Content generation: The low-hanging fruit

Medical affairs teams spend a lot of their time generating different kinds of content, including regulatory writing, educational materials, and conference presentations. Although this content certainly needs to be reviewed by humans for accuracy, AI can be, and in many cases already is being, used to speed up the process of content generation.

“If someone today is a medical writer at Pfizer and they’re drafting peer-reviewed publications for Pfizer’s phase 3 clinical trials, the drafting of that publication is a total waste of time,” Katz said. “AI could draft that publication quicker, faster, better, and the person should be a reviewer instead of an author, basically. Their function changes. That’s a good example. Another example is someone that’s creating a poster for ASCO. Why are you formatting this thing? AI can create a much prettier poster in 10 seconds than you can ever create, and you should edit it instead of author it.”

MSLs themselves also need to generate a lot of content to do their work connecting with HCPs.

“Generating content is a large part of what we do,” said Malloy. “Patient lay summaries, manuscripts, you name it. We are trying to augment their capability to do exactly what they’ve been doing for years.”

Administrative support: A huge opportunity

Stewart mentioned that her early career as an MSL involved a lot of spreadsheets and detective work. But increasingly, that work can be done by AI systems. 

“It’s cut the administrative work down significantly, and we’re able to do things much faster and be more agile with that work,” Stewart said. “I mean, of course, what I’m finding is there’s always a need to validate the information that’s given to us when it’s generated through an AI model, but it’s such a time saver. We’re able to get way more done with fewer people on the team than we were when I first started.”

But administrative support goes beyond insights generation. AI can also help MSLs manage their schedules and responsibilities more efficiently.

H1’s whole business is built around creating a database that makes it easier for MSLs to find out everything they need to know about healthcare providers, from their areas of expertise to what conferences they’re planning to attend. Now they are building AI agents that sit on top of that database.

“For BioMarin, we created a digital MSL who basically every day monitors the latest HCP’s information, recommends what you should do with that HCP, it predicts what that HCP will do,” Katz said. “It can then automate all your emails that you want to send them, do follow-up meetings, monitor the latest scientific literature. You can have it draft or automatically send the emails. It just reduces everything that an MSL needs to do. It sits on top of an MSL. Each MSL is paired with a digital MSL, and boom, that’s what they do for you.”

Training: Improving MSL soft skills

While information management is a huge part of an MSL’s job, equally important are the soft skills of connecting to healthcare providers, gaining their trust, and building a rapport. This is the part of the job where the human element is the most important. But AI can still be helpful.

“Because they are so well-versed and they do that constant upkeep and research into their field, MSLs have an understanding of the science that’s really very solid,” Malloy said. “But it’s in that relationship management, conversational skill building that we noticed a real gap. That’s where the MSL Interact tool really comes into play. It’s filling that gap, that need for MSLs to feel confident.”

MSL Interact is a training platform developed by Inizio that leverages the ability of large language models to take on different personas as they interact with users.

“What we’re doing is taking generative AI, training it, giving it a persona, ingesting information that is both publicly available and that is client-specific. We’re targeting different therapeutic areas. We’re engaging clients. They’re giving us their specific information,” Malloy explained. “We’re able to create this HCP persona that maps to different types of doctors or healthcare providers that these MSLs may be meeting in the field. Then the MSLs can come into the tool and actually have a practice conversation.

Artificial intelligence is no longer an experiment in retail. Nearly half of retailers now use AI daily or several times per week, according to data from Amperity’s 2025 State of AI in Retail survey.

From customer data platforms to predictive models and chatbots, the technology is today being embedded into everyday operations, reshaping how brands engage with customers and compete in a crowded market.

“We have AI embedded across many parts of the business, which makes it feel seamless rather than experimental,” Daniel Chasle, chief data officer at U.K. fashion brand New Look, told Newsweek.

“For example, we use Amperity to run algorithmic stitching of customer profiles, AI chatbots in Zendesk to support customer service deflection, and AI coding assistants for our developers. We’ve also rolled out Microsoft Copilot to a subset of employees to help with daily tasks. Together, these tools are becoming part of the new ‘norm’ in how we work,” he said.

The normalization of AI reflects a turning point for the industry. Ninety-seven percent of retailers plan to either maintain or increase their AI spending this year, with priorities focused on personalization, media spend and demand forecasting. Loyalty and customer service are also key targets, as executives look to reduce costs and strengthen relationships at the same time.

“What retailers are really asking for are demonstrable outcomes,” Tony Owens, CEO of Amperity, told Newsweek. “They don’t want AI for AI’s sake—they want proof it drives growth. Every use case has to tie back to revenue, efficiency, or loyalty in ways you can measure.”

A Shift in Omnichannel Strategy

One of the biggest changes in 2025 is how retailers think about omnichannel, astrategy for giving customers a consistent shopping experience, whether in person, online or through mobile.

“Omnichannel 1.0 was about being where your customers are—stores, websites, apps,” Owens explained. “Omnichannel 2.0 is about the customer journey itself, and AI is what makes it possible to personalize those journeys in real time. The customer decides the channel, not the retailer, and they’re voting with their wallets.”

Retailers see the potential: 63 percent believe AI will help improve customer loyalty, while 65 percent expect it to increase customer lifetime value. But fewer than half—just 43 percent—are currently applying AI in customer-facing applications.

