Agentic AI front and centre

Three months into 2025, it's unmistakably the year software vendors scramble to own agentic AI. Adobe has joined the throng, using this week's Adobe Summit to proffer its first 10 AI assistants across marketing, commerce and content supply chain management for B2B and B2C marketers.

According to Capgemini and Statistica reports, the value of the Agentic AI market hit US$5.1 billion in 2024 and is projected to blow past $47bn by 2030. Agents are defined as intelligent AI systems that can autonomously plan and execute specific tasks, make informed data-based decisions and sit within workflows.

In Adobe’s case, these agents come in the form of the Adobe Experience Platform (AEP) Agent Orchestrator and 10 purpose-built agents across the Adobe Experience Platform designed to tackle common, cumbersome marketing and CX tasks. The initial 10 cover data insights, audience, site optimisation, content production, journey management, data engineering, product advice, account qualification, workflow optimisation and experimentation.

Highlighting Audience Agent as an example, Adobe claimed this agent will be an “ally” for audience building, enabling users to cut the process of creating custom audiences from 4-5 days to just 1-2 hours by using natural language prompting in the agent to build out audiences. Product Advisor Agent, meanwhile, is designed to personalise experiences based on user context and brand preferences, anchored in first-party data and brand content. Audience Agent analyses cross-channel engagement data to create and optimise goal-based, high-value audience segments, which can then be activated for personalised campaigns.

Adobe dubs it the “evolution of transactional chatbots and web-based agents”, that draw on a company’s brand attributes, customer data and third-party sources to give one-to-one personalisation a steroid injection.

In an explainer on the Adobe Summit main stage, Adobe SVP cloud experience engineer, Anjul Bhambhri, said agents are both intelligent and autonomous, multimodal and directly connected into all of Adobe’s core products

“You continue to engage through conversations, but they go beyond answering questions; they can do things for you. You have to tell  hem what you want – goals, constraints – then they get to, often in background, proactively making suggestions, filling in gaps,” she told summit attendees. “It turns out, Adobe Experience Platform (AEP) applications are a natural fit for such agents. Sure enough, we are tightly integrating them in all your applications. These will be surfaced via the AI assistant, so you get a unified view.”

Meanwhile Adobe has launched a new app built on AEP, called Adobe Brand Concierge, which means users can configure and manage AI agents for their end users as part of custom experience plays.

These connect directly through to AEP data which then feeds the agent.

As a B2B example, Adobe said Brand Concierge will go beyond providing general product information on a brand’s website to deliver tailored content based on the existing account relationship, while handling tasks such as booking follow-up meetings. The aim is to optimise sales by generating more leads and boosting conversion through sharper personalisation.

IDC analyst, Roger Beharry-Lall, welcomed Adobe’s strategy of setting up 10 specific agents first, in contrast to the marketplace to other vendors, and on use cases that are causing headaches every Monday morning for marketing teams. But he said users will still need to do a lot of the unsexy, foundational work to utilise them.

“Having the 10 gives you some place to start, and an anchor around what this is. The challenge is those 10 are big – they’re chunky, meaty agents. It’s not like you flip a switch and you’re magically using the Audience Agent,” he told Mi3. “On the back end, you still have to do the data cleansing, governance, integration – if you don’t have a CDP, you have to think through that; if you haven’t got your CRM paired up, you have to think through that.

“You have to think through workflows, data and the cleanliness, which is the anchor to AI success.”

Excited but sceptical: Marriott marketer 

Running multiple AI agents across marketing and CX workflow management is a long way from last year’s Adobe Summit, where chairman and CEO, Shantanu Narayen, was urging attendees to start embedding gen AI into everyday tasks to have a hope of keeping up with the sheer volume of content required to realise the demand for personalisation at scale.

Enter marketing’s fear, uncertainty and doubt.

The most succinct summation of unsaid marketing sentiment on day two came from Marriott VP and head of global marketing orchestration, Hilary Cook. She was given a round of applause for asking how many marketers in the crowd were “excited but also sceptical about your ability to implement what you’re actually seeing?”

