Amazon Ads Agent: Can AI Run Campaigns From Ads to Streaming TV?

Amazon Ads Agent can already help eligible Australian advertisers plan, build, optimise and analyse some campaigns through natural language, particularly across Amazon DSP workflows and selected beta features. But the promise of one AI running everything from Sponsored Ads to streaming TV is not universally available in Sydney today. Full-Funnel Campaigns remains US-only, DVA+ is still rolling out, and advertiser-set budgets, creative approvals and manual controls remain central.

Amazon is quietly changing one of the most labour-intensive parts of digital advertising: the operating layer between marketing strategy and campaign execution.

At unBoxed 2026, Amazon renamed its unified advertising platform Amazon Ads Agent and positioned agentic AI inside the workflow used to plan, create, optimise and analyse advertising. The proposition is broader than another copy-generating marketing assistant. Amazon wants advertisers to be able to describe an objective, provide commercial parameters and let its systems coordinate more of the mechanics required to turn that objective into live media.

The significant phrase, however, is more of the mechanics.

Amazon Ads Agent does not mean every advertiser can hand an unrestricted budget to an AI system and walk away. Nor does the appearance of Sponsored Ads, display, online video, streaming TV and audio within one platform mean every automated capability has simultaneously launched in Australia.

For Sydney advertisers, agencies and national brands, the immediate issue is therefore not whether agentic AI is arriving in media buying. It already has. The practical issue is understanding which decisions the system can make now, which still require advertiser approval, and which headline features remain part of a staged international rollout.

Amazon Is Turning the Campaign Console Into an Operating Agent

Traditional digital advertising systems have largely required people to translate commercial intent into platform settings.

A marketing team decides it wants to increase sales. A media planner estimates audiences and channel allocation. A buyer creates campaigns. Somebody configures targeting, bids and budgets. Creative teams prepare assets. Analysts extract reports. Optimisation teams then interpret the results and adjust the campaign.

Amazon Ads Agent is designed to compress more of that chain into a connected workflow.

Amazon says its conversational experience can support tasks including audience sizing, reach forecasting, budget recommendations, campaign management, bid and budget recommendations, performance analysis and troubleshooting. In some workflows, advertisers can move from natural-language instructions into campaign execution rather than merely receiving an answer from an AI assistant.

That distinction matters. An AI that explains how to structure an advertising campaign is advisory software. An AI that converts a media plan into campaign structures, recommends targeting, adjusts optimisation parameters and helps deploy the result is becoming part of the operating system for the media budget.

What Is Actually Available to Australian Advertisers?

Amazon's announcements need to be read feature by feature. The Amazon Ads Agent brand now covers a broad advertising environment, but individual AI capabilities have different eligibility rules, beta stages and geographic availability.

Amazon Ads Agent conversational experience

• Status in October 2026: Closed beta for eligible advertisers
• Australian position: Australia is an included market, subject to account, capability and campaign eligibility
• What the advertiser still controls: Whether recommendations are applied, campaign parameters and account permissions

Agentic media planning

• Status in October 2026: Closed beta
• Australian position: Australia is included for selected advertisers, agencies and enterprise participants
• What the advertiser still controls: Objectives, final budget decisions and whether recommendations become an actual plan

Amazon DSP campaign creation from media plans

• Status in October 2026: Already expanded internationally
• Australian position: Available in Australia for supported Amazon DSP users
• What the advertiser still controls: Media plan inputs, review and approval before campaign launch

Natural-language analytics

• Status in October 2026: Closed beta
• Australian position: Australia is included for eligible hands-on-keyboard advertisers
• What the advertiser still controls: Interpretation, business decisions and resulting campaign changes

Full-Funnel Campaigns

• Status in October 2026: Open beta
• Australian position: Currently US-only
• What the advertiser still controls: Products, budget, creative inputs and optional performance thresholds

DVA+ Display, Video and Audio campaigns

• Status in October 2026: Extended closed beta in the US
• Australian position: International expansion is planned, but it is not yet a generally available Australian product
• What the advertiser still controls: Objectives, budget, creative approval and, with advanced settings, detailed targeting and delivery controls

This is the most important qualification for Australian readers. Amazon has already demonstrated Ads Agent to Australian advertisers and Australian DSP self-service users can access relevant agentic workflows, but the newest full-funnel and DVA+ experiences should not be treated as universally available Sydney campaign products simply because they sit under the same platform name.

Full-Funnel Campaigns Show Where Amazon Wants the Model to Go

Full-Funnel Campaigns provide the clearest picture of Amazon's longer-term advertising model.

Instead of running separate Sponsored Ads, display, video and streaming TV campaigns with separate optimisation logic, Amazon says an advertiser can provide products, a budget and an optional return-on-ad-spend threshold. Its AI can then select formats, allocate budget across stages of the funnel and continuously optimise toward sales growth.

