From AI Assistance to Agentic Execution: Skai’s Take on Amazon Ads unBoxed 2026

• Amazon Ads is moving from AI assistance toward agentic execution. Planning, targeting, creative, optimization, and analytics are becoming parts of a more connected workflow.
• The distance between insight and action is shrinking. Faster data, conversational analytics, and AI-assisted execution give marketers more opportunities to act while signals still matter.
• Commerce discovery is becoming conversational. Product education, comparison, recommendation, and conversion are starting to happen within the same guided experience.
• The bigger challenge is cross-platform. As every major platform develops its own AI capabilities, marketers need intelligence and governance that connect decisions across the entire media portfolio.

I’m writing this from unBoxed, where Amazon Ads just rolled out announcements across media buying, measurement, creative, commerce, and data. There is a lot to digest.

I also spent enough years at Amazon to know it’s easy to get caught up in individual product announcements. I think that misses the bigger story. When you put these announcements together, Amazon’s bet becomes pretty clear.

Amazon is making it easier to move from a question to an insight, from an insight to a decision, and from a decision to action. Planning, campaign creation, optimization, measurement, and creative development are starting to connect through AI. Amazon isn’t just adding AI features to advertising. It’s building toward an advertising environment where AI agents increasingly help marketers plan, analyze, optimize, create, and execute.

That’s a meaningful shift. But here’s what I’d actually take from unBoxed 2026’s announcements: as Amazon makes advertising inside its own ecosystem more connected and automated, advertisers have to think much more seriously about how those decisions connect to everything happening outside of it. Your customer doesn’t live inside one advertising platform, and neither should your strategy.

What does Amazon Ads Agent tell us about where media buying is going?

Amazon Ads Agent is probably the clearest example of where Amazon is headed. Amazon is bringing Sponsored Ads, programmatic capabilities, account management, campaign workflows, insights, and AI-assisted execution into a more connected advertising environment. Amazon DSP is also being retired as a name. The programmatic capabilities aren’t disappearing. They’re becoming part of this unified experience.

That’s more than a branding change. Amazon is bringing capabilities that used to require separate workflows closer together and putting a conversational layer on top of them. Among the capabilities Amazon announced:

• Agentic Media Planning uses conversational requests for reach forecasting, audience sizing, and budget allocation.
• Natural-Language Analytics lets marketers ask questions about campaign, portfolio, and brand performance through a chat-based interface.
• Agentic Guidance and Campaign Targeting brings recommendations and natural-language targeting into Sponsored Ads workflows.
• Full-Funnel Campaigns bring Sponsored Ads, Display, Video, and Streaming TV together under a single budget and objective.
• AI-powered creative tools can adapt and generate images, headlines, audio, and video within campaign workflows.

The direction feels pretty clear. AI is moving from something marketers use alongside the workflow to something increasingly built into the workflow itself. A lot of the manual work marketers have spent years doing probably shouldn’t be manual anymore. The more interesting question is what marketers do with the time and decision-making capacity they get back.

What happens when every advertising platform has its own agent?

Amazon isn’t the only company building this way. Every major media and commerce platform is moving toward more AI-assisted planning, optimization, analytics, and execution. That creates an interesting dynamic: individual platforms become easier to operate while the overall media environment can become harder to coordinate.

Amazon’s agent has deep knowledge of Amazon. Other platforms will have deep knowledge of their own environments. Each will have its own data, recommendations, objectives, measurement, and increasingly its own way of taking action.

But the advertiser is sitting above all of them:

• Where should the next dollar go?
• Which product has the biggest opportunity across retailers?
• Where do recommendations from different platforms agree or conflict?
• Is an Amazon recommendation still the right decision when you account for what’s happening on Walmart, Instacart, search, social, or CTV?

Those are different questions, and this is where I think the next phase of agentic advertising gets interesting. The goal isn’t simply to have more agents. It’s to connect the intelligence coming from those systems so marketers can make better decisions across the whole business.

That’s a big part of how we’re thinking about Celeste at Skai. Amazon can make Amazon advertising smarter. Celeste can help marketers understand Amazon alongside the rest of their commerce and media portfolio, connecting product, audience, creative, search, retail media, and performance signals to help determine where opportunities are emerging and where investment should go.

Native intelligence and cross-channel intelligence aren’t competing ideas. Advertisers need both.

How is Amazon changing the way consumers discover products?

The changes Amazon announced aren’t only happening behind the scenes. Branded Conversations stood out to me because it points to a different version of commerce discovery. Instead of moving through a familiar sequence of search, click, product detail page, compare, and purchase, shoppers can increasingly ask questions, learn about products, compare options, and receive recommendations through a guided conversation.