“The question I ask myself every day – and I’m curious if you do the same – is: Are we as marketers and technologists truly ready to take advantage of these innovations? I’m going to say what we’re thinking, which is we're not,” Cook said. “But readiness is in our control and it’s far simpler than we thought to get to personalisation at scale.”

Marriott has 30 brands, over 9,500 hotels globally, 350 campaigns annually, 220 million loyalty members, and thousands of marketers and associates. Immediately, the question is one of tech, operational and process legacy for Cook.

“But the biggest barrier is not the technology, it’s us. Marriott in 2027 turns 100. Whatever challenges you’re facing in your org, I promise you for us, it’s the same,” Cook said. “For brands around for more than 40 years, many of our databases and capabilities are built on idea of sending direct mail. It was a single touchpoint, easy process, simple assets. The digital industry not only grown but exploded. We sit on the precipice of how to leverage AI.”

Two years ago, Cook’s team “took off rose coloured glasses” and found objective ways to assess readiness. “AI is not a silver bullet for fixing decades of debt. But it’s an incredible amplifier if you’re willing to do the right work. Otherwise, I believe it’s just a shiny object that will never get adopted at scale.”

Admitting she finds things “other people find boring sexy”, Cook then presented four inevitable required steps to get there. The first is documenting process flows.

“The one thing as you ask people in org about how you take something from intake to in-market, is no one tells you the same thing twice,” she said. “When you get a report that gives you data insights you then make business business decisions on, trace that data back to its original source.”

Also as you implement technology, document the original set of use cases you used. “When you go to add on to a technology… how you have chosen to implement from the beginning can be a massive barrier to how you scale out,” Cook warned.

When Marriott started this process, it was taking 110 days from intake to in-market through 395 steps and 45 disparate processes, all interdependent upon each other, even if the 45 different teams behind them didn’t know it. In all, 43 artefacts, documents, 41 of which were duplicated in some manner.

“We chuckle because everyone recognises it in themselves,” Cook commented. “There is a real power in being able to name and quantify your pain points. By doing this, it gave us a map to take to leadership and come up with a game plan.”

Having gained exec alignment and unified data, Marriott could stand up a robust CDP and single view of customer across five regions. With a centralised but federated data approach, Marriott gained strong foundations to mature all marketing activation opportunities under its internal acronym, MAPA, or marketing and personalisation accelerator.

From there it was critical to take these marketing capabilities and match them with real use cases marketers were trying to drive across audience and data activation, content acceleration, orchestration and decisioning.

“When you build new purpose and partnership with marketers, experimentation fuels capability building and you build value while ensuring long-term success,” Cook said. “As we start to see that success, build more use cases into it, until enough scale to say it should be true for everything. That way we stare and step our way to maturity.”

Marriott’s welcome journey was historically three touchpoints; now it’s over 17 and across 100 days. Targeting, which Marriott has had since 2017, has been trimmed down from 45 processes to one, thanks to a unified view of the customer and set of guardrails. As a result, updating content or offers for an existing campaign in-market is 92.5 per cent faster.

“This means my marketers can actually prove their destiny. If the campaign is not working, they have a chance in real time to update that content,” Cook said. Marriott now has more than 500,000 variations of its welcome journey, and is on its way to 2 million. And it has tens of millions of content variations, all while achieving a 70 per cent reduction in time to market. This has led to a whopping 4,200x growth in personalised growth.

“We’re six times ahead of our incremental revenue goal, but because we did pilot scale approach and became more mature, we also started to accelerate. I went from in the last three months of the year from not knowing if we’re going to hit it, to exceeding it by six times,” Cook said.  

“We are all here because we want to be on the cutting-edge of innovation and make experiences better. I believe the future isn’t happening to us, I believe it’s ours to build.

“And remember: Hard work doesn’t have to be complicated, and taxonomy is sexy.”