In operational terms, the system is attempting to solve a problem that has historically been split between several teams: where should the next advertising dollar be placed?

A conventional campaign might allocate 20 per cent to awareness, 30 per cent to consideration and 50 per cent to performance activity because that was the media plan approved at the beginning of the month. An agentic campaign can potentially reconsider that allocation continuously using live campaign signals.

That creates a substantially different relationship between the advertiser and the platform. The human increasingly specifies the commercial envelope. The system increasingly determines how to operate inside it.

The Streaming TV Headline Needs an Australian Qualification

Streaming television is what makes Amazon's proposition particularly significant.

Automated search and sponsored-product optimisation are already familiar to performance marketers. Streaming TV has traditionally involved a different operating environment: larger creative assets, higher production costs, broader audience objectives and media-planning decisions that often sit well outside a retailer's Sponsored Ads team.

Amazon's Full-Funnel Campaigns attempt to connect those two ends of the funnel. A streaming TV impression can contribute data to the same campaign system that later influences lower-funnel activity.

But Australian advertisers should separate that strategic direction from current account access. Full-Funnel Campaigns are presently a US rollout. DVA+, which combines display, streaming TV, online video and audio into a simplified campaign, is also still in its US beta phase.

Amazon says DVA+ open beta is expected to begin in late October 2026, with international expansion planned through the fourth quarter of 2026 and general availability across 35 markets targeted for the first quarter of 2027.

Those are rollout targets rather than evidence that every Australian account already contains the functionality.

AI Optimisation Does Not Mean the Advertiser Loses Every Control

One of the more consequential parts of Amazon's design is that automation and manual buying are not presented as mutually exclusive operating models.

Within DVA+, Amazon describes a simplified AI-led path and a more detailed advanced settings path. Experienced advertisers can retain controls around targeting, deals, supply sources, frequency management, bidding strategies and other programmatic settings.

This produces three practical levels of campaign authority:

1. Human-directed automation: the advertiser defines the objective, products, creative and budget while AI handles much of the execution.
2. AI recommendations with human intervention: the system proposes audiences, budgets, bids or optimisation changes and a user decides whether to apply them.
3. Advanced manual control: experienced buyers directly configure targeting, inventory, deals, bidding or format-level settings where the campaign requires greater precision.

The significant change is therefore not that manual media buying disappears. It is that manual configuration becomes optional in a growing number of situations.

Creative Approval Becomes a More Important Control Point

Campaign automation becomes commercially more consequential when the platform can also generate the assets needed to extend a campaign into new formats.

Amazon says its Full-Funnel Campaign workflow can help create images and TV-ready video where additional assets are required. DVA+ is also designed to generate creative across supported formats using existing product information or a brand URL.

That removes a traditional bottleneck. A campaign no longer necessarily needs to wait for a separate production cycle before expanding into another media format.

It also makes approval more important, not less.

Amazon states that advertisers review and approve creatives before launch. Australian brands should treat that review as a genuine operational gate. AI-generated creative still needs to comply with brand standards, product accuracy, campaign terms, substantiation requirements and applicable Australian advertising obligations.

Automation can reduce the cost of producing another version of an advertisement. It does not transfer responsibility for the claim made in that advertisement.

The Real Change Is Budget Orchestration

The most strategically important part of Amazon Ads Agent may ultimately be neither the chat interface nor creative generation.

It is budget orchestration.

Digital advertising organisations have accumulated increasingly specialised campaign systems. Sponsored search, retail media, display, streaming television, online video, audio and remarketing can each have their own budgets, optimisers, reporting views and agencies.

Amazon's full-funnel model attempts to move the optimisation question upward.

Instead of asking, "What bid should this Sponsored Product keyword receive?", the system can increasingly ask, "Which combination of media is most likely to produce the advertiser's stated commercial result?"

That is a much larger decision.

For a Sydney retailer spending $20,000 a month, a poorly judged keyword adjustment has one scale of consequence. For a national brand shifting six figures between performance ads and streaming television, automated allocation becomes a material commercial-control issue.

Elyment has previously examined the same underlying boundary in Amazon's move toward AI agents with financial authority. Advertising presents a related problem: the budget may already be approved, but somebody still needs to decide how much discretion software receives inside that approved amount.

A Sydney Marketing Team Still Needs an Approval Architecture

The phrase "AI campaign optimisation" can make governance sound more complicated than it needs to be. In practice, a useful operating model separates reversible optimisation from decisions that materially change commercial exposure.

