See what we have shipped

Engaige now has its own chat widget, so you can put it on your shop and have your AI agent answer shoppers there directly. You set it up under Settings, Channels, Chat widgets: pick a brand colour, write the welcome message, add up to three conversation starters and an optional disclaimer, and watch each change in a live preview as you type.

Before it goes on your site you can try it on a test page, a demo storefront where you talk to your real agent. The widget only loads on the domains you list, subdomains included, and it answers once the agent assigned to it is published.

When the agent hands overWhat this meansThe moment the agent stops answering by itself and gives the conversation to a person., the widget asks the shopper for their name and email, or at the very start of the chat if you prefer. The handover email to your team carries those details and a reply goes straight to the shopper, and that email waits until the shopper has left their details, or up to an hour, so your team does not receive a handover with nobody to answer.

Any analysis you run with Engaige AI can now be saved as a skill and run again with one click, so the question you worked out once gives you the same answer on this week's data without rewriting it. Engaige AI drafts the skill's name and instructions from the conversation, and you edit them before saving.

After an answer that did real analysis, a "Save this as a skill" bar appears above the chat. To run a skill, type / in the chat and pick it, or open it under Settings, Skills, where your whole team can see, edit and run every saved skill. A skill only ever runs when someone runs it.

Shops using AfterShip can connect it so the agent answers delivery and return questions from AfterShip itself, rather than a colleague looking the parcel up. Return lookups need AfterShip Returns on your AfterShip plan.

Engaige now speaks MCP, so an assistant you already use can read your workspace directly instead of you going to the dashboard for it. It can search your conversations and open a full thread, follow how the agent reasoned and which actionsWhat this meansSomething the agent actually does in another system, such as issuing a refund or changing a delivery address, as opposed to something it says. it ran on a case, search your knowledgeWhat this meansThe documents and answers you give the agent to work from, such as your returns page or a product manual. documents, and read your agents, policiesWhat this meansA set of instructions you write for how the agent should handle one kind of request, such as a return or a late delivery. and configured actions.

Everything it can reach is read-only and bound to your own organisation: twenty-four tools that look, and none that change anything. You connect it once and ask your questions where you already work.

Engaige AI is now switched on for every workspace. It reads your own support history and tells you what to improve next, answers questions about your numbers and draws them as charts, explains what happened in any conversation, and proposes the instructions and knowledgeWhat this meansThe documents and answers you give the agent to work from, such as your returns page or a product manual. a new agent should open with.

The last stretch of work before this was about the hard cases. An investigation that stalls now picks itself back up instead of leaving you waiting, and the reasoning traceWhat this meansThe record of how the agent reached a reply: which policy it followed, which data it read, and what it decided. behind an answer shows each step as it happens, including the work Engaige AI hands to its own specialists, so a long answer is something you can watch being built rather than a spinner.

Shops shipping with PostNL can connect it so the agent answers where-is-my-parcel questions from the carrier itself, by barcode, rather than a colleague opening the tracking portal.

Shops on BigCommerce can now connect their store and put the agent to work on their orders, the way Shopify and Magento shops already could. The agent can look an order up and check it against the customer's email, list a customer's recent orders, cancel an order that has not shipped, correct a delivery address, check stock per variant, and issue a refund in full, in part or for a custom amount, with a free reshipment for what went missing.

A refund claim that arrives with a photo now gets the photo looked at. Every image a customer attaches is screened for signs it was AI-generated or AI-edited, and the result travels with the agent's reasoning, so a claim built on a fabricated picture no longer rides through on the strength of the text alone. There is nothing to switch on: it is part of how the agent reads an attachment.

The screening is a signal, not a verdict, and the agent is instructed accordingly: a high score never turns into an accusation towards the customer, it turns into a quiet escalation to your team. Before an image is checked, everything identifying is stripped from its metadata, and an image that cannot be cleaned is simply not sent.

Off by default. Turn it on per agent, in the agent's settings.

The agent can look up a customer's marketing profile and manage their email subscriptions from inside a support conversation, so unsubscribing someone no longer means a colleague switching tools. Subscribing honours the list's own single or double opt-in setting, and you decide how broad an unsubscribe may reach. Email only: SMS subscriptions are left alone.

Customers ask us for insights they can act on. Engaige AI could already answer questions about your support data with real numbers, but the answers arrived as text and tables. They now come back as a chart whenever the numbers suit one, in the conversation and on your home screen, so the shape of an answer is something you see rather than something you work out.

