AI for online stores: is your platform's AI enough, or do you need a custom agent?

Most e-commerce platforms already ship an AI for online stores inside the admin panel. The question we hear most in conversations with store owners is whether that's enough, or whether a custom-built AI agent for online stores adds anything. This article answers with five concrete differences and with what the two do well together.

Article summary

• Your e-commerce platform's AI helps the store's team with tasks the team itself asks for.
• A custom agent works without anyone asking and leaves proposals and approvals waiting for your team.
• The ShopGuru agent talks to the store's customers on the website, on WhatsApp and by email, before and after the purchase.
• The same agent prepares product pages from the questions customers couldn't find answers to.
• A custom agent connects to carriers, invoicing, helpdesk and CRM, on any e-commerce platform.
• The store's team chooses, for each type of action, what the agent does on its own, what asks for approval and what it never does.
• A 30-minute call with an engineer is enough to see whether the store needs both or just one.

What your e-commerce platform's AI already does well

The AI that comes with the platform knows the store's data, sits in the admin panel and, on many platforms, is included in the subscription. For drafting a description, summarising the week's sales or setting up a discount, it's real help. The overlap with ShopGuru doesn't worry us: it confirms the need exists.

The limit is in how it works: platform AI responds to whoever uses it. It waits for someone on your team to open the panel, know what to ask for and review the result.

"Platform AI helps your team do the work. ShopGuru does part of the work, talks to your customers and connects to everything else."

ShopGuru team

Five differences between platform AI and a custom agent

The five differences below don't compare models. They compare what each one does in the store when nobody is watching.

A custom agent works without anyone asking

A tool assumes your team knows what to ask for and has time to ask. In a store with 3 people, that time rarely exists at 11:12pm, when the message about a late order arrives.

ShopGuru works on its own initiative. The agent answers customers within the rules you set and prepares proposals for product pages, translations or prices, which it leaves in the app waiting for approval. Implementation and monthly improvements are on us, with engineers who put their name to what they do.

A custom agent talks to the store's customers

Most platform AI tools are built for the store owner. They talk to the person running the store, not to the people buying from it.

The ShopGuru agent talks to customers, in pre-sale and after-sale: it helps find the right size, answers about shipping to the Azores and Madeira, says where the order is and handles the exchange request. It does this on the website, on WhatsApp and by email, in the same conversation thread.

When it doesn't have the answer, the agent says "I don't know. I'll ask." and hands the conversation over to your team, with a summary and a suggested action. It's the same principle we describe in after-hours customer service.

A custom agent learns from customers' questions

Platform AI writes from what your team gives it: the product name, three features, a tone. The result depends on what your team remembered to say.

The ShopGuru agent writes from what customers asked and couldn't find. If five people ask it whether a coat is good for rain, that answer becomes part of the product page proposal. We call this the Cycle, and it only exists because the same agent is listening to customers. We explain the method in product pages that answer questions.

"The agent listens to customers' questions and turns them into product pages."

ShopGuru team

A custom agent sees beyond the store

Platform AI mostly sees what's inside the platform. The store also lives in the carrier, in invoicing and in email.

ShopGuru connects to the Stack the store already uses, whatever the e-commerce platform. It can check a shipment's status with the carrier or find the right invoice to send to the customer; the specific integrations are confirmed store by store in the Diagnostic. The agent also tracks the prices of the competitors your team chooses and proposes suggested prices with the reasoning behind them.

A custom agent follows the store's rules

Every store has exceptions it took years to get right: returns accepted after the deadline for long-standing customers, refunds that always go through the same person, discounts never given on new products.

At ShopGuru, the store owner picks one of three levels for each type of action: does it on its own, asks for approval, or never does it. Above a limit in euros or as a percentage, the action switches to asking for approval, and no approval goes without a name and a time, such as "Approved by Marta · 9:12am". On day one, everything asks for approval; autonomy is given one action at a time.

"You don't change how you work. The agent works the way your store works."

ShopGuru team

How platform AI and ShopGuru work together

The choice is rarely one or the other. The table sums up who does what in a store that uses both.

When the platform opens up an integration, for example through the Model Context Protocol, an open standard for connecting AI applications to external systems, ShopGuru can use it.

When an AI agent makes sense for your online store

In conversations with store owners, we see three signs. The first: it's the owner who answers WhatsApp at eleven at night. The second: the same questions arrive every week and the product pages still don't answer them. The third: the store has rules that no off-the-shelf software respects.

If none of these signs are there, your platform's AI is probably enough for now. If they are, the next step is to book a call of 30 minutes with an engineer, who looks at the store and tells you what they would and wouldn't do. For those who already want the full plan, there's the Diagnostic.

"First, a 30-minute call."

ShopGuru team

Frequently asked questions

I already use my platform's AI. Why would I need ShopGuru?

Platform AI helps your team do the work your team asks for. ShopGuru works on its own initiative, talks to the store's customers and connects to the rest of the Stack. Tools like Shopify Sidekick and ShopGuru can coexist in the same store.

Does ShopGuru replace my e-commerce platform's AI?

It doesn't replace it or ask you to stop using it. They cover different parts of the work: one sits in your team's admin panel, the other looks after customers and prepares proposals.

Is it a chatbot?

No. A chatbot answers from a script or a list of questions. The ShopGuru agent answers with the store knowledge, follows the store owner's rules, says when it doesn't know and hands the conversation over to your team with a summary.

Sources and editorial review

Written by the ShopGuru team based on how the service is designed. The examples of stores, times and names are illustrative. Published and reviewed on 16/09/2026.