Skip to main content
This guide walks you through building an agent in Agent Studio, from a one-line description to a deployed revision your team can use. You need the Department lead, Admin or Owner role to create agents.
1

Start a new agent

In Agent Studio, click + Create agent. You have three ways to start:
  • Describe it. Write what it should do in the box (“Describe what it should do…”), choose a Builder model, and Hive drafts the agent for you. For example: Chase overdue invoices in our tone and log every touch, Triage Gmail and draft replies, or Turn notes into client-ready proposals.
  • Choose a template from the template cards under the description box.
  • Create draft manually to start from an empty agent.
A generated agent opens as a reviewable draft, so nothing is live until you deploy it.
2

Write the instructions

In the Setup tab, give the agent a name, description and avatar, then write its Instructions: the job, the tone, what “done” looks like, and what it must never do. The base prompt can be up to 16,000 characters.You can refine the draft by chatting with the builder in the Build pane on the left (“Ask the builder to refine this agent…”). The builder proposes edits, and you review them before they’re saved.
3

Choose how people start it

Pick an input type:Add up to 12 Conversation starters, which are one-click example prompts shown to people using the agent.
4

Ground it in your knowledge

Attach Library files and folders the agent should read, such as policies, templates, price lists or past proposals. They keep their existing access rules, so someone who can’t open a file can’t read it through the agent either. See Library.
5

Connect its tools

Add the connectors the agent may use, for example Gmail or Outlook for mail, Xero for invoices, or HubSpot for deals. Each connection uses the connected account’s own permissions and the autonomy level set for it. See Accounts and autonomy.
6

Set its behaviour and limits

Click Advanced, then under Agent behavior choose the agent’s Autonomy:
  • Read: read and reason without performing actions.
  • Draft: prepare work for approval. This is the safe default.
  • Act: act only within approved operations, workspace policy and existing budgets.
Also under Agent behavior, set:
  • Web access
  • Creativity, from 0 to 1
  • When limits are reached: Block usage, or Switch model. Switch model tries Claude Haiku once, and only after a provider rate limit.
Open Technical configuration to set run budgets under Governance and budgets.
7

Test it

Try the agent before anyone else uses it:
  • Draft preview runs the unsaved draft without touching your tools.
  • Read-only connected test runs the saved draft against your real connected accounts, reading only and never writing.
Start from a conversation starter or your own question. Under Observed evidence and usage you can see which tools were actually used.
8

Save and deploy

Click Update to save a revision, then Deploy to make that revision live. Deploying pins the exact revision, so later edits don’t change what schedules and workflows run until you deploy again. Version history in the agent menu lists every saved revision.

​
Share the agent

Use Share to give named collaborators access. Collaborators can use the agent with their own connected accounts; sharing never hands over yours. Admins can see workspace agents but can’t take over your credentials.

​
Other things you can do

The agent menu also offers:
  • Pin
  • Pause
  • Duplicate
  • Export agent, which gives you a portable file that Import agent can bring back in as a draft
  • Configure delegations, which lets the agent hand sub-tasks to up to 20 other agents, up to 5 levels deep
  • Activity timeline
  • Version history
  • Archive
  • Delete permanently

​
Writing good instructions

“Draft a polite reminder for each invoice more than 14 days overdue, one email per client, and list the invoices you skipped” works better than “handle overdue invoices”.
Name the Library folder or template to follow (“Use the reminder wording in Credit control templates”), and attach it as knowledge.
List hard limits such as “never offer a discount” or “never email a client marked on hold”. These guide the model. The policy gate enforces actual permissions whatever the instructions say.
Start in Draft so every outgoing action is reviewed. Raise autonomy only after a track record. See Autonomy and blast radius.

Publish and evaluate

Keep quality high with test sets, analytics and schedules.

Agent Studio overview

Statuses, roles, safety and limits.