The Coverage view answers the question every support lead asks first: of everything that came in, how much did the agent actually take? It shows the share it handled end to end, the share it forwarded, and the reason it forwarded each one, so the number you report is one you can trace rather than one you estimate.

Forwarded conversations used to be missing from Insights altogether, which made every coverage figure quietly optimistic. They are counted now, and grouped by why they left the agent, so the gap between what it handles and what it could handle is the thing you are looking at.

A knowledge folder can now be pointed at your sitemap instead of pasting page URLs one at a time. Engaige reads the sitemap, picks up the pages behind it, and keeps the folder in step as the site changes, so the answers the agent gives come from what your site says today.

A folder that fails to sync says so on the folder itself, rather than going quiet and leaving the agent to answer from something stale.

Customers wanted finer control: the agent takes one part of a request while a person takes another. Every stepWhat this meansOne instruction inside a policy, like a single line in a set of instructions for your team. in a policyWhat this meansA set of instructions you write for how the agent should handle one kind of request, such as a return or a late delivery. can now be marked autonomousWhat this meansThe agent answers the customer itself, without waiting for someone on your team to check the reply first. or assistWhat this meansThe agent writes a reply and leaves it for someone on your team to send, change or discard.. When the agent is working on its own and reaches a step marked assist, it passes the conversation to a person at that exact point instead of carrying on.

So a refund inside the window you allow can run unattended, while the exception one step below it always reaches a colleague. Where that line sat used to depend on the agent's own judgement about when to ask for help. It now sits in the instructions you wrote, which means you can put it where you want it and see where it currently runs.

Knowing that the agent could do better is the easy part. Knowing what to change means reading conversations, and that was work you had to do yourself. Engaige AI now reads your own support history and comes back with what to change: the gaps in your instructions behind repeated handoversWhat this meansThe moment the agent stops answering by itself and gives the conversation to a person., the knowledgeWhat this meansThe documents and answers you give the agent to work from, such as your returns page or a product manual. it turns out to be missing, and the answers your team keeps rewriting. Its strongest signal is the difference between what the agent drafted and what your colleague actually sent, taken across many similar conversations rather than judged one message at a time.

Suggestions arrive as cards you approve or set aside, and nothing is applied on your behalf. Engaige AI then checks quietly whether the thing behind an approved card really did go away, so over time it gets better at knowing which suggestions are worth your attention.

A new agent gets the same treatment. Once enough of your past tickets are in, Engaige AI works through them and proposes the instructions and knowledgeWhat this meansThe documents and answers you give the agent to work from, such as your returns page or a product manual. to open with, so nobody starts from an empty setup.

Shops fulfilling through Ecomflow can connect it so the agent looks up shipment details, tracks a parcel and searches stock, instead of a colleague doing that in a second screen.

Shops shipping with QLS can connect it so the agent pulls the fulfilment picture for an order, warehouse side included, and the details of a specific shipment. A delivery question is answered from what QLS actually knows rather than from the last status a colleague remembers.

Signing in used to open a dashboard that showed you numbers and left the next move to you. It now opens Engaige AI, with what to improve next, the questions you ask most as ready-made prompts, your earlier conversations, and the drafts waiting on you, which you can approve without going anywhere else.

The point of it is that improving the agent and running it happen in the same place. You read what Engaige AI found, open the policyWhat this meansA set of instructions you write for how the agent should handle one kind of request, such as a return or a late delivery. or the conversation behind it, and act on it there.

Connect Logic4 and the agent works in it directly, so it can retrieve an order by its order number and look up a customer by email address, instead of a colleague opening a second screen to find out.

Ask Engaige AI what happened in any conversation and it comes back in plain language: what the customer wanted, what the agent did, which policyWhat this meansA set of instructions you write for how the agent should handle one kind of request, such as a return or a late delivery. it followed and where it handed over. You get the account of it without reading the thread yourself.

It is the fastest way into a conversation you want to understand. Instead of scrolling to find the moment, you ask, and then open the policyWhat this meansA set of instructions you write for how the agent should handle one kind of request, such as a return or a late delivery. or the step behind the answer from there.

Connect Adobe Commerce and the agent works in it directly, so it can retrieve an order with its full details and find all of a customer's orders by email address, instead of a colleague opening a second screen to find out.

Engaige works inside Belco: it answers in the channel your team already uses, and hands a conversation to a colleague where your policies say so.

