Shopify Knowledge Base for AI Channels

As part of its Fall 2026 Edition, Shopify has moved the Knowledge Base for AI shopping agents from a standalone app directly into the Shopify admin. This change centralizes how merchants control the official facts, like shipping policies and return windows, that AI assistants such as ChatGPT, Gemini, and Meta Muse use to answer customer questions. For merchants, this is a critical tool for managing how their brand is represented in the growing world of agentic commerce.

The new system automatically drafts facts from your public-facing store pages, but nothing is shared with AI agents until you review and approve it. This guide walks through how to navigate the new built-in Knowledge Base, what to do before the old app retires, and how to ensure your store's information is accurate for AI-driven discovery.

What is the Knowledge base under Sales channels > Agentic?

The Knowledge Base is a central page in your Shopify admin where you manage the official facts that AI shopping assistants use to learn about your store. It replaces the old, separate Knowledge Base app and is now located at Sales channels > Agentic > Knowledge base.

This move is part of Shopify's broader strategy for "agentic commerce, " where AI agents discover, recommend, and even purchase products on behalf of customers. The Knowledge Base acts as a verified source of truth for all non-product information about your business. Think of it as a briefing document you hand to an AI before it talks to your potential customers.

A verified source is crucial because the cost of AI error is high. If an agent incorrectly tells a customer you offer free international shipping, you bear the cost. You either pay for the shipping to honor the promise or deal with an angry customer who was misled. The Knowledge Base prevents this by providing facts you have personally approved.

It is important to distinguish this from your product feed. A product feed contains structured data about what you sell: price, SKU, inventory, and images. The Knowledge Base contains policy and operational data about how you sell: shipping rules, return windows, and contact methods. Both are essential for AI, but they serve different purposes.

Shopify automatically scans your public store pages and uses that content to create "draft" facts. These drafts are not visible to any AI until you personally review and publish them, giving you complete control over what information is shared.

What business details, policies and FAQs does it hold for AI channels?

The Knowledge Base holds facts scanned from your store's public pages across several key categories. This includes your shipping and return policies, contact details, product availability, and other frequently asked questions.

When you navigate to the Knowledge Base in your admin, you will see information sorted into categories like:

• Store Details (contact info, business name)
• Shipping & Delivery
• Returns & Refunds
• Product & Service Information
• General FAQs

Each fact is tied directly to the source page on your website where Shopify's crawlers found it, making it easy to trace where the information came from. This structured set of data is what AI agents query to provide accurate, real-time answers to shopper questions.

A worked example: a shipping policy fact

Imagine your shipping policy page states, "We offer free standard shipping on all domestic orders over $75." The Shopify scanner will parse this and create a draft fact. The fact might read: "Free shipping is available for orders over $75." It will be categorized under "Shipping & Delivery" and linked directly to your policy page.

This single, atomic fact is powerful. An AI agent does not need to read and interpret your entire page, with its exceptions and extra details. It can query for the "free shipping threshold" and get a direct, verified answer. This reduces the chance of misinterpretation and provides a faster, more accurate response to the end customer.

But automated scanning has limits. Consider an edge case where your policy says "Free shipping over $75, excluding Alaska and Hawaii." The scanner might only pick up the first part of that rule. It could create a draft fact that is technically true but dangerously incomplete. This is why human review is a mandatory step in the process.

In this scenario, you would not approve the fact as-is. You would use the manual edit function to amend the fact to be more precise, for example: "Free shipping is available for orders over $75 within the contiguous United States." You must fix the nuance the machine missed.

How do you review the scanned facts?

You can review all scanned facts by navigating to Sales channels > Agentic > Knowledge base in your Shopify admin. This page presents a list of all the facts Shopify has drafted from your storefront, organized by category.

The first concrete step for any merchant is to perform a "trust but verify" pass. Assume the scanner did a decent job, but look for the details it might have misunderstood. Your role is not just to approve correct information but to catch and fix subtly wrong information before it becomes a customer service problem.

