Understand the fundamentals
Learn the difference between an AI assistant and a role-focused agent, and why reliable workflows need defined inputs, outputs and limits.
Learn how the product journey works →Learn how to plan AI agent workflows
Build a grounded understanding of AI agents, choose suitable business workflows, plan integration and approval boundaries, and connect the ideas to relevant product roles.
Choose a learning path
Resources explain concepts and planning choices. For product configuration, use Documentation; for issue resolution, use the Help Center.
Find the right starting workflow
Begin with repeatable work that has a clear owner, known source data, a reviewable output and a sensible fallback.
Compare AI agent use cases →Plan controls before automation
Decide what an agent may read, draft or change, and identify actions that should wait for human approval.
Study permissions and human control →Practical implementation guides
Use these guides to turn a business goal into a bounded workflow before selecting tools or agent roles.
Define an agent-ready process
A useful starting process is frequent enough to matter and structured enough to inspect.
- Name the business outcome and process owner.
- List required inputs and their source systems.
- Define the output format and recipient.
- Record exceptions that need human judgment.
Map integrations and permissions
Connect only the systems a workflow needs and separate reading information from changing records or sending messages.
- Identify objects the agent must read.
- Identify proposed write or send actions.
- Use the narrowest practical permission scope.
- Plan for expired access and revoked connections.
Design human checkpoints
Approval is most useful where an action is sensitive, hard to reverse or dependent on business judgment.
- Review external communications before sending.
- Escalate policy or confidence exceptions.
- Require review for consequential record changes.
- Preserve context for the person making the decision.
Explore workflow examples by outcome
Each example starts with a business result and links to a fuller workflow page with relevant agents, integrations and human review points.
Build a lead research workflow
Structure account criteria, prospect research, enrichment, deduplication and CRM handoff without treating every discovered record as qualified.
Explore lead generation workflows →Plan a support workflow
Combine ticket intake, classification, knowledge retrieval, response drafting and escalation with clear policy boundaries.
Explore customer support workflows →Build a recurring reporting workflow
Define trusted sources, calculation ownership, reporting cadence, distribution and review before decisions are made.
Explore automated reporting workflows →Plan document processing
Map document intake, field extraction, validation, exception handling and downstream system updates.
Explore document processing workflows →Organize inbox automation
Separate classification and drafting from sending, deletion or other actions that need stronger controls.
Explore email inbox workflows →Design a content workflow
Connect briefs, brand guidance, drafting, review and publishing while keeping final claims and approvals with people.
Explore marketing content workflows →Move from learning to product discovery
Once the workflow is clear, compare specialized agent roles by business function. Agent catalog pages describe intended roles and workflow fit; availability can depend on your onboarding scope.
Looking for product instructions?
Documentation covers product setup and configuration. The Help Center covers symptoms, errors and recovery steps.
Browse AI Robot Agents product documentation