Audience recommendation

• Possible AI role: Analyse signals and propose audiences
• Sensible human control: Review exclusions, suitability and strategic fit

Bid optimisation

• Possible AI role: Adjust within approved campaign logic
• Sensible human control: Set limits and monitor abnormal movement

Budget allocation

• Possible AI role: Redistribute spend toward stronger opportunities
• Sensible human control: Define total budget, thresholds and escalation rules

Creative generation

• Possible AI role: Generate images, video or format variations
• Sensible human control: Approve claims, branding and final assets before launch

Campaign analysis

• Possible AI role: Identify trends and explain performance
• Sensible human control: Validate the business interpretation before major strategy changes

New channel activation

• Possible AI role: Recommend or configure additional media formats
• Sensible human control: Require approval where the spend profile or brand exposure materially changes

This is consistent with the broader control issue Elyment examined in designing shutdown and approval boundaries for AI agents. A marketing workflow does not necessarily need a manager to approve every bid adjustment. It does need a clear point at which automation must stop and seek authority.

Natural-Language Planning Could Change the Agency Workflow First

Agentic media planning may prove just as disruptive as campaign optimisation.

Amazon's closed-beta planning capability allows eligible advertisers to use conversational prompts to estimate audience size, forecast reach and receive budget-allocation recommendations. Australia is included among the participating markets.

That can collapse several manual planning steps into one exchange.

A Sydney marketing team could theoretically move from a brief such as "reach new customers aged 25 to 44 interested in premium home products, with a $60,000 budget" to a proposed reach model and channel allocation without manually assembling every intermediate report.

The value is speed. The risk is false precision.

A forecast remains a forecast. A recommendation generated from Amazon's advertising and audience signals does not know every commercial factor sitting outside the platform: margin pressure, retail stock, upcoming promotions, distribution constraints, competing media, regulatory sensitivities, internal brand priorities or the cost of servicing additional demand.

The planning function should therefore be viewed as a faster decision-support layer rather than a replacement for the commercial brief.

Amazon's Early Performance Figures Need Context

Amazon reports strong early results from Full-Funnel Campaign beta users, including higher long-term return on advertising spend and lower new-to-brand acquisition costs compared with the blended performance of independently configured formats.

Those results are commercially interesting, but they are Amazon internal beta data from a limited testing period. They should not be interpreted as a forecast of what an Australian advertiser will achieve once the product reaches this market.

Full-funnel automation changes campaign architecture, which means advertisers should evaluate more than headline ROAS. Useful questions include whether incremental customers were genuinely new, whether margin improved, whether total media efficiency changed, whether frequency remained acceptable and whether shifting budget between channels produced better commercial outcomes rather than simply moving attribution.

What Sydney Advertisers Should Test Before Scaling Agentic Media Buying

The safest way to evaluate Amazon Ads Agent is not to ask whether AI can run the entire advertising operation. It is to identify one workflow where automation can be measured against an existing process.

A practical pilot might begin with an established campaign or media plan, where the business already understands normal performance. Define the commercial objective, maximum spend, non-negotiable audience exclusions, brand rules and approval points before introducing additional automation.

Then compare the agent-supported workflow against the previous operating model: time required to build the campaign, configuration errors, speed of analysis, cost of creative production, optimisation frequency, campaign performance and the number of human interventions required.

The goal is not maximum automation. It is determining whether additional machine discretion improves the campaign without weakening control.

The same principle sits behind Elyment's analysis of why poorly designed automation can remain expensive even when AI itself becomes cheaper. Faster execution creates value only when the underlying workflow is sound.

Advertisers connecting AI systems to broader advertising tools should also understand where system actions occur outside the language model itself. Elyment has examined this separately in its analysis of why AI agent actions require controls beyond model-level guardrails.

Map the Approval Boundary Before AI Controls More of the Campaign

Sydney & NSW | AI Operations & Workflow Review

Review campaign objectives, budget authority, data inputs, creative approvals, optimisation rules and reporting workflows before agentic advertising becomes business-critical.

Request an Operational Review

The Next Test Is Not Whether AI Can Buy Media

Amazon has already answered the narrow technical question.

AI can participate in media planning, convert plans into campaign structures, recommend targeting, optimise bids and budgets, generate creative, analyse results and, in the newest workflows, coordinate advertising across several stages of the customer journey.

The more important business question is how much campaign authority advertisers are prepared to delegate.

For Sydney and NSW organisations, the immediate opportunity is real but uneven. Some conversational, planning, analytics and Amazon DSP capabilities are already accessible to eligible Australian users. Full-Funnel Campaigns remain a US product today. DVA+ is moving through its rollout schedule.

That distinction should shape procurement decisions, internal expectations and marketing strategies over the next several months.

Amazon Ads Agent is not yet an autonomous marketing department. It is becoming something more commercially important: an operating layer capable of translating advertiser intent into more of the decisions required to deploy media.

The competitive advantage will not necessarily belong to the organisation that gives the AI the most freedom.

It may belong to the organisation that defines the best objective, the cleanest data, the clearest spending limits and the right moments for a human to remain in control.

References

• Amazon's move toward AI agents with financial authority
• Designing shutdown and approval boundaries for AI agents
• Why poorly designed automation can remain expensive even when AI itself becomes cheaper
• Why AI agent actions require controls beyond model-level guardrails
• Elyment: Request an Operational Review