Workflow
Turn Team-Chat Commitments into an Aurora Review Queue
Turn permitted team-chat excerpts into a review queue that separates requests, ideas and accepted commitments.

Turn a permitted team-chat excerpt into a review queue by separating a request, a proposed idea and an accepted commitment. Keep the speaker, source, owner and date visible before creating tasks. A chat message alone does not establish that somebody accepted a deadline.
Source check: October 22, 2026. This workflow uses user-supplied notes as its baseline. Slack and Microsoft Teams were not present in the enabled public Aurora App catalog checked for this campaign. That does not prove every custom connection is impossible; it means this guide does not promise a native chat ingestion path or claim one was tested.
Bring a small, permitted conversation packet
Select the messages needed to understand one decision. Include surrounding replies that accept, reject or qualify an assignment. Remove credentials, unrelated personal discussion and material the intended reviewer may not access. Retain a source reference you can open yourself, while respecting the channel's sharing rules.
If a separate connection is available in your intended account, verify its identity, tool list and exact channel access first. A Slack app connected to another assistant or a Microsoft email connection does not establish that Aurora can read the same team's chat.
Classify the statement before assigning work
| Statement type | Evidence to retain | Queue treatment |
|---|---|---|
| Request | Who asked, what they need, stated timing | Needs an owner to accept it. |
| Accepted commitment | Named speaker's acceptance and any condition | Propose an action with the accepted scope. |
| Idea or possibility | Conditional wording and decision still open | Keep as a decision, not a task somebody promised. |
| Status update | What changed and the relevant source | Update the review context; check whether work remains. |
| Contradiction | Both statements and their order | Ask the responsible person to resolve it. |
Build a commitment review queue
Give each proposed item a source, an evidence state and a next decision. Avoid turning “we should” into “Taylor will.” If the owner is unknown, a useful output says so. If a relative date such as “next Friday” is ambiguous, retain the original phrase and ask for the intended date and timezone.
- Read the whole supplied excerpt and identify the decision it concerns.
- Group repeated messages about the same action so one commitment does not become several tasks.
- Separate explicit acceptance from inference. A mention, reaction or channel membership is not acceptance.
- Propose the smallest action and keep conditions attached.
- Ask the reviewer to confirm owner, date, source permission and whether a durable task is warranted.
- After an authorized task is created, retain its actual identifier and read it back. A draft table is not that receipt.
Fictional worked example: a deadline that nobody accepted
This example is invented. Casey writes, “Can we have the partner summary next Friday?” Jordan replies, “I can check the figures after finance confirms the final export.” Sam adds, “We could include a customer example if permission comes through.” None of those messages establishes an unconditional promise to deliver a finished summary on a particular date.
| Proposed item | What is supported | What remains open | Review action |
|---|---|---|---|
| Partner summary | Casey requested it | Accepted owner and exact Friday | Ask for acceptance and date. |
| Figure check | Jordan offered a conditional review | Finance export availability | Keep dependency attached; do not mark complete. |
| Customer example | Sam suggested an option | Consent and whether it is needed | Keep out of the task queue until decided. |
A weak extraction would assign all three items to Jordan and invent a deadline. A useful queue makes the missing decisions easier to answer.
Prompt for a human review queue
Use this with Aurora AI in Finta, Claude or Codex using only conversation material you are allowed to share. No chat connector is assumed.
Review this permitted conversation excerpt for commitments.
Purpose: [decision or project]. Source reference/date/timezone: [details]. Excerpt with speaker labels and surrounding replies: [text].
Return one row per distinct action or decision: source passage, request/accepted commitment/idea/status/conflict, explicit owner, explicit date, dependency, confidence and the human question needed.
Preserve conditional language. Mark unknown owners and ambiguous dates. Detect duplicate actions across messages.
Do not infer acceptance from mentions or reactions. Do not create tasks, change records, connect apps, read other channels or contact anyone. First return a review queue and ask me to confirm any next action.What makes the queue ready?
Each row should survive a reviewer asking who agreed, where they agreed and what would change the plan. Resolve the most consequential ambiguity first. Keep an item on hold if its source cannot be opened or the supposed owner disputes it. Record a correction without erasing the original evidence.
Common questions
Can I use Slack or Teams messages?
You can supply a permitted excerpt. This guide does not claim a tested Slack or Teams connection in Aurora. Verify any available connection separately; do not use another person's account or export an entire workspace to avoid a narrow access gap.
Should every action become a task?
No. Some entries are questions, ideas or resolved updates. A durable task is appropriate after its owner, meaning and timing are understood.
How is this different from meeting-note follow-through?
Chat unfolds across messages and can contain requests that remain unaccepted. This exercise centers on commitment formation. For a completed meeting, use meeting notes to CRM updates.
Keep the reviewed queue with the source packet. Aurora by Finta is the preparation surface; the people involved retain the commitments.