Adobe CMO wants to reinvent campaign ideation

Another marketer who knows all about tackling the overwhelming task of producing enough intelligent content and creative fast enough to achieve personalisation at scale and in as real-time a manner as possible – across as many channels as possible – is Adobe CMO and EVP, global marketing, Lara Balazs. She recently joined the vendor after a career in marketing across lifestyle, fintech, B2B and B2C, and described innovation as fuelling creativity and marketing coming together to make an impact.

Within the same breath, Balasz admitted marketers are overwhelmed by the task of trying to produce enough content to feed this modern marketing beast. Studies suggest creatives alone are spending 21 hours per week on repetitive tasks that could be automated.

“I know this job is not easy – being a marketer is not easy today. Marketing has transformed at a velocity we never thought possible, generally driven by gen AI, and a lot of tech coming to forefront. Along with this, we have social media apps, mobile apps, digital platforms beyond what we ever expected,” the CMO said. “That’s great for us as consumers… but breaking through Is harder than ever before. You have to find meaningful ways with creative that stand apart. It is a mandate. Personalisation at scale is a mandate – marketers know this, and that if they don’t get onboard they will get left behind.

“We know the companies that adapt, that adopt the tech, they will [grasp] possibilities and will scale further and faster than they thought possible.”

Adobe has experienced this firsthand, with in-house creatives working with marketers, agencies, and in teams. “We struggle to keep pace with the volume, variety and quality of content required to fill today’s digital channels,” Balazs continued. “We have gone through the long process to create the content through the brief, then we ideate, collaborate, review, and can take weeks, sometimes months to land. We know that it is just not acceptable in today’s environment. We said at Adobe it’s not good, we’re going take it as challenge.”

It’s led Adobe to a new project, dubbed Project Catalyse, and a new campaign ideation tool debuted at Adobe Summit. To do this, Adobe showed the process behind developing its new ‘Fantastic Voyage’ campaign for Photoshop.

In Adobe’s proposed ideation approach, high-performing social posts, content and insights, plus details on the target audience, are fed into a collaborative workspace where AI suggests formats and approaches to drive campaign ideation, including a mood board for brainstorming, pulling in brand approved assets and generating new brand-approved assets aligned with the campaign direction. Within the tool, a brief is also generated,

In future, Balazs expected the process to be done in the future in “record speed” with this kind of creative ideation in play.

“Our vision is to create a digital ideation space and tool that can reinvent campaign ideation. We believe this could be a game changer for companies,” said Balazs. “You can see with these tools that this could take days, even less than a day.

“We’re excited by the possibility of this tool... we’re keen to get your feedback.”

Adobe CEO: Acceleration, but safety first

In his keynote, Narayen’s message was AI is “accelerating and expanding the creative aperture for everyone, opening the content floodgates, tapping into everyone’s imaginations, and massively expanding the number of creative assets being designed and delivered”. It’s also bringing forward better ways of delivery and activation of campaigns, as well as analytics and insights for the entire content lifecycle.

“In the era of AI, we’re even more focused on enabling you to truly unify creativity, marketing and generative AI so you can deliver personalised, conversational experiences at a global scale,” said Narayan. “Adobe’s differentiated approach to AI is rooted in the belief creativity is uniquely human trait. But AI has the power to amplify and assist human ingenuity, to enhance productivity.

“We’ve made AI improvements across all four layers of Adobe’s AI stack spanning data, agents, models and interfaces, because we believe it’s the only way to deliver AI in a way that enhances your work.”

Narayen was among a throng of speakers also using terms like 'brand safe', 'commercially compliant', 'transparent', 'trustworthy' and 'enterprise-grade' as Adobe sought to reassure attendees that its expanding use of AI, generative AI, data, creative and marketing tooling is all safe and sound for blue chips.

Open systems, partnerships and the Microsoft alliance

Adobe isn’t stopping at its own front gate on AI, however. In recognition of the need for customised capabilities that address specific verticals, the vendor has struck new and deeper strategic partnerships with Acxiom, Amazon Web Services, Genesys, IBM, Microsoft, RainFocus, SAP, ServiceNow and Workday to enable seamless execution of use cases across agents for customer service, enterprise resource planning, human resources, collaboration and productivity, and data management.

