Published July 13
For all the hype around AI, it seems the productivity gains inside agencies are fairly modest right now.
Speaking at ATS Singapore last week, Omnicom-owned PHD’s APAC president, Eileen Ooi, estimated AI is currently delivering “about 10% to 20%” productivity gains across the agency’s automation of insights and reporting.
The reality, she said, is that the industry still has significant hurdles to overcome around AI adoption.
Her comments came during a panel where speakers also pushed back on AI “hype” and warned that, as AI becomes embedded in agency workflows, rising technology costs will force the industry to rethink traditional labor-based pricing models.
At Omnicom, for instance, Ooi said agentic capabilities are now integrated throughout the company’s Omni platform, supporting teams “all the way from looking at the brief … to actually now looking at activation.”
She was candid about where the industry stands though.
“Hand to heart, no,” she said when asked whether the organization had reached a gold standard in AI adoption. “We’re at a pivoting stage.”
The biggest myth about AI agents
Ooi also challenged the idea that agentic AI might simply replace human decision-making altogether.
“The biggest myth that we really have to break is the notion that agentic AI will just sort it. … It changes the way we work, but it doesn’t solve everything,” she said.
That measured view was shared by Ganga Chirravuri, president of product and solution development at Dentsu, who argued that the industry’s understanding of AI has become overly simplistic.
“The definition of AI itself is fairly bastardized right now,” he said.
Agencies have long relied on predictive analytics and data science, he noted, with generative AI representing the latest evolution rather than a new capability.
Chirravuri said the industry’s next challenge is building genuinely agentic systems.
“The independence of the orchestration layer is very important when you talk about being agentic,” he said, explaining that effective AI agents require “fine-tuning” to complete increasingly sophisticated tasks.
“Do you want to make everything truly agentic? It requires a little bit more work, to be very honest.”
AI won’t be free
While productivity remains a key measure of AI’s success, Ooi argued that the industry is overlooking another critical issue: the cost of deploying AI at scale.
“I don’t think as an industry we’re talking about it enough,” she said. “We’re unfortunately overly focused on … efficiency.”
“Tech cost has always been absorbed by [the] agency. … I think we now need to have a conversation more openly with our clients around the fact that tech does cost money to the organization.”
As AI becomes embedded across agency services, she said the industry will need to rethink traditional pricing models based around service. “There is a need to have a conversation around blended costs now,” Ooi added.
Costs notwithstanding, there’s no doubting the value of AI. Vincent Pang, APAC managing director at Locala, said the diversity of APAC makes AI particularly valuable.
AI allows planners to process far more local market signals than would be possible manually, he said, helping brands make better decisions about where to invest and how campaigns should be adapted market by market.
He added that AI can also help agencies overcome the operational complexity of omnichannel campaigns, ensuring detailed media strategies can be applied effectively across mobile, CTV, digital out-of-home and social.
“This is where AI and automation can be really practical, to make sure that a market-level plan can actually become a market-level setup, not just a beautiful strategy that disappears before it reaches execution,” Pang said.