Glossary
On screen, the same thing is sometimes called by different words in different places, and similar words sometimes mean different things. This page lists how they map. What each word means is covered in “How a project is built” on the Project page and in the How it works pages.
Different words for the same thing
The left column is the word these docs use. The screen mostly shows the same word, but some places show a different one. The table lists that label and where it appears.
| Word | Also shown on screen as | Where that label appears |
|---|---|---|
| A workflow started at a set time (a time trigger, or, when no enabled time trigger exists, the time on the first enabled step) | Scheduled | The “Request” of a run record, and Workflow history |
| A decision to advance a review request (the “Proceed to the next step”, “Proceed now”, “Complete the case” or “Proceed to the chosen destination” button, or an automatic action at a deadline) | Approved, Approved (to the next step), Approved (case completed), Automatic deadline action | A person's decision shows as “Approved” in the “Content” of Records (audit trail) and under “Confirmation” in a run record. In the state and “Activity” of a review request it shows “Approved (to the next step)” when the run moved on, or “Approved (case completed)” when the case was completed (“Approved” for a review request under “Notify and proceed after the deadline”, and for one with no record of which happened). One made at the deadline shows “Automatic deadline action” everywhere |
| A decision to send back or reject a review request (the “Send back” or “Reject” button, or an automatic action at a deadline) | Sent back, Changes requested, Rejected | “Sent back” and “Rejected” in the “Content” of Records (audit trail) and as the state of a decided review request; “Changes requested” and “Rejected” in the “Activity” of the review request |
| The results “Failed (reported by agent)” and “Failed (stopped)” | reported by agent, stopped | The description of “Failed / needs action” in the KPI |
| The result “Nothing to do” | No target | The “Excluded” counts in the KPI |
| Insufficient records | Records incomplete, no result was recorded | “Insufficient records” in the KPI “Excluded” counts and Insights; “Records incomplete” in Overview and the Agent completion rate; “no result was recorded” in the run record |
| External services (a kind of connector) | External integration | Forbidden operations in the execution rules, and operation kinds |
| Agent-to-agent integrations (a kind of connector) | External Agent | Forbidden operations in the execution rules, and operation kinds |
| Review step | Human review, Human check, Check settings | “Human review” is the kind chosen for a step, “Human check” adds one in the editor, and the heading of its settings is “Check settings” |
| Workflow run | case | “Complete the case” on a review request, and some descriptions on screen |
| Awaiting review | Waiting for a person, Waiting for confirmation | All mean a review not yet decided. “Awaiting review” appears in Workflow history and Review requests; the others appear in Overview and the run record |
| Run record | Record | The tab of the detail that opens from a line in Execution history |
Similar words for different things
These pairs look alike or use the same word, yet mean different things. They are easy to confuse, so the table sets out the difference.
| Word | Easily confused with | The difference |
|---|---|---|
| Execution | Workflow run | An execution is one piece of work by an agent. Each is a line in Execution history and opens as a run record. A workflow run is one pass through the steps, and the executions made during it are linked to it (how many depends on how steps and agents are laid out). Workflow runs are listed in Workflow history |
| Run record | Records (audit trail) | A run record is the detail of one execution. Records (audit trail) lists certain operations and events in the project (review decisions, changes to execution rules and members, and so on) |
| The result of an execution (Completed, Failed, Timed out and the rest of the seven) | The state of a workflow run (Running, Awaiting review, Done, Failed, Rejected, Cancelled) | They appear in different places. “Failed” is in both, but what is counted differs |
| Review request | Awaiting review | A review request is the request that reaches the reviewers. “Awaiting review” means not yet decided (the state of a workflow run, and the undecided section of the Review requests screen) |
| Review | Approve | Reviewing covers looking at the output and deciding. “Approved” is the screen's label for a reviewer's decision to proceed; one made at the deadline is shown as “Automatic deadline action” and is not an approval |
| Guardrail (an agent setting) | Stopped by the guardrail | An agent's guardrail is an instruction to the agent; writing it does not reliably stop an operation. “Stopped by the guardrail” is the reason given when the request was stopped by the platform's check on its input and was not carried out |
| “No target” as the Agent completion rate | The result “Nothing to do” | “No target” in the rate means there is nothing to count. “Nothing to do” as a result means the run never started |
| Basic rules | Global basic rules | The basic rules are chosen for a project. The global basic rules are set for the whole tenant, and a project cannot be looser than them |
| Review step | Run by Agent | Kinds of step. A review step holds no agent and a reviewer decides; a “Run by Agent” step holds agents |
| Project step | Playbook step | A project step is where agents are placed, and a unit of the project's flow. A Playbook step is one item in the procedure attached to an agent |
| Ledger | Records (audit trail) | A ledger is an external shared record that agents read and write as their goals (instructions) tell them. Records (audit trail) is what the platform keeps of certain operations and events in the project |
| Scheduled | scheduled run | Both mean it started at a set time. “Scheduled” appears for a workflow; “scheduled run” appears in the “Request” of a cycle sweep's run record |
| “Check settings” above a selected review step | The “Check settings” button in the editor | The former is the heading of that step's settings; the latter checks the whole draft |
The definitions and how things work are in the How it works pages, such as Projects and steps, Human review and Records and metrics.