“Customers don’t think of themselves as segments or cohorts. They’re on a journey with your brand,” Owens said. “AI helps retailers meet them in that journey by anticipating needs, tailoring offers, and staying consistent across every channel. People know when a brand truly ‘gets’ them. That’s when the relationship shifts from transactional to personal, and that’s what drives loyalty and lifetime value.”

Still, adoption is uneven. While enthusiasm is high, retailers are cautious about pushing AI directly into customer touch points, often holding back because of costs, skills gaps and infrastructure challenges.

Solving the Data Puzzle

The survey highlights one major obstacle: 58 percent of retailers say their customer data is fragmented or incomplete. That fragmentation raises IT costs, delays decisions and complicates personalization.

“The challenges are the acquisition of the data from the disparate systems and knitting the data together to give a consistent view of the physical customer behind the data,” Chasle said. “The opportunities are to have the unified view of the customer, their shopping behaviors and preferences, to be able to understand all our touch points and interactions with the customer. This becomes an incredibly powerful data set that can power our decision-making and our engagement with customers.”

The New Look brand tackled the issue by combining an enterprise data platform with Amperity’s identity resolution. “Amperity also makes the data seamlessly available back into our data platform for our data science teams to access,” Chasle said.

That effort already has delivered results. New Look is using real-time customer profiles to fine-tune marketing campaigns and improve personalization. According to Owens, the unified data helped the brand identify nearly 26 percent more high-value customers than it had recognized before, insights that led to stronger offers and higher conversions.

Owens said it’s “proof that when you put the right data behind AI, you deliver a better journey for the customer and measurable return on customer data for the business.”

And the results are tangible. “The newly created Real-time Customer Profiles with Amperity are already fueling our paid media suppression activity, CRM optimization and will soon start to power a new wave of personalization experiences,” said Chasle.

Owens said that New Look’s example illustrates the potential benefits. “By using Amperity to unify customer profiles and power predictive models, they uncovered nearly 25 percent more high-value customers than they knew about before. That insight led to better offers, stronger conversions, and proof that when you put the right data behind AI, you deliver a better journey for the customer and measurable return on customer data for the business.”

But not every retailer has made this leap. The survey found that only 23 percent are currently using AI in production to resolve customer identities or prepare data for marketing, underscoring how widespread the data challenge remains.

From Experimentation to Embedding

For many retailers, AI adoption is moving beyond pilot projects. Nearly half are already using AI weekly, and those with customer data platforms are far ahead of their peers.

Organizations with a customer data platform (CDP) are twice as likely to use AI daily (60 percent vs. 29 percent) and more likely to have full adoption across multiple business units (22 percent vs. 10 percent).

“We don’t have the luxury of budget to experiment, and so we are approaching it on a value basis as part of our transformation roadmap and the prioritization of the business value and alignment to the overarching strategy,” Chasle said.

Owens said the distinction between experimenters and leaders is becoming clearer. “Experimenters usually see productivity gains, such as reduced costs, faster workflows, or incremental improvements. That’s valuable, and every retailer should be experimenting with AI right now. But leaders take it further. They embed AI into the way the business runs. That’s when you move beyond efficiency to true personalization at scale.”

That gap is likely to widen. As some retailers build AI into core operations, others risk being left behind, stuck in pilot mode without the confidence or resources to scale.

What Comes Next

Both Owens and Chasle pointed to personalization as the next big opportunity.

“Yes, the personalization of the web experience is in our immediate roadmap, with a vision of this leading to a personalized AI-stylist capability supporting our customers both in the digital and retail channels,” Chasle said.

Owens predicted that the next wave will be even more transformative. “By 2026, retailers will start to democratize data across the entire enterprise, using it to orchestrate the customer journey end-to-end. That’s when AI will deliver the full return on customer data.

“And that’s the moment of separation,” he continued. “The retailers who master this will win the bulk of customers in their category and set the standard for the next generation of brands. The ones who don’t will fall behind. This is a defining moment for retail. There will be winners and there will be losers.”

The findings echo broader consumer research, such as Cognizant’s recent survey showing that shoppers increasingly expect AI-powered personalization in their retail journeys. Taken together, the two reports show both sides of the AI revolution: consumers demanding seamless experiences and retailers racing to build the data foundations to deliver them.

Whether those predictions materialize depends on how quickly retailers overcome the same obstacles that have slowed AI before: siloed data, high costs and employee training.

The survey underscores the tension between ambition and readiness. While 97 percent of retailers are ramping up AI investment, only 11 percent feel strongly that they are prepared to deploy AI tools at scale. High costs, technical gaps and fragmented data remain persistent hurdles.

Still, the direction is clear. “Being able to tackle these business processes and re-imagine them with AI is the biggest opportunity,” Chasle said.”It is going to require significant business buy-in with senior stakeholder sponsorship, a clear end-state vision and a roadmap of activity that progressively tackles the required change.”

• Adobe launches six new AI agents across customer support, product and data
• Enterprises will “soon” be able to customize their own agents, too
• Most customers are already interacting with AI chatbots

Adobe has made its AI agent technology generally available in a bid to help workers build, deliver and optimize customer experiences and marketing campaigns.

The agentic platform promises context-aware and multi-step actions, response refinement and ROI-driven workflows thanks to its AEP Agent Orchestrator underpinnings, meaning that businesses can tailor the AI to their own unique goals.