Agent Components

Engineering

Examines incoming messages to identify available information.

Places and confirms supplier orders for the store's stock, for example a batch of mugs.

Takes care of the Salesforce side of the work: bill uploads, field updates and lead routing.

Answers inbound leads in every time zone, writing a distinct reply for each one based on its context.

Interprets campaign dates, including relative dates.

Flags when schemas, code and docs fall out of sync; it has opened 55 analysis discussions.

Handles numeric questions by working out transcript points, priority bonuses and total eligibility scores.

Data collection.

Search engineer sub-agent equipped with its own set of Shopify-specific skills.

For issues scored 80 or above, searches for reasons the issue should stay open; if it also recommends closing, it picks the reason and writes the closing comment.

Analysis and synthesis.

Reviews security.

Security checks

Builds application features.

Combines OpenSEO with the author's Search Console and Google Analytics data.

Search metadata

Sub-agent dedicated to Shopify engineering.

Delegates and organizes work.

Simplification assessment

Reads each incoming incident and decides which specialist agents to call, based on the context that is still missing.

Sits on top of the other workflows and produces insights for a particular business need, depending on what the user asks.

Python-writing agent using code actions — minimal library, multi-provider support.

SMS agent that steps in whenever an email to a lead bounces.

Starts from a finished newsletter issue or an idea and produces social posts.

Implements code and adds tests.

Narrows retrieval to likely relevant policy documents.

Builds candidate search queries.

Reads the speech balloons and works out which character says each line.

A pool of more than 40 agents for AWS, CI/CD, Cloudflare, end-to-end tests and other jobs.

Engineering specialist of the Ari Collective — implementation, refactors, tests.

Looks at the agency's recent work and turns it into social posts without revealing client data; it can now also make short videos.

Checks the taxonomy itself for hierarchy gaps, naming inconsistencies and better ways to group related concepts.

Checks logs and metrics for evidence about a candidate cause.

Searches the web and returns findings to the lead researcher.

Combines diagnostic findings into reports.

Refines the other agents’ responses.

An internal assistant for call-centre consultants that tells them the next step and never addresses customers itself.

Allocates resources and queues specialist agents.

Receives sub-agent results through a shared mailbox.

Takes every customer query first, sends it to the right use-case agent, and handles conversation endings and handovers to human operators.

Coordinates Hiring Assistant’s agent activity.

Coordinates specialist investigations and synthesizes findings.

Submits the agreed data-preparation code.

Domain specialist receiving handoffs from the triage agent in OpenAI Swarm.

Stateless routing agent that hands off to specialists in OpenAI Swarm.

Software engineering agent with ACI — 12.5% SWE-bench, 87.7% HumanEvalFix.

Checks service metadata against maintenance workflows.

Build maintenance

Classifies test failures to guide diagnosis.