Gladly's agentic commerce capabilities let the same AI that handles your Customer Conversations also help shoppers find products, compare options, and add items to cart – all in one Chat, without adding a separate tool or a second Conversation. This article walks through what changes, what you'll need to decide before setup, how implementation works, and how to test before you go live.
What's changing
Once agentic commerce is turned on, a few things move or work differently:
Business Hours now lives under Settings > Business Hours instead of Settings > Chat, with more flexibility to schedule holidays and exceptions.
Chat can run 24/7. Gladly AI can keep helping customers outside your business hours. If a Conversation comes in after hours and the AI can't fully resolve it, Gladly creates a follow-up email for an Agent to pick up when your team is back online.
Quick Actions (like Track Order or Returns buttons) are going away as part of this change
Now that Gladly AI can handle those requests conversationally around the clock, a separate set of buttons isn't needed anymore – Customers get one seamless Conversation instead of a mix of buttons and Chat.
Chat onboarding is more customizable. Instead of a fixed setup in Settings > Chat, onboarding questions now live in your AI Agent's workflow, so you can add or adjust what's asked before a Customer reaches an Agent.
Product recommendations get their own home. You can fine-tune what Gladly AI recommends, and Customers can add a recommended item straight to their cart from the Conversation.
You may also notice "Experiment" tags on some conversations in AI Sessions for the first few days after launch. That's expected – it just means Gladly is watching that Conversation closely while performance is validated, not that anything is wrong.
Why it’s worth setting up
Handling shopping questions and support questions in the same Conversation means fewer Conversations get created in the first place, and more of them end in a resolution. Gladly has seen this combination lead to roughly a 20% increase in resolutions without a corresponding rise in conversation volume, since one Conversation can now do the work that used to take two.
Before you start: define your experience
Before implementation begins, you'll work with your Gladly team to make a few decisions. Having answers ready speeds things up considerably.
Which Chat experience do you want? Reactive is the more familiar, compact Chat window. Persistent is a fuller experience that stays available at the bottom of the page, closer to a search bar. Either way, the shopping and support capabilities are the same – this is purely about how the Chat looks and feels on your site.
Reactive Chat experience
Persistent Chat experience
Where do you want to start? Most customers implement directly in production, since that's the simplest path and reflects real traffic. Larger organizations sometimes prefer to validate in a staging environment first – more on this under Testing below.
What should happen after hours? If Gladly AI can't resolve a conversation outside your business hours, would you rather it auto-close, or hand off to an email Inbox so an Agent can follow up the next business day?
What should Gladly AI know? Let your Gladly team know which existing Answers and audiences it should draw on, so it reuses content you've already written instead of duplicating it. This is also the time to flag anything that isn't written down anywhere yet – for example, which products you'd like recommended first, or how much back-and-forth makes sense for a more technical product.
Do you want to A/B test the change? You can compare the new experience against your current one before fully switching over, or move straight to sending all chat traffic to the new experience.
Set up agentic commerce
Once your experience is defined, your Gladly team will work through setup with you. Here's what each part covers.
Set up business hours
Your Chat Business Hours are configured (or moved over, if you already had them set) under Settings > Business Hours. Because Gladly AI can now run outside those hours, it's worth double-checking this includes any holidays or exceptions you care about – this setting is what determines when a conversation gets the after-hours treatment you chose above.
Set up the workflow
Gladly configures (or updates) your Chat's workflow so onboarding and after-hours handling work the way you decided. In practice, this means the workflow checks your Business Hours and either lets Gladly AI continue helping the Customer, or gives them a friendly "we're offline" message with a follow-up email, depending on what you chose.
Set up the skill or connect it to Answers
This is where Gladly AI gets its shopping abilities – searching your catalog, sharing product details and reviews, making recommendations, and adding items to cart. Your Gladly team connects this to your product catalog and to your existing Answers and Audiences, so nothing needs to be rewritten. You stay in control of which capabilities are turned on – for example, you can allow product search and recommendations while leaving add-to-cart off. Review this list with your Gladly contact before launch.
Fine-tune the Questions & Recommendations Guide
Your existing Questions & Recommendations Guide is renamed (typically to something like "FAQ") and scoped down to policy and account-type questions only, since product questions are now handled by the new commerce capability. Your website is also removed as a source for this feature, since the AI now pulls product information directly from your catalog instead.
Update your Questions & Recommendations Guide using the suggested format:
Adjust the name and section header to “FAQs”
Adjust the Guide description to: Use this “Guide to answer policy questions, such as return and exchange windows. Do not use to give product recommendations.”
Turn on the A/B test experiment
If you chose to A/B test in the Data Collection step, your Gladly team sets this up on your Chat Entry Point so a portion of traffic sees the new experience while the rest continues as-is. If you chose to go straight to the new experience, all Chat traffic is routed there instead. Either way, don't worry if you see "Experiment" tags on some Conversations during this period – that's just labeling which setup a Conversation ran on.
Testing
How you test depends on your size and setup:
Complex customers with an involved QA and tuning process often want to validate first in a staging (UAT) environment before anything touches real customers. This is a good place to confirm shopping functionality works as expected, though full end-to-end service testing is best done in production, so plan for some testing to continue after go-live too.
Customers ready to get started without in-depth QA find it simplest to flip on directly in production. This also doubles as the most realistic test, since it uses your real site and real traffic from the start.
Whichever path you choose, plan to personally walk through the experience before considering it done:
Open the Chat and confirm you're greeted appropriately.
Ask a product question and confirm you get a recommendation with a link and image.
Ask a follow-up question about that product.
Add a product to your cart.
Ask a normal support question (like an FAQ or order status) and confirm it's still answered correctly.
Confirm you can be transferred to a live Agent when needed.
Try this outside your business hours to confirm the after-hours experience behaves the way you chose.
Check that the chat's colors and branding still look right.
Monitoring
Dedicated self-serve reporting for agentic commerce is on the way. Until it's available, your Gladly team can review individual AI sessions with you and share results directly, and will flag anything that needs your attention.
Things to watch out for
AI can move between topics more flexibly than before, so behavior may feel a little different even for FAQ-type Conversations you've had for a while. Your Gladly team will test extensively with you ahead of go-live to catch anything that needs adjusting.
Some sites' built-in bot or fraud protection can interfere with add-to-cart. If Customers report trouble adding items to their cart, flag it so your site can be allowlisted.
If your Chat runs 24/7, double-check your after-hours setting matches what you intended (auto-close vs. email follow-up) – it's an easy setting to get backwards, and it changes what happens to Customers who reach out overnight.
When to contact Gladly
The behind-the-scenes setup – connecting your catalog and site content into the AI's toolkit, enabling the underlying settings, and configuring the AI's shopping logic – is handled by your Gladly implementation team, so you won't need to touch this yourself.
If something Customer-facing breaks (for example, add-to-cart failing site-wide, or the Chat visibly breaking your page), contact your Gladly team right away – they can quickly revert to your previous experience while the issue is fixed. For smaller issues, like one flow underperforming or a styling quirk, there's no need to turn anything off – just flag it, and it's usually a quick fix on Gladly's side.