🧠 The Current State of AI in Major DSPs

Will It Replace Media Traders Anytime Soon?

Author

Vlad Chubakov
July 17, 2025

🧠 The Current State of AI in Major DSPs — Will It Replace Media Traders Anytime Soon?

AI is the hottest topic in every boardroom, strategy deck, and conference keynote. From "AI-first" platforms to predictions that AI will soon replace human marketers, it's easy to feel like you're already obsolete if you're not coding prompts or building chatbots.

But for those of us in the weeds of real-world programmatic, the view looks a little different.

Let’s cut through the hype and take a closer look at how AI is actually used in major DSPs today — and whether it’s anywhere close to replacing media traders.

📦 1. Campaign Setup

Still entirely manual.

Whether you're in DV360, Amazon DSP, or The Trade Desk, setting up a campaign in 2025 looks exactly like it did in 2018. You log into the UI, create line items, maybe use SDFs or bulksheets if you're scaling. No chatbots, no AI wizards, no “describe your goals and we’ll build your campaign” button.

Not even an autocomplete that suggests audiences or formats based on your brief.

❌ No AI involved in campaign setup across the major DSPs.

🔧 2. Optimization

Amazon and DV360 are still fully manual. If you want to adjust targeting, budgets, pacing, or frequency caps, you're doing it yourself.

The Trade Desk offers something more nuanced: KOA (their optimization AI) can auto-remove underperforming audiences and recommend tweaks. It’s better than nothing — but still limited. If your setup is flawed or missing strategic context, KOA won’t help. It doesn’t see the bigger picture.

✅ TTD offers some AI assistance via KOA.
❌ DV360 and Amazon rely 100% on manual input.

💰 3. Bidding

This is the one area where AI has been baked in for years - smart bidding models have long used machine learning to predict value and optimize pricing in real time.

But here’s the nuance: media buyers often assume bidding = optimization, when in reality, it’s only one part of it. AI handles the bids, sure. But strategy, targeting, pacing, budget shifts, and learning phase interventions? That’s still on you.

✅ AI is embedded in bidding algorithms.
⚠️ But it doesn’t replace human strategic oversight.

🕵️ 4. QA (Quality Assurance)

This might be the most outdated part of the workflow.

To QA a campaign, you download SDFs or Sheets and check everything manually - targeting, budgets, frequency, flight dates. Some teams use external scripts or tools, but nothing in the DSPs themselves is powered by AI.

Imagine a system that flags “you’re spending $50K on a test campaign with one audience” or “this campaign is targeting Peru but only in English.” That would be helpful. But it doesn’t exist yet.

❌ No AI-driven QA across any major DSP.

📊 5. Reporting

Still painfully manual. You download reports, format them, maybe build a Looker Studio dashboard. There are no built-in GPT-like copilots in DSPs that summarize performance, suggest insights, or automate reporting workflows.

Some agencies or brands are hacking this using external LLMs — but this isn’t native to any platform yet.

❌ No AI-native reporting or insights available yet.

🎨 6. Creative Development

Here’s where things get interesting.

  • Amazon DSP is far ahead - especially for endemic brands. You can log in, describe the creative you want, and the platform generates images. You can even create product videos.

  • TTD announced its Generative AI Creative Marketplace - sounds promising, but it’s not live yet.

  • DV360 offers nothing in this space — no AI creative tools, no GenAI integrations.

✅ Amazon leads in AI creative generation.
🚧 TTD is exploring it.
❌ DV360 has nothing yet.

🧵 TL;DR – AI in DSPs Today

Function

DV360

Amazon DSP

The Trade Desk

Campaign Setup

❌

❌

❌

Optimization

❌

❌

⚠️ (KOA)

Bidding (Algo-based)

✅

✅

✅

QA

❌

❌

❌

Reporting

❌

❌

❌

Creative Development

❌

✅

🚧 (coming)

So… Will AI Replace Media Traders?

Not anytime soon.

The platforms still rely heavily on human setup, QA, analysis, and strategy. AI is present - but mostly under the hood (bidding models) or in early-stage experiments (KOA, GenAI creatives).

In short: most media traders still spend their time debugging UIs, fixing 404 errors, uploading bulksheets, and analyzing performance in Excel — not chatting with AI assistants.

This isn’t a criticism of DSPs - innovation takes time, and none of us want rushed features that break our campaigns. But it’s important to ground our expectations. The hype cycle is loud. The reality is slower.

💡 Bonus Insight: Where AI Is Moving Faster

Interestingly, smaller players are moving quicker. New DSPs and adtech startups are integrating LLMs, QA tools, and AI copilots to fill the usability gap. These players are often more nimble and user-obsessed - and the big DSPs may end up copying their innovations down the road.

If you’re watching the AI wave in adtech, keep an eye on the underdogs.

Want more hands-on breakdowns like this?
Subscribe, share, and let me know what platform you want me to dissect next.

👋 Let’s connect

  • LinkedIn: @chubakovnt

  • All links & resources: vchubakov.carrd.co

    I offer consultations for programmatic marketers and ad buyers who want to get more signal, less fluff.
    Just fill out this quick form - I’ll personally reach out.