Blog Juliia.ai

Putting an AI bot on your eCommerce site is already has-been. Gulp, yes, I said it: nothing is simpler today than wiring a RAG on top of a product catalogue. A few weeks of project work, a PIM connector, and there you go: "we're good, we have an AI bot on the site. It answers our customers' questions on our products well." Box ticked, moving on.

Except an AI is not just a parrot, however knowledgeable it sounds.

You could say "it's still better than nothing." But even that old adage is up for debate. The real subject behind online customer experience isn't answering questions: it's accompanying a customer through their exploration, and above all through their purchase decision. That's why we have expert salespeople and customer advisors in the first place. Their added value isn't answering basic questions: it's building the confidence and reassurance that actually triggers the purchase. It's a craft.

The same is true for AI. The next competitive advantage in AI for eCommerce will come from re-humanising eCommerce, through advisory expertise.

A market that has settled the adoption question, not the trust one

This isn't just a gut feeling. Market data already backs it up in 2026:

71% of business and technology professionals report their organisation has already invested in chatbots for customer experience, and 64% of CX leaders plan to increase that investment further in 2026 (Master of Code, 2026).
Retail and eCommerce already account for 21% of the global conversational AI market, with retail/eCommerce chatbot spending projected to reach $72 billion by 2028 (Master of Code, 2026).
But adoption doesn't mean trust: 76% of consumers say they want an AI-powered shopping assistant, yet only 39% have actually used one to shop online, and 55% still say they distrust AI chatbots (Capital One Shopping Research, 2026).
Per commercetools (citing McKinsey), 62% of companies are already experimenting with AI agents, but only 23% have actually scaled them, and just 13% of consumers have completed a purchase off the back of an AI recommendation (commercetools, 2026).
Without real personalisation and advisory support, 62% of consumers say they're ready to abandon a brand after a too-generic experience (up from 45% a year earlier), while brands that genuinely personalise the experience capture up to 40% additional revenue (Hello Retail / McKinsey / Twilio Segment, 2026).

And the 2027-2028 forecasts don't point toward the "bot that answers questions" settling into the status quo. Quite the opposite:

Gartner predicts 40% of enterprise applications will feature task-specific AI agents by the end of 2026, up from less than 5% in 2025, and that by 2028, organisations running 80% of customer-facing processes through multi-agent AI will durably outperform competitors (Gartner, 2025).
AI agents could intermediate up to $15 trillion in B2B spend by 2028, with 90% of B2B purchases potentially handled by AI agents within three years (Gartner via Digital Commerce 360, 2025).
Forrester's 2026 predictions warn that the "agentic commerce race" is about to intensify, but that some initiatives will be, at best, learning experiences, and at worst genuine money pits for brands that jump in without a clear strategy (Forrester, 2026).
Morgan Stanley Research estimates agentic commerce could capture between $190 and $385 billion of US eCommerce spend by 2030 (a 10-20% market share) (Morgan Stanley Research, 2025).

The common thread across every one of these projections: the shift isn't toward more bots answering more questions. It's toward agents capable of advising, recommending with relevance, and reassuring: precisely the layer of value most of today's bots don't touch yet.

7 reasons we hear all the time, and that deserve a closer look

If you're a Head of eCommerce, chances are you've already heard, or said yourself, one or more of these 7 sentences to explain why your site isn't (re-)questioning its conversational AI strategy:

We already have one and it does the job.
We don't need one, our site already performs to its maximum potential as it is.
We don't have the budget to invest in additional gimmicks for our eCommerce.
We're unsure which one is the right one for us. There are many, they are all the same, and we have other priorities on our roadmap.
We don't trust AI to represent our brand and product expertise, and we don't have time to waste on testing it.
This is an IT project that requires proper governance, and our IT / DPO / Legal departments need time to fully assess risks before anything can be envisaged.
We'll just use ChatGPT or another LLM when the time comes.

All strong, practical, rational, and perfectly valid reasons. Until you look closer.

That's exactly the point of this series: dedicating one article to each of these 7 reasons, to explain why every single one of them is not only the wrong reason to hold back, but the right reason to start now. Do you recognise any of them? You're in the right place.

Next in the series: reason #1: "We already have one and it does the job."

Sources cited:

‍