A few of the announcements reinforce that shift:

• Branded Conversations creates multi-turn AI shopping conversations launched from Sponsored Brands Collections ads.
• Responsive eCommerce Creatives automatically assemble existing brand assets into display formats that can include conversational prompts.
• Review Requests gives advertisers a way to prompt verified purchasers to leave authentic reviews on high-traffic Amazon surfaces.
• Sponsored Services begins extending Amazon’s advertising signals into local services such as home services, dining, beauty, and events.

Put those together and the role of content starts to change. Product information matters. Reviews matter. Creative matters. The quality of the information a brand gives these systems matters. We’re moving toward an environment where the answer a shopper receives may matter just as much as the ad they see.

For brands, that makes understanding shopper intent even more important. Which questions are customers asking? Where are they hesitating? Which products need more education? Where are review gaps creating conversion risk? Which messages actually help move someone toward a decision?

Those signals shouldn’t live in isolation. The opportunity is to connect them with performance data across retailers and use them to decide which products, messages, audiences, and investments deserve attention.

Why does faster data matter when AI is making more decisions?

One part of unBoxed 2026’s announcements that shouldn’t get lost is the infrastructure underneath all of this. Amazon announced several updates that move data, measurement, and activation closer together:

• Amazon Marketing Stream v2 moves reporting from hourly JSON delivery to Apache Parquet files delivered every 15 minutes.
• Ads Data Manager provides a unified entry point for connecting first-party data into AMC and Amazon Ads Agent.
• Publisher Insights in AMC will add privacy-safe signals from participating third-party streaming TV and web publishers.
• Geographic Insights and Activation combines Amazon retail data with advertisers’ geographic first-party data to identify and activate against local opportunities.
• The AMC SQL Generator capability through Amazon Ads MCP Server enables partner AI agents to turn natural-language questions into validated, execution-ready AMC SQL.

That last one is particularly interesting for us at Skai. The Amazon Ads MCP Server itself isn’t new. Amazon launched it earlier this year. What’s important here is how those capabilities are expanding and what partners can build with them.

At unBoxed, we’re showing how Skai Studio can use Amazon’s MCP Server to bring AMC audience intelligence into a broader media-planning workflow.

That’s where MCP gets interesting. It’s not just that another system can generate a query faster. Data and intelligence that used to sit inside a specific tool can become part of a larger decision-making process.

The distance between measurement and activation is shrinking. If an agent can understand what happened, determine what it means, recommend what should happen next, and eventually take an approved action, measurement stops being something marketers look at after the fact. It becomes part of execution.

What does Amazon’s expansion beyond its own inventory mean?

Another announcement worth watching is Unified Buying. Amazon is bringing Amazon-owned CTV and audio inventory together with more than 43 connected streaming publishers through a single Programmatic Guaranteed deal, with one deal ID, frequency cap, and report.

That’s important because the Amazon advertising story increasingly extends beyond Amazon-owned surfaces. The same is true of Publisher Insights in AMC and Amazon’s expansion of interactive video experiences into third-party streaming environments.

For advertisers, that makes portfolio-level decision-making even more important. Amazon can optimize within its buying environment. Marketers still need to understand incremental reach, audience overlap, creative performance, and investment tradeoffs across Amazon, CTV, retail media, search, social, and other channels.

That’s where Skai’s role is different. The opportunity isn’t to recreate what Amazon does natively. It’s to help advertisers understand how Amazon fits into everything else they’re doing.

Does more automation mean marketers give up control?

No. I think it makes control more important. Amazon itself acknowledges that AI-generated responses can contain mistakes and that important information should be verified before action is taken. That’s worth paying attention to.

As AI moves from answering questions to changing bids, budgets, targeting, creative, and campaign settings, governance becomes part of performance. Marketers need clear answers to questions like:

• Who can approve an action?
• How much can an agent spend?
• Which data can it access?
• What happens when two systems recommend different things?
• Can you understand why a decision was made?
• Can you stop it?

Those questions aren’t barriers to agentic advertising. They’re what allow companies to use it at scale.

At Skai, we’ve been thinking a lot about this as we build agentic capabilities. The goal isn’t autonomy for the sake of autonomy. It’s giving marketers the ability to move faster while keeping the right controls, transparency, and accountability in place.

What should advertisers take away from unBoxed 2026?

After hearing the announcements, I wouldn’t look at them as a checklist of features to turn on. Look at the direction they’re pointing.

Amazon is connecting more of its advertising ecosystem. Data is moving faster. AI is becoming part of planning and execution. Commerce discovery is becoming conversational. Amazon’s media footprint is expanding. Measurement is getting closer to activation.

That’s a lot of change, but the bigger change may be what happens to the marketer’s job. If platforms increasingly handle the mechanics of optimization themselves, the value shifts toward deciding what should be optimized, how investments should move across platforms, what business outcomes matter, and where human judgment needs to stay in the loop.

This story isn’t about marketers handing everything over to AI. It’s about marketers operating at a different level.

Amazon made its bet on that future clear. Now advertisers need to decide how all of those increasingly intelligent platforms work together.

---

Frequently Asked Questions