Skip to content

Glossary

The words Lainstack uses, as the app spells them. Each links to where it is explained in full.

The run state of a published automation whose triggers are on. Publishing makes an automation active; Paused turns its triggers off. See Build your first automation.

An AI assistant in a workspace that people can chat with. It is a persona assigned to a published automation that starts with the Agent work item trigger. See Build your first agent.

A connection to the AI service your AI blocks and agents use, such as an OpenAI-compatible or Anthropic-compatible model, or your own gateway. See Use AI blocks.

A block that pauses a run until a person approves or rejects. It leaves by the Approved, Rejected or Timed out exit. See Put a person in the loop.

A workflow built in the editor from blocks, started by a trigger. Each automation has a draft you edit and, once published, a live version that runs.

One step of an automation, such as HTTP Request or Condition. Every block has settings, outputs and exits, all listed on its reference page.

The panel on the left of the editor that lists every block you can add, triggers first.

An automation that starts with the Called by an automation trigger, so other automations can run it with the Run automation block and get its returns. See Branch, loop and reuse.

A stored sign-in that blocks use to reach another system: an API key, a password, an OAuth sign-in, a mailbox or a storage bucket. Its secrets cannot be read back. See Connections and credentials.

An escalation that asks a person to fix the input a block failed on, after which the block runs again. See Put a person in the loop.

The workspace section that holds reference tables. See Data Lake.

The editable version of an automation. Edits save as you go and stay in the draft until you publish.

A run of the draft from the editor. Reads and computation run for real; blocks that write, send or delete are held, and pass on their pinned output, their last real output or their plan when they have one. See Runs in the editor.

The last part of a webhook URL, chosen on the Webhook trigger. See Start runs from another system.

A point where a run stops and waits for a person: an approval, a correction or a review. Escalations are listed under Escalations in the workspace rail. See Escalations and approvals.

Where a run leaves a block. Most blocks have one, Out; others have several, such as True and False on a Condition, and each can lead to a different block.

The failed runs of one automation that failed at the same block with the same failure code and message pattern. See Failure groups.

A block that writes, sends or deletes in a draft run. It performs nothing, records what it would have done, and passes on a stand-in if it has one. See Held writes.

A value a setting takes from the trigger or an earlier block, inserted by typing @ or by picking it from the Inputs list.

A form people in the workspace fill in to start a run. It comes from an automation’s Operation trigger and is listed under Operations. See Triggers.

Your company or team in Lainstack. It holds members, teams and roles, and contains workspaces. See Organizations and workspaces.

The run state of an automation whose triggers are off. It stays published; switching back to Active turns the triggers on again.

The name, avatar, role and monthly budget an agent works under.

Example output saved on a block’s Samples tab, also called its pinned output. A draft run can start from the trigger’s, later blocks map from it, and a held write passes it on as a stand-in. See The Samples tab.

The column that identifies each row of a reference table. A written row with a key that already exists replaces that row.

Make the current draft the live version of an automation and turn its triggers on.

A table of shared data that automations write to and look records up in. See Look up reference data.

Start a new, real run from where a failed run stopped, reusing the outputs before that point. Resume as is uses the version the run first ran; Resume on live uses the current published version. See Recover failed runs.

What a person may do at one level: super organization, organization or workspace. See Organizations and workspaces.

One execution of an automation, from its trigger to its last block. Every run keeps its steps with their inputs and outputs. See Runs.

Start a draft run from one block of a past run, reusing the outputs before it. See Run from here.

Whether a published automation’s triggers are on (Active) or off (Paused).

The editor showing one run on the canvas, with a run bar, each block’s status and its values on the Run tab. See Runs in the editor.

What a held write passes on so later blocks can run: its pinned output, its last real output, or the plan some blocks give. It is not a real response. See Held writes.

One block’s part in a run, with its status, duration, input and output.

The top level of an account. It holds single sign-on, domains and security policy, and contains organizations.

A group of organization members, for example to receive escalations together.

The block an automation starts from: Webhook, Schedule, Email received, Operation, Called by an automation or Agent work item. See Triggers.

A trigger that gives an automation a URL another system can call to start a run. See Start runs from another system.

Where the work lives: automations, connections, reference tables, runs, escalations and agents. Every workspace belongs to an organization.

Where escalations with no assignee wait until someone claims them.

The AI Agent setting that decides whether an agent can change anything: Off, Ask or Smart. See Build your first agent.

Last updated: