More AI won’t fix advertising. Coordination will.

Artificial intelligence is now embedded in every layer of advertising.

Creative platforms use it to generate and test ad variations. Audience tools use it to build segments. Measurement providers use it to model attribution. And media buying platforms use it to optimize bids in real time.

None of that is the problem. The problem is that these systems don’t talk to each other.

Each vendor has built its own AI, tuned to its own data and optimized for its own objective. The result is a stack full of intelligence that operates in isolation and, in many cases, actively works at cross-purposes.

The next phase of AI won’t be defined by building more intelligence. It will be defined by how well these systems work together.

The breakdown isn’t theoretical. It’s already happening.

An advertiser deploys an AI-powered creative platform to optimize ad variations for engagement. Meanwhile, the DSP’s audience models are optimizing for conversions. The creative AI is pushing in one direction. The audience AI is pulling in another. Neither system knows the other exists, and the marketer sitting between them has no way to see the conflict, let alone resolve it.

Multiply that across measurement partners, data providers and supply platforms, and the real risk comes into focus. It’s not bad AI. It’s uncoordinated AI, where the “why” behind decisions quietly disappears.

For the open internet, this matters more than most people realize.

Walled gardens can coordinate internally because they control the entire stack. The open internet doesn’t have that luxury. If we want AI to be an advantage for the open ecosystem, and I believe it can be, we need coordination, not just more algorithms.

Your bidder is your edge. Don’t smother it.

A DSP’s bidding algorithms aren’t generic models. They’re built on years of signal, derived from bidstream data across dimensions like publisher performance, geography, device type and ad format. That’s intelligence that no general-purpose large language model (LLM) can replicate. It reflects years of refinement and deep integrations across the programmatic ecosystem.

When teams layer on disconnected AI tools without understanding what the bidder already does, they risk undermining the most refined intelligence in their stack. AI shouldn’t compete with the bidder. It should extend it.

This is also why the DSP should sit at the center of AI orchestration. It’s where the media dollars are actually spent. It’s the system that translates signals into decisions. If AI systems don’t connect back to that layer, it’s just automation running in a vacuum.

Start with your data. Let your DSP orchestrate.

The practical path forward starts with where data already lives.

Increasingly, that’s cloud data platforms — where customer records, conversion events, and campaign signals are centralized. That’s the connective tissue.

But the value of that data isn’t unlocked by having outside agents make optimization decisions and push them into your DSP. It’s unlocked by giving your DSP direct access to those signals so it can incorporate them into its own decisioning.

Creative performance, audience insights and measurement signals should all flow into the decisioning layer, not sit in separate tools issuing competing instructions.

The question I’d challenge every marketer to ask: Is your AI strategy giving your DSP better inputs or trying to override its outputs?

Connection without governance is its own risk

But let me be clear: Connecting AI systems is not the same as opening the floodgates. As agents increasingly interact with other agents across the stack, the potential for unintended data exposure grows. Automated interactions are harder to monitor than human ones, and without proper guardrails, the same integrations that enable coordination can also create risk — data leakage, signal loss or unintended inference.

Orchestration has to be intentional. Every connection point needs transparency into what’s being exchanged, why and under whose control. The goal isn’t connectivity for its own sake. It’s coordination with accountability.

The opportunity ahead

The open internet has always stood for choice and transparency. AI should reinforce those principles — not erode them by creating more black boxes.

The platforms that will matter are the ones that connect partners deliberately, with clear boundaries, enabling AI across the ecosystem to work in concert rather than in conflict.

At The Trade Desk, this is the vision behind OpenTTD — a foundation designed to connect data, AI and partners in a way that enables orchestration with purpose, transparency and control.

Not more AI for the sake of it. Better-coordinated AI, in service of outcomes that actually matter to the business.

The opportunity isn’t artificial intelligence. It’s coordinated intelligence, built on trust.

The Current is owned and operated by The Trade Desk Inc.