Engaige works inside Front: it answers in the channel your team already uses, and hands a conversation to a colleague where your policies say so.

Connect Skio and the agent works in it directly, so it can retrieve a customer's subscriptions by email address and change quantity, price, or product variant on a subscription, instead of a colleague opening a second screen to find out.

Connect WooCommerce and the agent works in it directly, so it can retrieve a specific order from your store and pull up a customer's recent orders by email address, instead of a colleague opening a second screen to find out.

Connect Returnless and the agent works in it directly, so it can find any return by order number plus email or postcode, and answer instantly on status, items, compensation and refund details and give live tracking on the return shipment, down to confirming that it has arrived at your warehouse, instead of a colleague opening a second screen to find out.

Connect Recharge and the agent works in it directly, so it can pull up a customer's subscriptions and schedule and update the shipping address on a subscription, instead of a colleague opening a second screen to find out.

Connect Picqer and the agent works in it directly, so it can look up the order, picklist, shipments, backorders, and returns in the warehouse and check live stock levels, instead of a colleague opening a second screen to find out.

Connect Loop Returns and the agent works in it directly, so it can look up a return by order name and customer email and flag a return for human review and add an explanatory note, instead of a colleague opening a second screen to find out.

Connect Juo and the agent works in it directly, so it can pull up a customer's subscriptions, schedule, and next delivery and skip one of a customer's upcoming orders, instead of a colleague opening a second screen to find out.

Engaige works inside Kustomer: it answers in the channel your team already uses, and hands a conversation to a colleague where your policies say so.

Connect Shopify and the agent works in it directly, so it can pull any order with line items, SKUs, prices in the customer's currency, discounts, financial and fulfilment status, shipping address, tags, and fulfilment tracking (carrier, number, status, URL) and look up a customer by email, with their profile and linked orders, instead of a colleague opening a second screen to find out.

Engaige works inside Trengo: it answers in the channel your team already uses, and hands a conversation to a colleague where your policies say so.

Engaige works inside Freshdesk: it answers in the channel your team already uses, and hands a conversation to a colleague where your policies say so.

Connect ReturnGO and the agent works in it directly, so it can list a customer's returns (RMAs) and get the details of a single return, instead of a colleague opening a second screen to find out.

Engaige works inside Zendesk: it answers in the channel your team already uses, and hands a conversation to a colleague where your policies say so.

Connect ShopWeDo and the agent works in it directly, so it can retrieve order details by Shopify order ID and cancel an order before fulfilment starts, instead of a colleague opening a second screen to find out.

Engaige works inside Salesforce Service Cloud: it answers in the channel your team already uses, and hands a conversation to a colleague where your policies say so.

Engaige works inside Richpanel: it answers in the channel your team already uses, and hands a conversation to a colleague where your policies say so.

Engaige works inside CM.com: it answers in the channel your team already uses, and hands a conversation to a colleague where your policies say so.

Engaige works inside Re:amaze: it answers in the channel your team already uses, and hands a conversation to a colleague where your policies say so.

Engaige works inside Help Scout: it answers in the channel your team already uses, and hands a conversation to a colleague where your policies say so.

Engaige works inside HubSpot: it answers in the channel your team already uses, and hands a conversation to a colleague where your policies say so.

Connect Next Chapter and the agent works in it directly, so it can retrieve order details by order code and customer email, instead of a colleague opening a second screen to find out.

Engaige works inside Intercom: it answers in the channel your team already uses, and hands a conversation to a colleague where your policies say so.

Engaige works inside Gorgias: it answers in the channel your team already uses, and hands a conversation to a colleague where your policies say so.

Connect Rever and the agent works in it directly, so it can look up returns and their progress by order ID and customer email, instead of a colleague opening a second screen to find out.

Connect Monta and the agent works in it directly, so it can pull up an order with its fulfilment status and update the delivery address on an order, instead of a colleague opening a second screen to find out.

Engaige works inside eDesk: it answers in the channel your team already uses, and hands a conversation to a colleague where your policies say so.

Connect SendCloud and the agent works in it directly, so it can pull up an order with its outbound shipments and get the latest tracking events for a parcel, instead of a colleague opening a second screen to find out.

Engaige works inside Dixa: it answers in the channel your team already uses, and hands a conversation to a colleague where your policies say so.

Engaige works inside Crisp: it answers in the channel your team already uses, and hands a conversation to a colleague where your policies say so.