legible to its new audience. Accessibility work, long justified as simply doing the right thing, now has a hard commercial case: it is the shelf from which agents shop.

The Ranking Factors of Conscience

Consider how discovery already works inside agentic commerce. When OpenAI launched Instant Checkout in ChatGPT, built on the Agentic Commerce Protocol with Stripe, it made the ranking rules explicit: product results are organic and unsponsored, ranked purely on relevance, with the agent weighing availability, price, quality, whether the merchant is the primary seller, and whether checkout is enabled. No media budget moves those levers.

Values enter this path the same way price does: as criteria the agent applies literally. A customer who tells their agent “only repairable electronics from companies that publish supply-chain audits” has converted a value into a filter. A brand whose repairability and audit history live in structured, verifiable product data passes the filter. A brand whose values live only in a manifesto does not underperform; it simply never appears.

Verifiable Beats Virtuous

Agents did not invent skepticism about values claims. A 2020 European Commission study found that more than half of examined environmental claims were vague, misleading, or unfounded, and roughly 40 percent had no substantiation at all. Regulators from the FTC to the UK’s Advertising Standards Authority have spent years policing the space. Consumers were trained by that history, and the agents serving them inherit the training: a comparing system treats an unverifiable claim not as a positive but as noise, or as risk.

The practical consequence is that third-party verification travels further than prose. A B Corp certification, a published audit, a material disclosure tied to a product identifier, a warranty a support bot will actually honor: these are the artifacts an agent can cite when it explains a recommendation to its user. Brand trust becomes an input alongside availability and price, and inputs have to come from somewhere checkable.

Keep the Human Convinced, Too

None of this retires the human audience. OpenAI’s own launch notes stress that users confirm each step before anything is bought, and that payment tokens are scoped to a specific amount and merchant. The moment of confirmation is still a human moment: the person reads the agent’s summary, sees the brand name, and decides whether to tap buy. That summary is increasingly drawn from your verifiable facts, but the story around them still has to earn the final click.

So the values layer has to work at two levels. Machine-checkable evidence: structured product data, certifications linked as evidence, consistent policies, clean markup, plain claims tied to sources. And human-credible narrative: the same values, told warmly where people actually look. The two must agree, because agents cross-reference channels, and inconsistency is exactly what a comparing system notices first.

Conclusion

Values-based marketing is not dying in the agent era. It is being held to its word. The brands that thrive will be those whose claims survive translation into data an agent can verify and a story a human can believe at the moment of confirmation. Write your values where both readers can find them.

References

• AI Agents as Users, Nielsen Norman Group: https://www.nngroup.com/articles/ai-agents-as-users/
• Buy It in ChatGPT: Instant Checkout and the Agentic Commerce Protocol, OpenAI: https://openai.com/index/buy-it-in-chatgpt/
• Greenwashing, Wikipedia: https://en.wikipedia.org/wiki/Greenwashing

Research and written by Peter Jonathan Wilcheck