Ecommerce Task Automation Explained: Tasks, Workflows and Paz Routines

A task states the result you need: find the priority bottle descriptions that are missing capacity. A workflow is the steps to get there: pick the products, compare the catalog with the PDP, find the gaps, draft the correction. An AI agent chooses how to carry out those steps based on what it finds along the way.

A fixed rule handles predictable work, such as flagging every empty field. An agent handles the parts that need judgment: reading a PDP and three reviews to confirm a fact, or deciding which of two conflicting values to trust. Either way, the task defines what done looks like. A consumer shopping agent works for the shopper. A commerce task works for your business.

In Paz, a Routine saves what to check, which products and channels it covers, and when to run. Some run on a schedule you set. Others fire on a trigger: a catalog sync, a Merchant Center disapproval, a change in a tracked signal.

Each run ends in a conclusion: nothing to report, a finding your team should know about, or a fix to check against agreed guidelines. Fixes within those guidelines are published to your catalog or Merchant Center and checked. The next run builds on what you decided, so each question is asked once.

In the managed service our experts set up the Routines, review what they find, ship routine fixes within agreed guidelines and bring exceptions to you. Your team sets priorities and resolves those exceptions.

• Goal: the question to answer or the product-data gap to close.
• Trigger and scope: the schedule or event that starts it, and the products, market and channel it covers.
• Inputs: the sources it may read and the business context it should apply.
• Action: a finding, a fix shipped within agreed guidelines, or an exception brought to you.
• Finish line: how you know the output met the goal.
• Exceptions: what happens when sources disagree or one is unavailable.

Write down the expected result for a complete description, a missing fact and an unavailable source, and the task becomes easy to check every week.

A Routine checks a collection each week for descriptions missing verified material or capacity. When it finds one, it drafts the correction from your catalog and PDP and checks it against agreed publishing guidelines with the source information attached. Complete descriptions are left alone. Conflicting source values become exceptions for your team to resolve.

Business context decides the order: the collection you are promoting this quarter comes before a range you are phasing out, even when both have the same gaps. Approval and publishing permissions decide which destinations and fields the fix may change.