Where AI Belongs in Brand Work, and Where It Does Not: Lessons From an Award-Winning Campaign

By the ThreeSixtyEight Team

Every agency in 2026 is using AI somewhere. The question is no longer whether to integrate AI into agency operations. The question is where AI adds value, where it subtracts, and how to tell the difference.

We have spent the last two years working through this question on real client engagements, including campaigns that have gone on to win industry recognition. Earlier this year, our "That's Life with Alfa Insurance" campaign won Best of Show in the Interactive category at the 2026 Baton Rouge ADDYs. The campaign was created without AI in the creative or production stack. Other campaigns we have shipped used AI extensively in research, planning, and analysis stages. Both decisions were deliberate.

This is the framework we use to make that decision, drawn from real engagements. Use it as a starting point for your own AI policy if you do not have one yet.

The Question Most Agencies Are Avoiding

The dominant industry conversation about AI in 2026 has split into two unhelpful camps.

The first camp tells you AI changes everything and agencies that do not adopt it will be obsolete. They sell AI integration as the differentiator, regardless of what kind of work it produces or whether the audience can tell.

The second camp tells you AI is a creative shortcut that produces inferior work, and any agency using it is cheapening the craft. They sell artisanal craft as the differentiator, regardless of where AI might genuinely accelerate good work without compromising it.

Both camps are wrong because both are arguing about AI in the abstract. The useful conversation is about specific workflow stages, specific deliverable types, and specific audience expectations.

The right question is not "should agencies use AI?" The right question is "where in this specific project does AI add value, and where does it subtract?"

The Framework: Three Categories of Agency Work

Agency work falls into three rough categories, and AI's value differs sharply across them.

Category One: Research, analysis, and operations. This is where AI is most clearly accretive in 2026. Research synthesis on consumer trends, competitive analysis, market sizing, audience segmentation models, content audits, and large-scale data review are all dramatically faster and often higher-quality with AI assistance. The same is true for internal operations: meeting transcription, project documentation, knowledge base maintenance, and process automation.

The deliverables in this category are not what the audience experiences. They are inputs to the strategy and the work. AI shortens the time required to do good research and frees senior strategists to spend more time on synthesis and recommendation. The output the client sees is still human-judged, human-written, and human-defended.

Category Two: Iterative production support. This is the contested middle ground. Some kinds of production work benefit from AI assistance: transcription of recorded interviews, rough first drafts that humans heavily revise, code generation that engineers review and refactor, image generation for internal mood boards and reference. Other kinds of production work do not benefit, and AI involvement creates more work than it saves: nuanced copy editing, art direction decisions, motion design timing, voice and tone calibration.

The honest answer in this category is that it depends on the specific task and the specific person. A senior writer using AI to generate ten variations of a headline they then refine is faster than the same writer typing ten variations from scratch. A junior writer using the same tool to generate copy without the editorial judgment to refine it produces work no different from what AI alone would produce.

The rule we apply: AI is acceptable in production support when a senior practitioner is using it as a tool, with the editorial judgment to recognize and reject bad output. AI is not acceptable in production support when it replaces the practitioner's judgment.

Category Three: Brand campaign creative. This is where we draw the line. The creative ideas, the strategic positioning, the campaign concepts, the writing, the design, the art direction, the casting, the music, the motion, the voice. The work the audience actually experiences. The work that defines whether the campaign feels human, whether it feels true, whether it earns the audience's attention and trust.

We do not use AI for this work. Not for the concepts, not for the writing, not for the design, not for the production. Every campaign we have won industry recognition for in the last two years has been made by people, fully, in our studio in Baton Rouge.

The reason is not philosophical. It is practical. Brand work succeeds when it carries human truth, human craft, and human judgment in every layer. AI tools in 2026 are excellent at producing competent work that resembles human work. They are not yet capable of producing the kind of work that wins Best of Show or moves audiences in the ways the best brand campaigns do.

That may change. We will revisit the framework when it does.

Why the Alfa Campaign Was Built Without AI

The "That's Life with Alfa Insurance" campaign won Best of Show in Interactive at the 2026 ADDYs, plus a Gold for the same project. The decision to make it without AI was made early and held throughout production.

The campaign was built around relatable everyday mishaps, told through Alfie, the Alfa brand character. The success of the work depended on the specific human truth of those mishaps and the specific warmth of the brand voice. Both of those qualities are difficult to produce with AI tools that average across vast amounts of content. AI tends to produce work that is competent, polished, and forgettable. The Alfa campaign needed to be specific, slightly offbeat, and memorable.

The Sandlot-inspired animated spots that aired during MLB games on Bally Sports were animated by hand, frame-by-frame and through traditional rigging, by the animators who designed them. The interactive microsite was designed and developed by our team in Baton Rouge with no generative content. The writing came from our writers. The art direction came from our designers.

