This is an excerpt from the Digiday+ research report, 2026 Marketer’s Guide to AI Applications, Agent AI, AI Search, and GEO/AEO, which examines how marketers are navigating the opportunities and challenges posed by AI as an integral part of marketing. The report is based on a survey of 142 brand and agency professionals and individual interviews with marketing and technology executives responsible for AI investments and application development..
A Digiday study (conducted annually starting in 2022) found a significant increase in the adoption of AI technology by marketers. In 2022, 44% of brand and agency professionals said their companies are investing in AI technology. This percentage rose to 57% in 2023, 71% in 2024, and reached 86% in 2025.
The growing importance of AI to marketers is evidenced by the number of companies that have created chief AI officer positions in the past two years. In 2024 and 2025, brands General Motors, Mastercard, and ZockDoc appointed heads of AI, as did agencies Golin, Lackey & Company, and Horizon Media.
“In every industry, there will be a certain percentage of companies that: [AI] If not, most companies don’t. And I think the economic upside of understanding that is what makes such a huge competitive difference,” Wesley Ter Haar, co-founder of digital agency Monks and recently named chief AI officer, told Digiday in April.
Consumer adoption of AI has also increased significantly, with the result that many brands now regularly use AI in their consumer applications. For example, PetSmart has relaunched a membership program that uses AI to customize deals based on customers’ past purchases. Guitar Center also launched a chatbot called Rig Advisor to help customers choose the right product for their needs.
But as AI technology evolves and becomes more complex, many marketers Digiday spoke to for this report said employee training on how to best use AI tools lags behind overall adoption.
Dan Gardner, co-founder of creative agency Code & Theory, said this is especially true when it comes to upskilling and reskilling team members. “Anyone can learn new tools. Upskilling and reskilling multiplies the value of human ingenuity,” Gardner says. “Designers, for example, are trained in communication design. Using AI tools to make design a little easier doesn’t improve their skills. Designers are just using new tools. The way they improve their skills is to multiply the value of understanding communication design. There isn’t enough emphasis on implementing new ways of working and tools.”
Matt Maher, founder of M7 Innovations, an independent research and development company, said that while individual users may be comfortable with AI tools, businesses generally aren’t making the most of them. “There is no question that users are savvy at a baseline level, but there is a difference between being able to understand and get the most out of a tool like ChatGPT, which has 800 million weekly active users, and Gemini, which has 400 million monthly active users,” said Maher.
“When a company hires [Anthropic’s] Claude uses APIs for all internal software. [Microsoft] “It’s basically a copilot for everyone, and it feels like a big machine. There’s little imagination to how much you can actually use these tools if you push them to their limits,” Maher added. …Big tech companies aren’t very good at showing people the great things they can do. …At baseline, we are all getting used to AI, but there are still gaps. ”
According to Mark Marais, global CTO at design and technology agency Huge, when organizations, including marketing teams, move to implement AI technology before fully developing a plan for how the technology will be used, there is often a difference between the amount invested in AI tools and the return on that investment.
“There are huge investments being made in AI, but the foundation of it all is the infrastructure layer, or TPU. [tensor processing units] GPU from Google [graphics processing units] “With NVIDIA, someone has to pay for it. Every time an agency or a brand wants to deploy a model, the Googles and Amazons of the world have to find a way to monetize the infrastructure layer and every layer within the AI ecosystem,” Maleha said. Paying for that infrastructure is becoming a big concern for brands, Marais explained. “What if I want to activate another 500 seats of Claude code? What will that mean financially? Will I get that money back if it only increases productivity by 30%?” Marais asked.
“GPUs and TPUs are being funded and recognized for their economics,” he added. “Suddenly, these models have to be monetized in a real way. AI is and always will be a shiny object. Brands thought, ‘I want a press release, so let’s worry about GPU and TPU fees and API calls later.'”
Digiday’s research found that the majority of marketers continue to introduce AI technology into their workflows using ready-to-use AI tools, rather than building tools in conjunction with existing large-scale language models such as Google’s Gemini or OpenAI’s GPT, or building and training their own LLMs in-house.
85% of survey respondents said their organization uses AI tools out of the box. Less than half of respondents (40%) say their company uses existing LLMs to build its own tools, and only 19% say they build and train their own LLMs.
The cost of building a customized AI tool with an existing LLM or building and training your own LLM and the learning curve associated with implementing these options are likely reasons why most marketers choose to use off-the-shelf AI tools. Smaller companies may not have the luxury of having an AI team dedicated to creating custom tools.
Huge’s Maleh noted that over the past year, marketing teams have had several new AI tools available at their fingertips. “What has happened is that there are now many more models available out of the box,” Marais said. “You can start with a basic model that someone else has already taken the time to create, whether it’s Adobe or Google, and customize it to fit your needs. It has many of the capabilities that Google’s Cloud Platform has in ready-to-use tools like Vertex.”
Google Vertex AI is an AI development platform that allows users to build their own custom AI or machine learning models using Google Cloud’s infrastructure. The platform provides pre-built models that serve as a base for users to build custom tools and functionality.
Marais said another change in the AI landscape that has taken shape over the past year is the democratization of AI models and collaborations between some major industry players, such as the recent partnership between Adobe and Google, which integrated Google’s Gemini, Imagen, and Veo models into Adobe’s creative tools. “If you are currently using Adobe’s Firefly, but would like to use Google’s Nano Banana as your asset generation model, you can do so within the Firefly console,” Maleh explained. “That wasn’t the case a year ago. It wasn’t Firefly or anything. … We’ve gone from where a lot of the platforms were walled gardens to more open spaces.”
M7 Innovations’ Maher said some tech companies are lowering the barriers around AI services, allowing brands to build on existing capabilities. “What I’m starting to see now is the technology stack,” Maher said. “There’s not going to be anyone who rules them all. I’ve seen brands say, ‘Copilot is our base, and we’re stacking Claude on top of that with a bunch of really smart APIs.’ Or, “I use Adobe Firefly to create and Canva to complete.”
Maher added that this could lead to significant savings for brands. “We’re starting to see cost efficiencies by partnering with large companies but also creating our own versions and sandboxes. … And we’re saving a lot of money because we don’t have to build from scratch,” he said.
