Manifest AI
Shopify shopping assistant that trains itself on your catalogue and answers buyers in the store
Shopify shopping assistant that trains itself on your catalogue and answers buyers in the store
Fast-read signals for fit, pricing, and trust.
Best for a Shopify merchant whose catalogue is large enough that browsing is the problem — the assistant is trained on your own products and policies on install rather than on a knowledge base you have to write. Two things to establish before committing: the paid plans are not published anywhere a buyer can read them, so the price is a conversation; and anything beyond the storefront, such as WhatsApp or Instagram, is routed to a second product rather than included here.
There is nothing to author. Most storefront assistants ask you to build a knowledge base before they are useful; this one imports your products, collections, policy pages and articles when you install it and answers from them. For a merchant, that turns a project into an installation — and it also sets the ceiling, because an assistant trained on your catalogue is only as good as your catalogue. Thin descriptions and inconsistent attributes produce a thin assistant, and no amount of configuration fixes that.
Separate what it does with that training from ordinary support chat. The pre-built agents are aimed at the browsing shopper rather than the one with a problem: sizing recommendations from height, weight and fit preference, discovery quizzes, nudges before a drop-off, product comparisons drawn from the catalogue. These run on the product page itself, not only in a chat bubble, which is a different and more commercial intervention than answering tickets. Support is handled too, with unresolved conversations passing into Gorgias or Zendesk rather than sitting in a queue only this tool can see.
Installation is a Shopify App Store action, and the vendor says other platforms need its team, which makes them a services arrangement rather than a self-serve option. The channel is the other boundary to map before a trial: the vendor's own page routes anyone wanting the same AI on WhatsApp, Instagram or email to a second, separately branded product, so a plan that starts on the storefront and ends in a messaging strategy will involve more than this one purchase. Neither is a flaw; both change what you are agreeing to.
Quick fit check against how you actually work.
What you can actually do with this tool.
Install it and the store's own content becomes its training data, with no knowledge base to author first. That removes the usual weeks of preparation, and it means the assistant's quality follows the quality of your product data.
The product is a library of pre-built commerce agents with a no-code studio for building your own by setting a goal, success criteria and instructions. You are configuring specific jobs — sizing, discovery, recovery — rather than tuning one general assistant.
Quizzes, nudges and embedded widgets sit on the product page itself, aimed at the browser who has not asked a question. That is a different intervention from a support chatbot and the one most likely to affect conversion.
Unresolved conversations pass to human agents through Gorgias or Zendesk rather than into a queue only this tool can see. If you already run a helpdesk, that is what makes it adoptable without changing how support works.
Installation is a Shopify App Store action. The vendor says its team can integrate other platforms, which is a services conversation rather than a self-serve install — worth knowing if Shopify is not where you sell.
The listing names Klaviyo, Yotpo, Gorgias, Shopify Inbox and Checkout alongside OpenAI and ChatGPT, so reviews, email and checkout data are reachable rather than siloed from the assistant.
Tools and platforms this connects with.
Common questions about this tool, answered.
The training step is automatic: the app reads your catalogue and policy content when it is installed, so there is no knowledge base to write first. What does take work is deciding which agents to run and where — the sizing, quiz and nudge placements on product pages are configuration choices, and they are where the commercial effect comes from.
Not published. The Shopify listing states that a free plan is available and links to further pricing options without setting them out, and the vendor's own site carries no pricing page — its calls to action are to request an agent or book a demo. Expect to install the free tier to evaluate, and to negotiate the paid plan rather than compare it.
Only with the vendor's help. The documented installation path is the Shopify App Store, and for other platforms the vendor states that its team can integrate it. Treat that as a services engagement with its own timeline and cost rather than as a supported self-serve option.
The Shopify listing states that the AI replies in all languages on live chat, while the listing's own language field is English. For a store selling across markets that is worth testing directly on your own catalogue during the free tier, since answer quality in a second language depends on how your product data is written as much as on the model.