Walk into any adtech conference in 2026 and you will hear the same pitch on repeat: AI is going to revolutionise creative for digital out-of-home. Generative models will spin up thousands of ad variants on demand. Computer vision will read the street, the weather, the mood of the crowd. Machine learning will pick the perfect frame for the perfect moment. It is a compelling story, and parts of it are already true. But a lot of it still belongs in the demo reel.
At Lumos we operate one of the larger programmatic DOOH platforms in Australia and New Zealand, and we have spent the past 18 months pressure-testing AI-powered creative optimisation across live campaigns. This post separates what is genuinely working from what is still vapour, so media planners and brand teams can invest in the right places.
What AI-powered creative optimisation actually means in DOOH
The phrase gets stretched in every direction. In its most useful form, AI-powered creative optimisation in DOOH refers to systems that automatically adapt what appears on a digital screen, when, and to whom — using data signals such as time, weather, location, audience composition, and real-time performance. The simplest version is dynamic creative optimisation, or DCO, which has existed for years. The newer version layers in machine learning, computer vision, and generative AI to pick or even produce the creative that is most likely to drive the campaign outcome.
According to IAB Australia's 2026 Outlook, 71 per cent of media agencies plan to increase investment in AI-driven creative tools this year, up from 48 per cent in 2025. The appetite is clearly there. The question is which capabilities are mature enough to bet a campaign on.
What's real: dynamic creative tied to live signals
The most mature use case is rules-based and ML-assisted dynamic creative tied to environmental and audience signals. This is no longer experimental. It is shipping at scale across major DOOH networks in Sydney, Melbourne, Auckland and beyond.
Weather-triggered creative — a beverage brand running 'Stay cool' frames when the temperature passes 28°C and 'Warm up' frames when it drops below 14°C
Time-of-day adaptation — quick service restaurants serving breakfast, lunch and dinner messaging on the same screen, automatically
Audience-indexed rotation — using anonymised mobility and panel data to lean into creative variants that index higher for the audience present at a given location and hour
Stock and inventory-aware messaging — retailers updating product visuals or pricing in near real time as supply changes
Sports and event reactivity — sponsors triggering 'we won' or 'live now' frames the moment a result lands
These workflows are no longer cutting edge — they are table stakes for serious programmatic DOOH campaigns. The AI here is mostly about the orchestration layer: making sure the right asset reaches the right screen at the right second, and learning which combinations actually move the needle.
What's becoming real: generative AI for asset production
Generative AI for ad creative is the area moving fastest. Tools like Adobe Firefly, Runway, and a wave of purpose-built ad creative platforms are now being used to produce DOOH variants at a pace that was unthinkable two years ago. Industry research from WARC in early 2026 found that brands using generative AI for creative production reported an average 38 per cent reduction in turnaround time and a 24 per cent reduction in production cost across digital channels.
In DOOH specifically, generative AI is genuinely useful for three jobs: scaling localised variants (one master concept, fifty suburb-specific versions), producing rapid weather and time-of-day adaptations, and creating the long tail of audience-indexed creative that human studios cannot economically produce. We are seeing FMCG brands move from four creative variants per campaign to forty without a corresponding blow-out in studio cost.
Generative AI hasn't replaced the creative team — it's freed them up. The big idea still comes from humans. AI just lets us run that idea at a level of personalisation we used to only dream about. — Eric Fan, CEO, Lumos
What's still hype: 'fully autonomous' creative AI
The pitch where an AI watches a screen, reads the crowd, generates a brand-new ad in real time and serves it on the spot — that one is mostly demoware. There are real reasons it is not ready for serious campaigns yet.
Brand safety — generative models still produce off-brand or off-tone variants frequently enough that human approval is required before public placement
Legal and compliance — Australia's AANA and ACMA frameworks, plus emerging AI disclosure rules, mean uncontrolled generative output is a serious risk surface
Quality at large format — 4K and street-format DOOH placements expose AI artefacts that look fine on a phone screen and terrible on a digital billboard
IP and rights clearance — talent, music and image rights are not magically resolved by a generative model
The brands getting this right treat generative AI as an accelerant for human-led creative, not a replacement for it. Every variant still passes through a brand and legal QA gate before it ever lights up a screen.
How to invest in AI creative without buying the hype
If you are a brand or agency planning a DOOH investment in the back half of 2026, here is the honest scorecard for where AI creative dollars are well spent and where they are not.
Spend with confidence — DCO tied to weather, time, location and audience signals; generative AI for variant production with human QA; ML-driven creative rotation that learns from in-flight performance
Spend cautiously — generative AI for headline-only adaptation (it works, but the lift over a good copywriter is small); AI-driven copy translation across markets (still needs native review)
Avoid for now — fully autonomous real-time creative generation; AI 'mood detection' targeting; anything that promises to remove the creative team from the loop
The best programmatic DOOH partners will be transparent about which capabilities are production-grade and which are experimental. If a vendor cannot show you the human-in-the-loop steps in their AI creative workflow, that is a signal to push harder on questions before signing.
The Lumos take
AI is genuinely changing how creative gets produced and deployed in DOOH, and the brands that move first on the mature use cases will compound an advantage. Dynamic, signal-driven creative paired with generative variant production is already delivering measurable lifts in attention, recall and short-term sales for our advertisers. But the leap from 'AI assists the creative team' to 'AI is the creative team' is much further off than the conference circuit suggests.
If you are exploring AI-powered creative optimisation for an upcoming DOOH campaign, the team at Lumos is happy to talk through what is working in market right now — including live case studies from FMCG, retail and QSR. Visit spotlumos.com or get in touch with our commercial team to start a conversation.
[SCHEMA_ORG]{"@context":"https://schema.org","@type":"Article","headline":"AI-Powered Creative Optimisation in DOOH: What's Real, What's Hype","description":"AI-Powered Creative Optimisation in DOOH: What's Real, What's Hype","image":"https://images.unsplash.com/photo-1639762681485-074b7f938ba0?w=1200&q=80&fm=jpg","datePublished":"2026-05-08","dateModified":"2026-05-08","author":{"@type":"Organization","name":"Lumos","url":"https://spotlumos.com"},"publisher":{"@type":"Organization","name":"Lumos","url":"https://spotlumos.com","logo":{"@type":"ImageObject","url":"https://spotlumos.com/logo.png"}},"mainEntityOfPage":{"@type":"WebPage","@id":"https://spotlumos.com/au/resources/ai-powered-creative-optimisation-dooh-real-vs-hype"},"keywords":"programmatic DOOH, digital out-of-home, audience intelligence, media planning","about":{"@type":"Thing","name":"Programmatic DOOH Advertising"},"isPartOf":{"@type":"WebSite","name":"Lumos","url":"https://spotlumos.com"}}
