A
- Act on this — Asks Hive to work a signal: it prepares a plan and comes back to you for any decision. See Triage signals.
- Agent — A specialised assistant you configure in Agent Studio for a job, grounded in your Business Brain and held by your rules. See Agent Studio.
- Analyst — The read-only workspace role. Analysts can see why Hive proposed something but cannot act on it. See Roles and permissions.
- Approval — A decision on a proposed action: Approve & run now, Decide later or Do not run. An approval is bound to the exact parameters you saw and expires after seven days. See Review and approve.
- Ask Hive — The chat at the centre of Hive, opened with New chat. It answers from your connected tools, the Business Brain, the web and your files, with citations. See Ask Hive.
- Audit log — The immutable record of every action, tracing signal → plan → decision → result, with one-click undo for reversible actions. See Audit log.
- Autonomy ladder — The five rungs an action type can sit on: Shadow, Suggest, Approve each, Auto in limits and Autonomous. See Autonomy and blast radius.
- Autonomy level (connector) — How far Hive may go with one connected account: Read (search and summarise only), Draft (prepare actions for approval) or Act (carry out approved actions). New accounts start on Read. See Accounts and autonomy.
B
- Baseline — What “normal” looks like for a metric, learned from its history. Hive raises a trend signal only when readings move well away from the baseline, and keep doing so. See Trustworthy signals.
- Blast radius — How much damage an action could do if it were wrong, from B0 (invisible or reversible) to B4 (money, legal or irreversible). See Autonomy and blast radius.
- Business Brain — Hive’s shared memory of your business — entities, facts and relationships, each with its source and trust level. See Business Brain.
- Bring your own key (BYOK) — Funding a seat with your organisation’s own OpenAI or Anthropic key instead of Hive credits. See Billing and plans.
C
- Channel — Slack or Microsoft Teams, used to ask Hive questions from chat. Signals and approvals stay in Hive. See Slack and Teams.
- Citation — A link from a sentence in an answer to the evidence behind it — a document, email, live read, signal, fact or memory.
- Connector — A connection to one of your business tools, over managed OAuth or MCP. See Integrations overview.
- Coverage — The Signals tab that shows what Hive is watching in each connected tool, so an empty feed is never mistaken for an all-clear. See Coverage.
- Credits — AI usage allowance included with each seat, topped up in £25 steps. See Billing and plans.
D
- Decide later — Snoozes an approval or signal so it comes back at a time you choose.
- Department — A slice of the business (Finance, Sales, Customer Success and so on) that roles can be limited to.
- Deploy — Turning a workflow draft into a numbered, live release. See Test and deploy.
- Dismiss — Closing a signal, ideally with a reason. The reason decides what Hive raises next. See Triage signals.
E
- Execution layer — The part of your plan that sets how many agent and workflow runs your firm gets each month.
- Execution memory — See Learned behaviours.
I
- Incoming webhook — A signed URL any system can post events to, feeding Signals. See Incoming webhooks.
K
- Kill switch — Halts every autonomous action and approval across the workspace until an Owner or Admin resumes. See Safety and control.
L
- Learned behaviours — Rules, workflows and exceptions your team has taught Hive, each Active, Proposed or Shadow. See Learned behaviours.
- Library — Your files, folders, templates and saved prompts. Workflow-generated images and PDFs are saved here. See Library.
M
- MCP (Model Context Protocol) — An open standard for connecting tools to AI. Hive can connect to any MCP server by URL. See MCP servers.
- Mute — A time-limited suppression of a signal type or record — 14 days by default, up to 90.
P
- Personal memory — Notes you ask Hive to remember about you, used only in your own chats. See Personal memory.
- Project — A group of chats with shared instructions and files. See Projects.
- Public form — A link anyone can submit to start a workflow, without signing in. See Schedules and forms.
- Pulse — The workflow runs meter shown in the builder.
R
- Receipt — The record left by an action or a finished workflow run.
- Reversible, compensatable, irreversible — How an action can be undone: fully (a draft), by a compensating action (a sent message), or not at all (a refund). Irreversible actions always need a human.
- Run loop now — Runs Hive’s collect-and-detect loop immediately instead of waiting for the hourly run. Admins only.
S
- Scheduled task — A scheduled agent run or workflow. See Scheduled tasks.
- Seat — A paid member of the workspace, on the Basic, Pro or Max tier.
- Severity — How urgent a signal is: High, Medium or Low.
- Shadow — Hive simulates an action without carrying it out, proving what it would do before it is given permission.
- Signal — Something Hive caught that is at risk of being missed, with evidence, why it matters and a proposed next step. See Signals overview.
- Signal Case — A durable situation that keeps evidence, decisions and outcomes together. A pilot feature, off by default. See Signal Cases.
- Snooze — Hides a signal until a set time — between five minutes and 30 days. It returns early if its severity rises.
T
- Template — A ready-made workflow you can start from. See Templates.
- Test run — Running a workflow draft with every action shadowed, so nothing is sent. See Test and deploy.
- Trigger — How a workflow starts: Manual, In-app form, Public form or Schedule. See Triggers and steps.
- Trust tier — How much Hive relies on a fact: verified, reported, inferred or untrusted.
W
- Workflow — A graph of steps that runs from one trigger. See Workflow builder.