Skip to content
All offers
01 / 04

Custom agents

I build an agent for one repetitive job in the tools you already use. A person approves spend, messages and changes to records.

The job to solve

The same checks keep coming back. Someone compares listings, watches prices or assembles a report by hand. Start with the job you can name, the inputs it needs and the person who will review the result.

Example jobs

  • Compare catalogue listings with an approved record and prepare a queue of changes for review.
  • Check prices against margin floors and flag the exceptions.
  • Bring catalogue health checks into one ranked list.

How we work

  1. 01

    Name one job and agree what a useful result looks like.

  2. 02

    Connect the inputs and build around the systems you already use.

  3. 03

    Review the output together. Keep approval with a person before anything changes spend, messages or records.

  4. 04

    Put the job into use, with instructions for reviewing exceptions and keeping it running.

What you leave with

One working agent, a review step and the notes to run it. It handles the routine checking and drafting. You keep the decisions.

Scope, agreed up front

  • Start with one named job. A wider implementation is a separate scope.
  • Agree the approval points before connecting actions that change anything.
  • Discuss ongoing support after the first job is live.

See the related work

Existing work that shows the approach. The scope for your job is agreed separately.

Autonomous Catalogue Agent

An agent that checks marketplace listings and prepares corrections to copy and imagery. A person approves changes before they go back through Amazon SP-API. Catalogue QA takes minutes instead of days.

Read case study

Repricing Engine

A daily repricer for thousands of SKUs. It checks margin floors, buy-box position and competitor price changes before setting a price. Exceptions go to a person.

Read case study

Catalogue Health Dashboard

A dashboard that ranks listings by risk and flags broken listings, missing variants and policy violations before suppression.

Read case study

Send the job you want gone.

Tell me what you do, the tools or channels involved, and what is stuck.

Start a conversation