Configure
Create or update an agent through the API, scoped to the active workspace.
API overview · African voice AI
Build customer voice workflows on Magana.
Create and configure voice agents, connect business actions, and inspect what happened after a conversation. The API sits alongside the dashboard so your team can build the integration path it needs.
From configuration to review
- 01Create an agent in a workspace
- 02Connect an approved voice channel
- 03Run an interaction and configured actions
- 04Read the call record and outcomes
Voice-agent lifecycle
Define an agent's instructions, persona, greeting, and relevant actions. Test the conversation, then attach a channel and review call records as your workflow runs.
Connect and test
Use a browser test call or configure a phone path for an inbound or outbound conversation.
Inspect
Read the call record, transcript when saved, and analysis after the interaction.
Actions & integrations
Configure webhook actions so an agent can request information or perform an approved operation in your system. Keep credentials in workspace settings and decide which actions may run during a conversation.
Webhooks & outcomes
Configured webhook actions call your endpoints as part of the agent workflow. Review call outcomes, action results, and follow-up needs in the records available to your workspace. Check the current documentation for the specific event and response contracts your integration uses.
Design the next action
- Identify which information the agent may retrieve
- Define the operation and response your system accepts
- Map uncertain results to a person or follow-up queue
Authentication & workspace boundaries
Authenticate API requests with a workspace-scoped key. Each request resolves to one workspace; records and actions are scoped to that workspace. Keep keys on the server and rotate them through Developer Settings when needed.
Dashboard member access and API-key access are separate paths. Choose who can configure agents and who can read interaction data before connecting a live workflow.
Build a small voice workflow first.
Start with one agent, one action, and an outcome your team can inspect.