The process is designed for clarity. Each drafted fact is displayed alongside the content and a link to the exact page on your store where the information was found. This allows you to quickly verify the accuracy and context of each piece of data before deciding whether to publish it.

Consider the cost of not reviewing carefully. If you approve an old, scanned fact about a 30-day return window when your policy is now 14 days, you will be forced to honor the outdated policy for any customer who was informed by an AI. This creates financial loss and operational headaches for your support team.

The speed vs. accuracy trade-off

When you first open the Knowledge Base, you may face dozens or hundreds of draft facts. This presents a trade-off between speed and accuracy. You could quickly approve everything to get your store represented on AI channels, but you risk publishing errors. Or you could meticulously review every fact, which takes time.

A practical approach is to prioritize. First, review and approve facts in the most critical categories: Returns & Refunds and Shipping & Delivery. These have the biggest financial and customer experience impact. Then, work your way through less critical categories like General FAQs. This balances immediate presence with long-term accuracy.

Shopify's Help Center documentation provides a detailed walkthrough of the interface, showing how to view both the facts and their sources. It is a good practice to regularly check this section, especially after updating your store's policy pages, to ensure the drafted facts reflect your current business operations.

How do you approve and publish, or reject, a fact?

For each scanned fact, you have the option to "Approve & publish" to make it available to AI agents, or "Reject" to discard it. You can also edit information manually if the scanned text isn't quite right.

Here are the steps to manage your facts:

1. Go to Sales channels > Agentic > Knowledge base in your Shopify admin.
2. Find the category and the specific fact you want to manage.
3. Review the drafted content. If it is accurate, click Approve & publish. The fact is now live for AI agents to query.
4. If a draft is incorrect or irrelevant, you can reject it. Shopify may ask for a reason to help improve its scanning process.
5. If a draft is mostly correct but needs a small change, you can switch the source to "Manual" and edit the text directly before publishing.

When to edit manually

Manual editing is for adding the precision a machine scanner lacks. The scanner might draft a fact like, "We ship on business days." You should edit this for clarity to, "Orders ship Monday through Friday, excluding federal holidays." This specificity prevents customer confusion around holiday shipping times.

You should also edit to remove marketing language. A scanner might pull "Our amazing 30-day happiness guarantee." from your homepage. A better, more factual version would be, "A 30-day return policy is offered on all products." AI agents need plain facts, not sales copy. Use manual editing to translate your marketing into their language.

Rejecting a fact is for information that is completely wrong or no longer relevant. For instance, if the scanner creates a fact about a "Black Friday 2025 Sale" from an old blog post, you should reject it. Editing it would not make sense. When you reject, Shopify may prompt for a reason; choosing "This information is outdated" helps train the scanner.

This review-and-approve system ensures that you have final say over every piece of information that represents your brand in AI-driven conversations. Nothing is published without your explicit consent.

The line between a customer's question and a machine's query is gone. Your store must speak a language both can understand: clear, consistent, and provably true. The Knowledge Base is where you teach it to speak.

What happens to the old Knowledge Base app on 15 Jan 2027?

The standalone Knowledge Base app will be officially retired and will no longer be available after January 15, 2027. The app is already unavailable for new installations from the Shopify App Store.

For most merchants, this transition will be seamless. If your store had 1,000 or fewer facts in the old app, Shopify has already migrated those facts to the new in-admin Knowledge Base and published them automatically. No action is required on your part, though it is a good idea to review the migrated facts for accuracy.

If you never installed or used the old app, you can disregard any mention of migration. Your journey begins with a clean slate in the new built-in tool. Navigate to the Knowledge Base in your admin to see what facts Shopify has already drafted for you from your public site pages. Your task is simply to review and approve them.

The cost of inaction for those who used the old app is significant. If you fail to migrate necessary facts or engage with the new system, your store's presence on AI channels could degrade. After the retirement date, any facts residing only in the old app will no longer be served, effectively making your store invisible to certain AI queries.

What should a store with more than 1,000 facts do?

Stores with more than 1,000 facts in the old app must manually move any facts beyond the first 1,000 into the new Knowledge Base. The automatic migration did not apply to these larger accounts.

