LG Ad Solutions’ Dave Rudnick On Why AI Is About Amplifying Experts, Not Replacing Them
“AI isn’t one giant brain that replaces people. It’s a set of focused agents that make experts more powerful,” notes Dave Rudnick, CTO, LG Ad Solutions. “The companies that get their data ready and start deploying agents now will pull ahead fast.”
We sat down with Rudnick to discuss how LG Ad Solutions is using AI to make advertising workflows smarter, faster, and more collaborative.
ALAN WOLK (AW): When you think about AI in the advertising workflow, where do you see the most immediate impact?
DAVE RUDNICK (DR): Advertising has always had the same rhythm. A campaign starts with a request, then insights and planning, then the IO, then the run. It performs well or needs adjustment, and then you look back and ask, “What happened?”
AI lets us make that process more efficient. First, you need your data in a place where you can do something meaningful with it. Then you start inserting intelligent agents into workflows to solve specific problems. These agents make repetitive, data-heavy work faster. Calculations that used to take days now take hours.
The result is people spending less time on manual tasks and more time thinking about new opportunities. Everyone becomes a conductor, orchestrating their own small set of agents. The best part is seeing employees take that time and use it to generate ideas — new audiences, new recommendations, new kinds of associations. That’s where creativity starts to accelerate.
AW: You’ve said AI doesn’t replace expertise. How does that work in practice?
DR: AI has to be trained and constantly guided. It drifts if you don’t. The expert keeps it on track. The agents will always need “parents” and those parents have to be experts in what they do to keep the agent performing.
Take pacing as an example. Before, someone would check performance each morning and make manual adjustments. Now, you can tell an agent, “Alert me if we’re off track and make small corrections automatically.” That’s 24-hour oversight, not replacement.
When people use those systems to deliver better outcomes — say five or ten percent more efficiency for a client — that’s real impact. It’s not about removing people. It’s about giving them better tools.
AW: How does that differ from the large language models people talk about?
DR: Large models like ChatGPT are built to do everything, which makes them generalists. They can be wrong because they draw from a huge range of information. But when you build an agent trained only on fixed internal data — for example, HR policy or ad rules — it becomes extremely accurate.
Advertising data works the same way. We tell it what the data is, what the limits are, and how to use it. Later, we can connect one agent to another, and that’s where things get interesting. Two agents, each with their own knowledge, can find relationships no one saw before. But experts still manage the guardrails. That part never goes away.
AW: Tell me about how LG Ad Solutions is working with Databricks.
DR: Every company ends up with data in a dozen different systems — from different departments or acquired companies. Databricks lets us bring all of that into a single environment called a lakehouse. Think of it as a giant pool of unstructured data.
Historically, sharing or analyzing large data files required manual transfers through secure File Transfer Protocol (FTPs) or other limited-access systems—a process that was time-consuming and difficult to scale. The collaboration is designed to make it easier, faster, and more efficient for brands, agencies, networks, and measurement partners to access its global viewership dataset and combine it with their own first-party data. Through this integration, contracted data partners can streamline analytics, enhance planning, and make more informed media decisions—all within Databricks’ governed, privacy-first environment.
You don’t define it upfront; you store it as-is. The platform sits above and lets you access what you need on demand. It’s faster and more flexible because you’re not constantly reprocessing data. That’s huge for us, with petabytes of information across ad delivery, planning, and performance.
AW: The Databricks release also mentioned being privacy-first. How does that play out?
DR: The software runs entirely within our environment, so the data never leaves our control. We can tag sensitive data dynamically and use clean rooms to collaborate with partners. Clean rooms are key because they let two companies find overlap or create shared segments without exposing PII.
It’s like the name implies — a controlled environment where the data can’t be contaminated. That allows innovation while staying compliant with privacy laws globally.
AW: Are clients already familiar with this kind of setup, or is there a learning curve?
DR: There’s definitely an education phase. The industry has always been data-driven, but now the infrastructure itself has become strategic. Everyone is racing to get their data environments right so they can build agents on top of them.
The pattern we’re seeing is clear: most companies have moved to the cloud or are halfway there. Then comes organizing data into lakehouses. After that comes agent development. Once you’ve built that foundation, things scale fast.
We’re deploying about 30 agents and multiple MCP connectors this year. The companies doing that kind of work now will have a serious head start. Those still cleaning up on-prem systems will be playing catch-up while others are already measuring AI-driven efficiency gains.
Every major holding company and Fortune 500 brand is prioritizing AI as an efficiency driver for 2026. It’s now part of every strategic plan.
AW: How do you see this evolving over the next year?
DR: Internally, we’re building agents to make our teams more effective. The next step is collaborative agents that work across partners — agencies, brands, tech companies — solving shared problems together.
The goal is for LG Ad Solutions to be recognized as a partner that uses AI to improve how the entire ecosystem operates. That’s where we’re focused: applying intelligence not just for our own workflows, but for how we collaborate across the industry.
