Stanley 1913 is using AI in marketing operations while keeping it away from the advertising consumers actually see. According to Digiday, the brand allows the technology in ideation, early concept work and personalization, but its chief brand officer says consumer-facing creative remains human-made.
That position is more useful than either extreme in the current debate. “Never use AI” ignores real efficiency gains. “Use it everywhere” ignores the fact that different stages carry different levels of brand, legal and cultural risk. A serious policy needs a boundary, not a slogan.
Why Stanley’s line matters
Stanley has invested in internal creative teams and a photo studio, so the boundary is backed by capability rather than nostalgia. The company can automate upstream work without outsourcing the final expression of a brand built on community, creators and authenticity. It is protecting the part of the process where authorship is visible to the customer.
The exact line will not suit every company. A performance retailer producing thousands of simple variants has a different risk profile from a lifestyle brand. The lesson is to define the boundary from brand value and evidence, rather than letting tool availability decide.
One universal AI rule fails
Marketing work includes research, synthesis, planning, production, adaptation, distribution and measurement. Each stage has different failure modes. A model that summarizes interview notes creates a lower public risk than a model that invents the face, voice or product claim in a national campaign.
Policies should therefore classify uses by reversibility, visibility and consequence. Can a human detect the error before release? Will the customer see the output? Could failure damage a person, misrepresent a product, break a contract or weaken a distinctive brand asset? Those questions are more actionable than arguing whether AI is creative.
A four-zone governance matrix
- Green — assist freely: transcription, tagging, format conversion, internal search, first-pass synthesis and administrative work, with normal quality checks.
- Amber — assist with accountable review: ideation, audience variants, draft copy, localization and personalization. A named specialist verifies facts, tone and rights.
- Red — restricted use: final brand ideas, realistic people, sensitive claims, signature visual assets, regulated categories and work whose value depends on human testimony.
- Black — prohibited: fabricated endorsements, unlicensed likenesses, deceptive product representation, hidden manipulation and any use that violates law or contract.
The matrix should live inside the workflow. A policy document that nobody sees during briefing, production and approval will not change behaviour.
Questions for every stage
Before using a model, ask what data enters it, whether the vendor can retain that data and which rights cover the output. During production, record the tool, model and human changes. At approval, require the reviewer to assess factual accuracy, brand distinctiveness, representation, disclosure and product truth. After launch, monitor complaints and unexpected patterns, not only speed and cost.
Regional teams need the same core rules with room for stricter local requirements. Stanley’s distributed creative centres illustrate the governance challenge: efficiency improves only when every team interprets the boundary consistently.
Measure the policy, not just the tool
Track hours saved, revision cycles and cost per usable asset, but pair them with rejection rate, correction rate, legal escalations, brand-consistency scores and performance after human review. A system that generates ten times more drafts but increases rework has not created ten times more value.
Review the boundary quarterly. Some uses will become safer as tools and controls improve; others may prove riskier after customer feedback. The goal is not to freeze a 2026 opinion. It is to make deliberate, documented decisions.
The strategic takeaway
Stanley’s approach is not anti-AI. It treats automation as infrastructure and human authorship as a brand choice. That distinction gives teams permission to gain efficiency without quietly handing over the most visible expression of the company.
A strong AI policy should tell marketers where to experiment, where an accountable reviewer must intervene and where the answer is no. The competitive advantage will not come from using the most AI. It will come from knowing which work should remain recognizably yours.
This image matches the article because it puts human-made consumer creative in front while keeping automation visible but deliberately behind the boundary.
