Use case · Ecommerce operations
Workflows confirm COD orders on WhatsApp, follow up failed deliveries, and keep a store dashboard current, while refunds and cancellations wait for a person.
An illustrative ecommerce operations use case for Agent Serve, the AI workspace where teams build, deploy, and manage AI agents: Shopify and WhatsApp workflows, order tables, a store dashboard, monitors and an order operations Relay.
At a glance
The problem it removes
Cash-on-delivery orders ship before anyone checks the buyer meant it. Failed deliveries come back as courier reports that someone reads days later, and the store numbers live in three exports that never agree.
Step by step
A WhatsApp message asks the buyer to confirm or cancel, and the reply tags the Shopify order.
A monitor flags the rise in failed deliveries, and the buyer gets a WhatsApp follow-up the same day.
The Relay drafts the refund or cancellation, and it runs only after someone approves.
How it fits together
Workflows send WhatsApp confirmations and NDR follow-ups, and tag the Shopify order with the reply.
Orders, replies and courier statuses land in workspace tables as they happen.
The store dashboard reads those tables, so orders, confirmations and RTO update as rows land.
Monitors re-check RTO rate and unconfirmed orders as often as every 5 minutes.
An order operations Relay reads return requests and drafts the refund for a person to approve.
How it runs
Agent Serve runs the messages and the checks, and anything that cancels an order or moves money waits for a person.
A Shopify webhook points at the workflow’s Webhook trigger, and the order is written to the orders table.
The WhatsApp block sends an approved template with Confirm and Cancel reply buttons.
The WhatsApp webhook trigger receives the reply, and the workflow tags the Shopify order and updates the table.
The API block reads shipment status from the courier’s API, with its key stored as a workspace secret.
What to measure
Example figures for a brand shipping a few thousand orders a month, many of them cash on delivery.
−20%
RTO on COD orders
Unconfirmed orders are held before dispatch.
85%
COD orders confirmed
Buyers reply to the WhatsApp message before dispatch.
−40%
Time on NDR calls
Most failed deliveries are resolved on WhatsApp.
100%
Refunds approved
Every refund and cancellation has a reviewer.
Illustrative example. Figures show how the workflow is measured; they are not customer results.
In depth
Agent Serve puts the order workflows, the store dashboard and the Relay that handles returns in one workspace.
Build the COD confirmation and NDR follow-up in the Agent Serve workflow builder, with Shopify, WhatsApp and API blocks.
Agent Serve dashboard panels read the tables workflows write to, and data sources refresh every 15 minutes or query live.
An Agent Serve Relay classes each call as a read, write, destructive or spend. Refunds and cancellations wait for approval.
The approver sees the order, the amount and the reason the Agent Serve Relay drafted, then approves or declines.
Examples
Three ecommerce teams running order operations on Agent Serve, each with a person approving refunds and cancellations.
An Agent Serve workflow asks every cash-on-delivery buyer to confirm, and holds orders with no reply for the team to call.
An Agent Serve workflow reads courier statuses through the API block and messages buyers whose delivery failed.
An Agent Serve Relay reads each return request against the returns policy and drafts the refund for approval.
Examples show what teams can build in Agent Serve. They describe patterns, not named customer deployments.
Before and after
| Work | Before | With Agent Serve |
|---|---|---|
| COD orders | BeforeShipped on trust, or confirmed by calling each buyer. | With Agent ServeConfirmed on WhatsApp, with the reply tagged on the order. |
| Failed deliveries | BeforeRead from a courier report days later. | With Agent ServeCaught from the courier status and followed up the same day. |
| Store numbers | BeforeExports from the store, the courier and the payment gateway. | With Agent ServeOne dashboard over tables that update as rows land. |
| Refunds | BeforeApproved in a chat thread with no record. | With Agent ServeDrafted by a Relay, approved by a person and audited. |
Governance
Agent Serve keeps buyer data where it belongs, and anything that cancels an order or moves money waits for a person.
Security and governanceAgent Serve Relays hold refunds and cancellations for approval unless they fit inside delegated limits you set.
Agent Serve stores courier and gateway API keys as workspace secrets, out of workflow text and logs.
Relays never send email or book meetings without an approval, under any autonomy setting. An approval covers the exact payload the reviewer saw and expires if nobody decides.
Proposals, approvals, declines and executed actions are written to the workspace audit log, which admins can export.
Every workflow run records each block’s inputs, outputs and errors, so a reviewer can trace how a result was reached.
A Guardrails block detects personal data and blocks or masks it before text reaches the next step.
Admin, write and read roles per workspace, plus permission groups that limit which models, blocks and integrations a team can use.
Further reading
Every COD order that ships without a real buyer behind it costs two shipping legs and a blocked unit of stock. Three workflows ask, listen and sweep, so only confirmed orders reach the warehouse.
D2C brands · 5 min readMost returned-to-origin parcels had a failed delivery attempt first, and nobody asked the buyer what went wrong. This playbook catches every non-delivery report, asks the buyer, and sends the answer back to the courier before the next attempt.
D2C logistics · 5 min readMost D2C teams still start the day by stitching a Shopify export, a courier report and two ad accounts together in Excel. Here is a dashboard that fills itself as orders land and tells the team when something moves.
Fashion & D2C · 6 min readQuestions
Point a Shopify webhook at an Agent Serve workflow’s Webhook trigger. The Shopify block then reads, tags, updates or cancels orders, and manages products, customers, inventory and fulfillments.
Not natively. Agent Serve connects to couriers, payment gateways and WhatsApp providers through the API block, with the key stored as a workspace secret, or through Tool connections where the provider offers a server for AI tools.
Near real time. Agent Serve panels over workflow-written tables update as rows land, and data sources refresh as often as every 15 minutes or query live on each view.
Not by default. An Agent Serve Relay holds refunds and cancellations for approval unless you give it delegated limits and the call fits inside them.
No. The figures are illustrative examples of what an ecommerce team might measure with Agent Serve, not customer results.
Last updated
Enterprise access
Agent Serve is available to teams by invitation. Tell us what you want to build and we will set up your workspace with you.