Agentic commerce has emerged as the new frontier in e-commerce, with some surveys showing that up to 30 per cent of shoppers already use AI to discover or buy products online. The shift is accelerating faster than most retailers expected – and it’s changing the way consumers move from discovery to checkout.
Where shoppers once visited a website, browsed a marketplace, or typed a query into a traditional search bar, the journey has splintered into dozens of micro-pathways. Often, those pathways no longer end on a brand’s own channels.
Platforms like Firmly AI are fueling that shift by embedding shoppable functionality inside search tools, AI chatbots, messaging platforms, content sites and social feeds. As Scott Hendrickson, CRO at Firmly AI, puts it, the company is focused on making “AI chatbots, search tools, messaging and social media shoppable,” turning moments of curiosity into “instant purchasing.”
It’s a direction that global analysts are watching closely. A recent Boston Consulting Group research feature warned retailers that AI shopping agents are already “autonomously influencing consumer purchasing decisions, and redefining online shopping interactions.”
BCG cautioned that retailers risk losing direct customer engagement and brand loyalty if they become increasingly dependent on third-party AI gateways, unless they build their own AI experiences and strengthen their operational infrastructure.
Hendrickson says the industry’s urgency is apparent. “The acceleration of the role that AI search plays is just one segment within AI commerce that has grown massively over the course of the past six months,” he says. “Put on your helmet, keep your eyes wide open: We are set to be living in a very fluid environment for a while.”
The pace of change and the danger of waiting
The speed at which agentic commerce is evolving is now one of retail’s biggest strategic tensions: adopting too early creates risk – but waiting may not be an option.
“No matter how quickly you jump in, the pace at which this world is evolving is so blistering that if you think of waiting three months or a month, you’re still playing catch-up,” Hendrickson says. “I’m not sure there is a catch-up. It is an ongoing evolution that we all need to jump on and help create together.”
Yet enthusiasm varies sharply between retail categories. Hendrickson is “super excited at the evolution of AI within commerce,” but says not every journey benefits from agentic shopping – at least, not today.
“When we think about the role that AI plays within commerce, you have to segment it,” he explains. For habitual purchases like groceries, restaurant orders, or weekly essentials, AI agents make immediate sense because “the buyer consideration is already done.” A trusted agent embedded within a retailer or delivery platform simply removes friction.
“But it’s a very different thing when you’re looking at travel or clothing – anything where there is consideration required,” he explains. “A lot of the conversations are coming through a technology-first lens. If you look from a consumer lens and ask what actually creates value, I’m not sure the efficiency is yet there in many use cases.”
His own behavior tracks that evolution. “How I use Perplexity or ChatGPT is very different today than it was six months ago, versus a year ago.”
Fear vs opportunity in the merchant response
Hendrickson says that while many retailers see the potential, others remain hesitant.
“There are a lot of conversations we have with merchants today where they’ve gone shy. They’re operating out of fear versus out of opportunity. That mindset needs to be switched. It’s okay to make some mistakes.
“The quicker merchants can get through that evaluation period and jump in, is ultimately what’s going to dictate success.”
Security is often the sticking point – and for good reason, he continues. “The security of transactions is non-negotiable.”
Notably, he argues, retailers don’t need to dismantle their infrastructure to adopt agentic commerce. Instead, they need to connect what they already have.
“One lesson we learned is that if you can connect the existing ecosystem with the existing relationships consumers have with a brand, it’s going to work,” he says. Whether the checkout is enabled on Best Buy, Aeropostale or Abercrombie content inside a social network or publisher, “there is trust. That trust is vital in e-commerce”.
Maintaining brand identity in an AI-mediated world
As AI agents increasingly mediate product discovery, retailers are raising a critical question: How do you protect brand voice, identity and customer relationships?
Hendrickson sees multiple layers to the answer. “It’s not just your brand identity. It’s also your customer relationships.” He points to TikTok Shop as an example of a powerful platform that ultimately owns the customer journey it facilitates.
“There is no lifetime value generated through those orders. That is inherently problematic, especially for large merchants. Everyone is cognizant of that.”
In contrast, tools like Firmly, are designed to keep merchants in control. “In our solution, the merchant is always the merchant of record… It’s the merchant’s content, the merchant’s product pages, the merchant’s customer.”
Not all retailers can adopt these systems at the same pace. “Marketplaces with complex back ends might find it harder… compared to those using a platform like Shopify whose APIs easily allow innovation.”
A future of niche, value-driven AI experiences
Despite the momentum behind AI, Hendrickson is cautious about assuming uniform adoption. “I’m not sure I’ll go to ChatGPT for my fashion advice,” he jokes. But he believes the future lies in specialized, consumer-centric agents.
“There’s a plethora of niche experiences that consumers will flock to in their own way to get the value they’re desiring – whether that be travel, fashion, raising children, or who knows what. I don’t think it’s a one-size-fits-all.”
At the end, the customer journey is no longer owned by any single channel, and agentic commerce is accelerating that shift. To Hendrickson, the priority for retailers now is not achieving perfection in the execution of agentic AI, but participation. The landscape is moving too fast for spectators.
