The habit we said we’d keep

Last January we said this wouldn’t be a one-off. We’d grade ourselves against the six calls we made a year ago and publish a new one every year after. So before a single new call, here’s the report card, unedited, including the ones we didn’t get exactly right.

What we got right

We said disclosing AI use would stop being a liability and start being a trust signal, and that at least one major platform would roll out a formal AI-content disclosure label before the year was out. That happened fast. Meta began labeling a wide range of AI-generated video, audio, and image content as "Made with AI" starting in April 2024, expanding well beyond its earlier, narrower manipulated-media policy (Meta). Then in August, the FTC finalized a rule banning fake reviews and testimonials outright, including ones generated by AI or written on behalf of people who don’t exist, backed by real civil penalty authority (Federal Trade Commission). We called the direction. We didn’t call how quickly the floor would become a legal one instead of a reputational one. That’s on us, in a good way.

We also called a swing back toward bold, expressive design after a decade of flattened minimalist logos. We didn’t get three quiet, incremental examples. We got one loud one: Jaguar’s November 2024 rebrand, built around what the company called "exuberant modernism," saturated color, and a full identity overhaul ahead of its 2026 relaunch as an electric-only brand. It became one of the most argued-about rebrands in years, praised by some as bold repositioning and mocked by others as brand suicide. Either way, nobody scrolled past it. That’s the maximalist bet paying off, just louder and narrower than we framed it.

What we got wrong, or can’t fully prove yet

We predicted a high-profile brand would face public accountability for a stated value it never backed with action, sustainability, equity, community investment, and that the backlash would land in mainstream business press. The accountability moment arrived almost exactly on schedule. We had the direction backwards. The pressure campaign that hit brands hardest in 2024 didn’t come from people demanding companies live up to equity commitments. It came from organized backlash against having made them at all, and it worked. Tractor Supply eliminated its DEI roles, stopped submitting data to the Human Rights Campaign, dropped its net-zero pledge, and pulled Pride sponsorships in a single June 2024 announcement, framing the reversal as realigning with customer values. We said accountability was coming. We assumed it would come from the left. It came from the right, and it landed harder than either of us expected.

Our election-year and mobile-first predictions were directionally right and impossible to prove with one clean data point, which is its own kind of lesson: a prediction vague enough to dodge scrutiny isn’t much better than one that’s wrong. We’re tightening the language on both fronts below.

Six calls for 2025

Here’s what we’re willing to be held to this year.

01. AI agents stop being a demo and start being a daily tool

2024 was the year every platform announced an agent. Salesforce launched Agentforce in September, positioning autonomous, multi-step AI agents as what CEO Marc Benioff called "the Third Wave of AI," and set a public goal of powering a billion agent interactions by the end of 2025 (Salesforce). We predict at least one major creative software platform, Adobe, Canva, or Figma among the likely candidates, ships an agentic feature in 2025 that completes a full multi-step creative task inside a live project, not a single generated asset, and that at least one agency our size publicly documents cutting production time on a routine deliverable by half using it. The interesting question stops being whether you use AI. It becomes what a human actually decided.

“Fake reviews not only waste people’s time and money, but also pollute marketplace competition.”

Lina Khan, Chair, Federal Trade Commission

02. AI disclosure moves from brand choice to legal floor

We called this shift starting a year ago. In 2025 it finishes. With the FTC’s fake-review rule now enforceable, we predict at least one public enforcement action in 2025 against a company for undisclosed AI-generated reviews, testimonials, or fake engagement, and that "how we disclose AI use" becomes a standing line item in brand counsel conversations, not a marketing nice-to-have. Quiet stopped being neutral last year. This year it starts being risky in a different, more literal sense.

03. DEI messaging gets quieter, and the quiet gets misread by everyone watching

Pew Research found overall support for workplace DEI efforts sits at 56 percent among employed U.S. adults, but the number hides a split that makes caution the rational move for a lot of brands: Republicans divide almost evenly, 30 percent calling DEI a good thing and 30 percent calling it bad, while Democrats back it 78 percent to 4 (Pew Research Center). No safe middle exists inside that math. We predict a dozen or more additional publicly traded companies quietly soften DEI language in earnings calls, job titles, or annual reports in 2025 without formally ending anything, folding programs into broader "culture" or "belonging" umbrellas, and that at least one major brand gets criticized from both political directions for the same single decision, in the same month. Rebrand the department. The pressure doesn’t go anywhere.

04. Search stops meaning ten blue links

Google’s AI Overviews rolled out nationally in May 2024 with a stated goal of reaching over a billion people by year’s end, and Google itself claimed the feature drove more search use and higher satisfaction (Reid). We predict that by the end of 2025, a majority of informational Google searches in the U.S. surface an AI-generated summary before a single organic result, and that at least one major publisher or brand publicly reports a double-digit drop in organic click-through rate they attribute directly to AI Overviews. Ranking first stops being the whole game. Getting cited inside the summary becomes the new first page.

05. Mobile stops being a design consideration and becomes the only one

We said mobile-first would become a floor in 2024. We’re sharpening that for 2025: any brand or agency that launches a significant new digital product this year treating mobile as a secondary breakpoint, built for desktop first and adapted down, will show a visible, public performance or usability problem within its first quarter live. Not a hypothetical risk. A real one, provable in analytics within ninety days.

06. The small-brand design gap keeps closing, and it starts showing up in the data

We were proudest of this call last year, and we’re extending it with a harder edge. We predict that small business and independent brand work shows up more often, and more visibly, in 2025 design roundups, local business awards, and national trade coverage than it did in 2024, not because independent brands hired bigger agencies, but because better tools closed the distance. This is the one we’d bet the name on twice.

“We don’t make predictions to sound smart in a deck. We make them because six of last year’s are still sitting in public, waiting to be checked.”

Marisol Vega, Creative Director, HAUS XXIV

What we’re not predicting

Same three we refused last year, still refusing them: AI doesn’t replace the creative director, short-form video doesn’t kill every other format, and brand doesn’t die in favor of pure performance marketing. Every one of those calls gets made every January and none of them has landed since we started counting. We’re also not predicting the DEI pullback reverses course in 2025. The math in the Pew data doesn’t support that bet, and we’d rather be honest about a hard trend than optimistic about one we don’t believe.

We’ll grade this list next January, the same way we graded 2024’s, and keep going. See how far the pattern holds in 2026 predictions: what’s actually going to change this year.