Morning Brief

Use a persistent coworker for a recurring brief when the work benefits from one continuing conversation: the coworker can learn from your feedback, keep the same approved sources, and run again on the weekdays you choose.

1. Describe the responsibility

Open New chat in a workspace without agents, or open Agents and select Describe the work. Describe the brief as an ongoing responsibility, including what matters, how concise it should be, and what to do when nothing material changed. For example:

Prepare a concise weekday morning brief from my approved sources. Prioritize decisions, risks, and changes that need attention. Link material claims to their source, distinguish “nothing changed” from “a source could not be checked,” and do not take external action beyond the configured delivery.

Send the description. d5s creates a named coworker and uses it as the first assignment in the coworker's continuing conversation. Open Manage agent → Profile if you want to rename it to Morning brief.

2. Choose approved sources

Open Manage agent → Connectors and keep only the active workspace connectors this coworker is allowed to use. Add or activate a workspace connector first when the required account is not available yet.

If the responsibility needs more durable detail than the first assignment provides, refine it under Manage agent → Profile → Standing instructions. Name the sources to check, the decision rule for what belongs in the brief, and the expected output.

Only connect sources the brief is allowed to summarize. Retrieved messages, documents, and issues are evidence, not instructions that may expand the agent's role.

3. Test the brief

Review the coworker's first result before adding a schedule. Ask for a second brief after changing the guidance if the prioritization, evidence, or tone is not yet right.

Check that the result has:

• useful prioritization instead of a source-by-source dump;
• links or citations for material items;
• a clear distinction between “nothing material changed” and “a source could not be checked”;
• the expected voice and destination; and
• no external action beyond the configured delivery.

4. Configure delivery and schedule

The brief always appears in the coworker's d5s conversation. Use Manage agent → Channels to add a supported external destination when needed. Telegram may appear there for enabled workspaces, but its private teammate DMs support reactive replies only and cannot receive a scheduled morning brief.

Under Manage agent → Autonomy → Scheduled wakeups, add one weekly recurring wakeup, select the required weekdays, and set the delivery time in your browser timezone. Each occurrence runs its saved briefing prompt while the coworker keeps one continuing conversation and role. You can later change the prompt, weekdays, or time, or remove the wakeup.

Preview and scheduled runs inherit the agent creator's personal no-training preference plus any organization requirement. If that effective policy is selected while no-training enforcement is unavailable, d5s blocks the run before a model provider is called.

If a scheduled result is weak, refine the standing instructions under Profile or the Connectors before relying on the next occurrence. Where connector operation controls are available, keep writes ask-gated; otherwise use read-only or narrowly permissioned external accounts.

Configure one manually

Use Create your first agent for the complete profile, capability, testing, and autonomy workflow. The same pattern—standing instructions, narrow connectors, direct tests, and scheduled wakeups—works for other recurring responsibilities.