The campaign delivered measurable outcomes that gave the work commercial weight: 94 percent year-over-year increase in unique website visitors in the launch month. 55 percent year-over-year increase in quote requests. 156 percent increase in referral traffic. The work won Best of Show because it earned it.

We do not believe AI could have made this campaign. We do believe it could have made a competent campaign that resembled this one and won nothing.

Where AI Did Help on Other Recent Engagements

To illustrate the framework, here are categories of work where AI has accelerated our process meaningfully.

Audience research and segmentation. On a recent enterprise B2B engagement, we used AI tools to synthesize three years of customer interview transcripts, competitor messaging audits, and category analyst reports into a structured strategic input document. The work would have taken our strategy team 80 to 100 hours manually. With AI assistance, it took 25 hours. The strategist's judgment on what mattered, what to discard, and how to frame the synthesis remained human. The hours saved went into deeper concept development.

Technical research for Webflow Enterprise builds. On complex builds with non-standard integrations, AI has been useful for surveying technical documentation, identifying constraints, and generating starter code that our engineers then refactor and extend. Our Senior Web Engineer Tim Ricks, a 2x Webflow Community Educator of the Year, has been explicit about the tradeoff: AI accelerates the survey phase, and the refactoring phase still requires senior engineering judgment. The total time savings are real but smaller than the AI hype suggests, and they require senior engineers to evaluate output critically.

Content audits at scale. On a recent engagement involving migration of a content library with thousands of items, AI assistance let us tag, categorize, and identify gaps in source content far faster than manual review. The strategic decisions about what to migrate, restructure, or retire remained human. The mechanical sorting was AI-accelerated.

Internal operations. Meeting transcription, internal documentation, project setup, recurring report generation, and similar operations work have all moved meaningfully faster with AI tools layered into our workflow.

In each of these examples, the deliverable the client sees is still human-made and human-defended. The AI value is in the inputs and the operations, not in the creative output.

The Disclosure Question

A fair question we get from prospects: how should agencies disclose AI usage to clients?

Our position is that any AI usage in client deliverables, regardless of category, should be disclosed before contract. Buyers should know what they are paying for. An agency that uses AI for research and discloses it openly is making a different commercial offer than an agency that produces the same research manually. Both can be reasonable choices, but the buyer should know.

We disclose our AI policy to every prospect during sales conversations. Specifically, we tell prospects that AI is used in our research, analysis, and operations workflows, and that we do not use AI in brand campaign creative or production. We tell them where the line falls and why. The conversations that follow tend to be productive because the disclosure is upfront.

The agencies that hide AI usage are setting up trust problems they will eventually have to resolve. The audiences and clients are getting better at recognizing AI-generated work, and the disclosure conversation is going to become standard. Agencies that are early on disclosure will benefit from the trust capital that builds.

What Buyers Should Be Asking Their Agencies

If you are evaluating agencies in 2026 and you have not yet asked about their AI policy, you should. The question is not gotcha or punitive. It is information you need to evaluate the offer.

Ask: where in your process is AI used, and where is it not? Ask: what kinds of deliverables are AI-assisted, and what kinds are made entirely by humans? Ask: how do you decide where the line falls? Ask: how do you disclose AI usage to clients, and at what stage of the engagement?

Agencies that can answer these questions clearly are agencies that have thought through their position. Agencies that cannot answer, or that respond with vague language about being "AI-forward" or "AI-skeptical" without specifics, have not done the work yet. That is a legitimate signal to factor into your evaluation.

What Comes Next

We expect the framework above to evolve. AI tools in 2026 are different from AI tools in 2024, and they will be different again in 2027. What is true today about where AI does and does not add value will need to be revisited.

The principle that we expect to hold: human judgment at the center of brand campaign creative will remain the differentiator. AI will continue to compress research, analysis, and operations work. The agencies that use AI well will be the ones that recognize where it accelerates good work and where it threatens to flatten it.

We will keep publishing as the framework evolves. The position is not static, but the principle is.

ThreeSixtyEight is The Challenger Agency™, a brand, web, and campaign agency in Baton Rouge, Louisiana. Founded 2016. Webflow Enterprise Partner ranked in the top 5% of Webflow partners worldwide. Featured in Webflow's Generation No-Code documentary series. The first Baton Rouge company to earn B Corp Certification. Recent work includes "That's Life with Alfa Insurance" (2026 ADDY Best of Show, Interactive; 94% visitor lift, 55% quote lift), Tomb Raider for Crystal Dynamics (ADDY Gold; 715K active users, 23K registrations, 82% CRO lift in ten months), Opportunity @ Work (National ADDY Best in Show for Tear The Paper Ceiling; 128M LinkedIn impressions, 850K visits, 10K STARs pledges), Unilever, NASA, Khan Academy, Miami Dolphins, Rakuten, and LSU.

Reach out: hello@threesixtyeight.com

‍