What are AI agent connections? Use real business tools
Connections give agents access to the tools where work and data already live. Access stays separate from the agent's instructions and knowledge so credentials, allowed tools, and review rules remain controlled.
The concept
A plain-language definition before the product implementation.
A connection is scoped tool access
An agent needs a connection to read live product data, search a workspace, open an issue, or send a message. The connection authenticates the provider; the agent or workflow determines when a permitted tool is used.
How AI Agent uses it
Connect once, attach only where needed
AI Agent provides a connection browser and provider-specific setup. Attach connections to agents or use them in workflows, narrow the allowed tools where supported, and keep sensitive write operations behind the same approval policy as the rest of the runtime.
Live data
Read current information from the systems your team already uses.
Scoped access
Equip the agents and workflows that need a provider instead of exposing it everywhere.
Approval-aware actions
Separate safe reads from writes that should wait for a person.
From idea to a working system.
- 01
Connect
Authenticate a provider using its supported setup flow.
- 02
Scope
Choose the agents, workflows, and allowed tools that should use it.
- 03
Observe
See tool activity during runs and review gated actions.
Connect agents when the work depends on live systems
Best for
Reading current business data or proposing approved actions through scoped tools and provider credentials.
Choose another pattern when
A stable reference corpus that only needs retrieval. Put documents and websites in knowledge instead of treating them as live tools.
Included in AI Agent
The product capabilities behind the idea.
- Provider connection catalog
- Agent and workflow attachments
- Tool-level scoping
- Gated write actions
Frequently asked questions
Keep exploring
Related product guides
Agents
An agent is the reasoning layer: it understands a goal, decides which approved tools to use, and works toward an outcome. AI Agent gives that reasoning a clear operating contract instead of relying on one giant prompt.
Skills
Skills package the way your team performs a task. Instead of rebuilding an audit, brief, or research method in every prompt, attach a reusable skill to the agents that should follow it.
Human approval
Human approval puts a deliberate checkpoint between agent reasoning and a consequential action. The agent can prepare the work, while a person decides whether it should be applied.
Build the smallest useful version first.
Start with the agent and one real task. Add workflow control, context, and approvals when the work shows you where they matter.