Adobe also announced expanded agency and system integrator partnerships with Accenture, Deloitte Digital, EY and IBM to drive customisation across industries and use cases.

All of these partnerships are a mechanism to connect what marketing and CX teams are doing to the wider enterprise too, none more so than the ongoing strategic partnership with Microsoft. In what’s now a seven-year integration and partnership, Adobe said it’s going deeper on integrating Microsoft 365 Copilot with Adobe Express Agent. This includes activating Adobe Marketing Agent in Microsoft Teams, Microsoft PowerPoint and Microsoft Word applications through Microsoft 365 Copilot.

For example, Adobe said businesses can refine audiences for targeting by accessing data and insights from Adobe Experience Platform, unlocking operational insights and completing more complex analysis tasks and supporting audiences creation for widescale personalisation campaigns. Adobe Marketing Agent simplifies the audience analysis process by quickly surfacing insights in Microsoft 365 apps through Copilot. For example, marketers can prompt Adobe Marketing Agent to instantly uncover and pull insights from Adobe Customer Journey Analytics directly within Copilot, or create campaign performance reports in Microsoft apps, such as PowerPoint and Word.

It was only last year Adobe confirmed work with Microsoft to bring together Adobe Experience Cloud workflows and insights into Microsoft’s Copilot for Microsoft 365 and break down application silos. In a Microsoft survey referenced last year, 43 per cent of marketing and communications professionals reported having to switch between digital programs and applications and said it’s disrupting creativity.

For Constellation Research’s Liz Miller, it’s clear Adobe is placing its bet on open ecosystems. “All the infrastructure, apps, data – all of this comes together and this is how,” she commented as things kicked off at Adobe Summit. “The foundation, schemas, measures, intention in reconfiguring, re-innovating what we wanted to be as marketers when we grow up is coming to a head. We all know how hard the marketing job is.”

More broadly, Adobe has struck a new partnership with Amazon Web Services and Amazon Ads. The strategic collaboration builds on the availability of Adobe Experience Platform (AEP) on AWS and ushers in new integrations with AWS's generative AI services, Amazon Connect and Amazon Ads. For one, the companies will integrate AEP with Amazon Connect to provide end-to-end visibility into the customer journey while protecting customer privacy by combining AEP's customer profiles with Amazon Connect's customer care insights. Adobe also plans to leverage Amazon Q in Connect and Adobe's AEP AI Assistant alongside AI Agents, so businesses can deploy tailored experiences based on unified customer data.

Holdcos are another spoke in Adobe’s partnership wheel. Under a new partnership with Publicis, Adobe will build solutions that combine Publicis Groupe’s delivery capabilities across its intelligent content operations with Adobe Firefly Services directly into Publicis Groupe’s Core AI platform.

AI threaded through the content supply chain

Among the slew of other AI-powered capabilities announced at this week’s summit were Adobe GenStudio, Firefly Services and Custom Model advancements. These were all positioned as further ways to optimise the content supply chain for businesses, remove bottlenecks across content management, content production and improving workflows and collaboration while retaining brand safety and compliance.

Adobe is also expanding its generative AI platform, integrating foundational models, including commercially safe AI through Adobe Firefly to scale content production that powers customer experience management. Adobe Firefly is a collection of creative and generative APIs and services for enterprises.

According to Adobe, using this combination of tools with its Workfront solution has led T-Mobile to chalk up a 275 per cent increase in the number of marketing campaign projects the telco executes each month, plus a 10x reduction in campaign production time and 30x improvement in email open rates.

Adobe GenStudio for Performance Marketing was another innovation welcomed by marketers. It's aimed at further enabling businesses to scale content production by automating resizing and replication: Adobe cited Japanese ad holdco Dentsu achieving 70 per cent faster asset delivery across 145 markets.