Quick summary:
- 70% of organizations use AI in at least one business function, but agency adoption lags significantly behind other industries
- Change management matters more than technology — success requires shifting mindsets and work habits, not just buying tools
- AI amplifies subject matter expertise rather than replacing it, making experienced professionals more valuable
Tim Donovan was scrolling LinkedIn when he spotted yet another agency leader dismissing AI as a passing fad. The post claimed human creativity could never be replicated, that clients would always prefer the “human touch,” and that agencies should focus on their core strengths instead of chasing shiny new technologies.
Donovan, CEO of Seek Argus and former president at two machine learning companies, had seen this pattern before. During his time in credit underwriting and anti-money laundering technology, he watched entire industries transform while holdouts insisted their traditional methods were irreplaceable. The agencies clinging to old models reminded him of steam-powered factories refusing to adopt electricity.
His response was direct: “If agency leaders don’t realize that AI is like the invention of fire or the steam engine, they’re going to find themselves in a tough position.” The post sparked hundreds of comments and connections from agency owners hungry for practical AI implementation advice.
Speaking to host Greg Matusky on The Disruption Is Now, Donovan broke down why most agencies are falling behind, what successful adopters do differently, and how the industry can catch up before it’s too late.
Watch now:
Key takeaways:
Agencies are missing out on AI while other industries surge ahead
McKinsey’s 2024 State of AI report shows more than 70% of organizations now use AI in at least one business function. Marketing and sales, product and service development, service operations, and software engineering lead adoption.
But many agencies are laggards.
“The larger agencies and holding companies have been able to invest over time in technology and talent,” Donovan explains. “Smaller agencies haven’t. They typically don’t have an IT team. They don’t have anybody who’s an AI expert.”
Meanwhile, clients become more sophisticated with baseline AI tools daily, putting pressure on agencies to match or exceed their capabilities.
Agencies built on hourly billing models in particular face existential questions when AI can compress weeks of research into minutes or generate first drafts that previously required extensive human labor.
Change management beats technology every time
Matusky learned this lesson through trial and error. His firm now sees 13% revenue increases per employee, but the path required more than purchasing software licenses.
“You have to create an atmosphere of joy,” Matusky says. “You need to go out there and show people how they can advance their careers and do really cool things that will surprise and delight the client.”
Donovan agrees completely: “Change management — you can bring tools into an agency until the cows come home, but it really is about changing people’s mindset and their work habits.”
He argues that doesn’t happen through AI councils, where senior leaders meet monthly, discuss tools, then delegate implementation to junior staff. “That model does not work,” Donovan warns. “It’s a waste of time.”
Success requires subject matter experts in senior leadership roles mastering workflows first, then sharing practical guidance with their teams. Connor Grennan from NYU Stern School calls this “agency over AI” — humans maintaining control over tools rather than surrendering decision-making to machines.
AI amplifies expertise instead of replacing it
The biggest misconception about AI is who should use it. Some leaders assume junior staff should handle AI implementation while seniors focus on strategy. Others worry AI will commoditize their expertise.
Both perspectives miss the mark.
“I think it can be a tool for everyone depending on the use case,” Donovan says. “I have clients that are senior-level executives that want to formulate a strategic idea, and they use AI to take that idea and give it a body and give it context. I also have clients with junior staff members using AI for particular workflows like social media scheduling.”
Matusky discovered this principle through his own experience with messaging creation. For decades, he struggled to teach writing skills to talented team members. Finding great designers and videographers was straightforward. Finding exceptional writers proved nearly impossible.
ChatGPT changed everything. “When I saw it, I realized that I had spent most of my career trying to train other people to no avail,” he says. “Really with ChatGPT, it is teaching the machine how to write. If you do it through the prompt, you’re telling this individual how to write or rewrite.”
The technology doesn’t replace subject matter expertise. Experienced professionals who understand audience, tone, and strategy can guide AI to produce work that junior team members couldn’t create alone.
Large and small agencies face different AI barriers
Agency size determines specific implementation challenges. Large agencies and holding companies possess resources for technology investments but struggle with institutional inertia.
“They have an arrogance about them that prevents them from fully implementing,” Matusky observes. “Every time I talk to the head of a large agency, they talk in a way that I’m certain they don’t want to give up their role as an expert.”
These leaders often emphasize AI’s risks — hallucinations, client concerns, reduced billable hours — without acknowledging straightforward solutions like informed prompting and human oversight. Fear tactics protect their perceived expertise while slowing organizational adoption.
Smaller agencies operate with different constraints. Time scarcity tops the list. “You have very busy teams,” Matusky notes. “The time is hard to find.” While AI tools aren’t expensive — $30 per seat monthly for most platforms — training requirements and workflow changes demand significant attention from already stretched resources.
Nimbleness is an advantage for smaller agencies. They can experiment with new approaches, pivot quickly when something works, and avoid the bureaucratic approval processes that slow larger organizations.
Key moments:
- Donovan’s first realizations of the power of AI (3:01)
- The state of AI adoption and what’s hindering it (5:57)
- Big firms vs small firms in AI implementation (8:19)
- Why change management is so important (13:41)
- Is AI a bigger boost for experts or novices? (16:07)
- How AI surprised Donovan when he first saw it (20:49)
- Does AI show consciousness or just pattern recognition? (22:36)
- Machines and storytelling (26:59)
- Using AI to capture records of the human experience (34:18)
Q&A with Tim Donovan
Q: How should communicators think about their relationship with AI tools?
A: Agency over AI — AI is a tool. What we can’t have happen is people just give up their agency. This happens a lot where it’s like, I’m going to use ChatGPT, I’m going to do research, write a press release, and then take the output as gospel. That’s not the best use case. You have to have human-in-the-loop oversight and drive the value of the AI tool you’re using.
Q: What advice do you give agencies hesitant about AI adoption?
A: Don’t let perfection be the enemy of the good. You’re going to get good content if you know how to prompt and you know what I call an LLM stack, where you can go between different tool sets to verify data and citations and writing styles. If you’re not taking AI seriously as an agency, you’re going to struggle. There’s an immense amount of opportunity if you implement AI in the right way.
Q: What does successful AI implementation look like in practice?
A: You need to have subject matter experts in senior leadership roles who are the tip of the AI spear. If they master workflows or processes or procedures using particular tools, they can share and guide their junior staff on how to take advantage of those things. It has to start at the top.
Q: Will AI ever truly replicate human creativity and nuance?
A: Do machines understand sort of the human experience nuances at this moment? No, not at all. We’re not there yet. I think over time though, the more data that is put into these LLMs and the more guidance and use cases, you’ll start to see machines provide maybe that kind of tonality and be able to discern that. But we’re a ways away from that.
