Agentic commerce - AI agents autonomously guiding or even completing purchases - has been hyped as the future of retail. In practice, however, it’s proving far more complicated. AI is making discovery and product recommendations smarter, but full autonomous checkout has struggled to gain traction. OpenAI’s recent retreat from Instant Checkout, rerouting purchases through Instacart and Target, highlights the limits of large-scale AI shopping right now.
“AI will make it easier for consumers to buy, more conversationally based, but people buy from people, and we risk losing that human connection.”
Says Stuart Otter our Head of Technical
The promise is clear: AI agents can take over time-consuming tasks, from comparing products to organising purchases, offering shoppers a frictionless, hyper-personalised experience. But as Ben Wake points out,
“Most people use AI to get a collated overview. OpenAI hasn’t seen the uptake they expected - many users are still going straight to third-party sites to complete their purchases.”
Why It’s Struggling
Several barriers are slowing adoption. Trust and security are obvious issues - handing over payment details to AI raises fraud concerns. At the same time, merchant product data must be standardised and constantly updated. Pricing, stock levels, SKUs, GTINs - all of it needs to be accurate for AI agents to function reliably. Without this, shoppers risk being steered toward competitors with cleaner, more accessible data.
Adam warns of another challenge:
“As AI use increases in commerce, those who engage the most might come out on top, even if they’re not the best fit for the customer.”
Meanwhile, Charlie Archer highlights potential behavioural risks: over-reliance on AI could make shoppers less engaged in research, or create mismatched expectations when AI makes a decision that doesn’t align perfectly with the shopper’s intent.
Liliia Maliutina adds that retailers could lose first-party relationships if the entire journey happens inside an AI interface.
“It works if product metadata is perfect. Otherwise, AI agents may skip the store for a competitor that’s easier to read.”
Where It’s Working
Despite the challenges, AI is already delivering results. It’s most effective for discovery, comparison, and utility shopping – replenishing staples, finding replacement parts, or collating product options.
Liliia Maliutina notes,
“For zero-friction shopping, agentic commerce is a massive win”
Merchants with strong product data infrastructure – accurate SKUs, GTINs, UPCs - are best positioned to benefit. Chris Munday points out that unique or luxury products may not see the same advantage.
“It’s effectively another marketplace with guided purchasing assistance. Merchants with good merchandising will benefit most, while those with unique products may not.”
Partial automation, like price alerts, subscription reminders, and predictive replenishment, is already showing tangible results. Retailers like Walmart, Instacart, and Expedia are experimenting with pilot integrations that influence purchases while keeping final checkout secure.
What Retailers Should Do
The advice from our experts is consistent: focus on incremental adoption. Improve discovery, assist recommendations, and build trust with shoppers before attempting full autonomous checkout. Prepare your systems for multi-agent ecosystems and dual storefronts - one for human shoppers, one for bots - rather than assuming full autonomy will be accepted immediately.
Daniel Mulley says,
“The goal is a day when you can tell an AI exactly what you need and it produces the right results - that’s where agentic commerce can truly shine,”
Looking Ahead
Agentic commerce has enormous potential. AI agents could anticipate needs, manage inventory, and streamline everyday purchases. But for complex, emotional, or luxury shopping, humans remain essential. Retailers need to focus on data quality, trust, and designing experiences that work for both AI agents and human shoppers. Those who get the balance right will set themselves up to thrive as AI becomes a bigger part of the commerce landscape.
What do you think about agentic commerce? Should we focus on it or should we just ignore it like we did with 3D TVs...
If you need help with your eCommerce business, or just want to chat to see where we can help you grow, get in touch.