Shopify has directly emailed affected merchants with instructions. If your store falls into this category, you will need to add the facts you want to keep to the new system before the January 15, 2027 deadline. After this date, any information left only in the old app may become inaccessible.

A practical migration strategy

For these large stores, a concrete plan is essential. Before the deadline, log into the old app and take screenshots or export your list of facts. You need a record of what you have. Then, triage this list into three categories: "Must Keep, " "Nice to Have, " and "Outdated." This is your governance step.

Focus your energy on the "Must Keep" list. For example, if you have 1,600 facts, the first 1,000 were migrated. You need to address the other 600. After triaging, you might find only 200 of them are truly critical. Manually recreate those 200 in the new system first. This is more manageable than trying to move everything.

This manual migration presents an opportunity for content governance. It is a chance to review which facts are most critical for your business and ensure that the information presented to AI agents is prioritized, accurate, and up-to-date.

This process highlights the trade-off between completeness and clarity. The 1,000-fact limit in the new system encourages you to be concise. It forces a shift in mindset from documenting everything possible to curating only the most vital information. This constraint is a feature, pushing brands toward a clearer, more focused AI presence.

Ensuring your store's policies are clear for AI is the new frontier of customer service. But clarity before the sale doesn't stop questions after. A customer who understands your return policy perfectly might still order the wrong size. When that happens, the best experience isn't a support ticket, it is self-service. Tacey's order editing tools work on the order status page, after checkout, letting customers resolve their own issues, like changing a size or quantity, before the order ships. You control the rules, they control the fix.

Frequently asked questions

Can I add my own facts to the Knowledge Base?

Yes. In addition to approving Shopify's drafted facts, you can manually create new facts. This is useful for information that customers frequently ask about but which doesn't have a dedicated page on your site. For example, you could add a fact about your brand's sustainability practices or ethical sourcing policies.

This proactive approach can reduce support tickets. By adding facts that answer common questions you see in your customer service inbox, you allow AI agents to handle those queries before they ever reach a human. The new Knowledge Base supports up to 1,000 additional facts per store.

Does the Knowledge Base affect my store's SEO?

The Shopify Knowledge Base is distinct from traditional SEO structured data. While both use structured data, they serve different goals. SEO data, like Schema.org markup, helps search engines create rich snippets in search results. It is about presentation and ranking in a list of links.

The Knowledge Base, in contrast, is for conversation. It feeds AI agents the specific answers they need to respond to a direct user question like, "Does this store ship to Australia?" While one does not directly affect the other, maintaining accuracy in both is part of a healthy digital presence.

What happens if I do not approve any facts?

If you don't publish any facts, AI agents will have no verified information about your store. When a customer asks an AI a question, it will likely respond with, "I'm sorry, I cannot find that information." This creates a dead end for the shopper and a lost opportunity for you.

In a worst-case scenario, the AI might try to guess based on outdated or third-party information, leading to incorrect answers. The immediate cost is a poor customer experience, which can push a potential buyer to a competitor whose policies are clear and accessible to the same AI.

Is the new Knowledge Base free to use?

Yes, the integrated Knowledge Base is a native feature of the Shopify admin and does not have an additional cost. It replaces the previous free-to-install standalone app. It is included as part of your core Shopify subscription, reflecting its growing importance in the e-commerce landscape.

How often does Shopify scan my site for new facts?

Shopify periodically re-scans your site. When it detects changes, it may create new "Suggested update" drafts for you to review. However, this process is not instant. Do not expect a policy change made today to generate a draft fact within minutes; it can take hours or even days.

Because of this lag, if you make a critical and time-sensitive update to your store, such as changing a shipping price, the best practice is to manually edit the corresponding fact in the Knowledge Base immediately. Do not wait for the automated scanner to catch the change.

Where can I get help with my Shopify store's AI strategy?

Shopify provides extensive documentation through its Help Center, which is the best place for technical questions about the Knowledge Base interface. For broader strategic advice on how to position your brand for agentic commerce, consulting with a Shopify expert agency can provide tailored guidance for your specific business needs.