Create an agenda with an agent

A practical workflow for turning unstructured event material into speakers, sessions and their connections.

Yes: an appropriately authorized agent can create an agenda and connect speakers using the Core API. For a customer working against a short deadline, start with their existing notes or documents, a reviewed plan and a small verified sample. A custom integration or MCP server is not required if the agent already has a suitable HTTPS execution tool.

Prepare the material

Provide the agent with the relevant agenda notes, speaker biographies, organization ID, event ID, event dates and timezone. Describe which existing records should be kept or changed. The Core Events resource exposes the dates but not the timezone; obtain the latter from the organizer.

Keep API credentials in the execution environment’s secret store. Only share source material that the customer intends to process in that AI environment.

Use this prompt

Use the Brella Core API skill to prepare this event's agenda.

Organization ID: [our organization]
Event ID: [our event]
Event dates and timezone: [confirmed values]
API origin: [the HTTPS address provided by Brella]

Extract sessions and speakers from the attached material. Produce a table
with source IDs, title, local start time, explicit timezone offset,
duration, track, location, description and speaker identities/roles.
Mark missing values and ambiguous matches; do not guess.

Read existing speakers, tracks and sessions if you have authorized access.
Propose which records to reuse, create or update. Preserve existing
speaker assignments unless our reviewed plan explicitly replaces them.
Show one complete example and the remaining batch plan for approval.

After approval, execute the sample if your environment can call the API,
then verify it before the remaining approved records. Keep source IDs,
returned Brella IDs and per-item results so an interruption can be
reconciled without blindly retrying writes. Otherwise give us runnable
requests and clearly state that nothing has been imported.

Download the agenda plan template. It is a planning document, not a payload to submit to the API.

Follow the supported sequence

1. Review the plan. Resolve the timezone, dates, durations, identities, track and intended changes.
2. Resolve or create speakers. Save their actual Brella IDs. Speaker creation needs a first name.
3. Resolve a content track. Read an existing content/external-content track, or create a content track when authorized.
4. Create or update one session. Use start_time, duration, the track ID and optional speaker_assignments containing the real speaker IDs.
5. Verify. Read the session with include=speaker-assignments,track, check the associated IDs/roles, and confirm its displayed time in the organizer agenda.
6. Complete the approved batch. Record each result, stop on uncertainty and reconcile before retrying.

The interactive agenda recipe explains the necessary scopes and provides request builders.

Keep these limits visible

• The agent needs an HTTP-capable execution environment and a correctly scoped Core token. Reading the skill alone does not connect it to Brella.
• Session descriptions require DraftJS JSON, while speaker biographies can be text.
• speaker_assignments replaces the session’s entire assignment list when supplied. [] removes every assignment.
• Core has no general session draft/publish input and no speaker DELETE action. Do not assume an import is hidden or automatically reversible.
• Core POST is not a general upsert. Existing external IDs should be reconciled before creating records; do not retry a timed-out create blindly.
• Each request commits separately. A partially completed import needs a resume log, not a claimed all-or-nothing rollback.

The documentation and skill are grounded in the Core route, wrapper, mutation and service implementation. Actual success still depends on the customer’s token configuration, event state and runtime verification.