Inbox triage gets described as an email problem, but the queue that needs attention might be LinkedIn messages, connection requests, or three channels you use for different relationships.

Getting it under control doesn't mean reaching zero. It means knowing the important things have been answered, delegated, or consciously left, and that a bulk cleanup hasn't buried something you needed.

Strawberry works inside the real, logged-in inbox with you. It can read the whole thread, use the calendar context, draft the reply in the right channel, and show you the exact controls it proposes to use.

Choose the queue and decide what done looks like

Start with one inbox and a manageable time window. You may want a priority view, drafted replies, delegation, cleanup, or a combination. Strawberry adapts to the service rather than assuming that email labels, LinkedIn connection states, and another provider's assignment controls mean the same thing.

Want to try it?

Ask your Strawberry companion: “Help me triage this inbox. Bring forward what needs attention, draft the useful replies, and show me any cleanup or delegation before taking action.”

Skill

Triage an inbox

Prioritize the selected email, LinkedIn, or other communication queue and prepare the right replies, delegation, and cleanup for review.

For a large or unfamiliar queue, begin with a varied sample. A few messages from important relationships, obvious noise, unresolved threads, and ambiguous cases reveal much more about the right sorting rules than a blind archive of everything old.

Read the thread before deciding what matters

An unread message is not automatically important, and an unanswered one is not automatically an obligation. Strawberry can inspect the full conversation and sent history, then bring in a meeting, calendar event, project record, or file when it changes the meaning of the message. That makes the draft reply more useful than one produced from the latest line alone.

Because this happens in the visible browser, you can inspect the exact thread, profile, account, and available controls. You can correct the classification or take over without trying to recreate the state from a separate AI summary.

Separate the priority decision from the action

A compact control view should lead with messages that could change what you do next. It can include why each item matters, a reply in your established voice, a delegation or follow-up suggestion, and a cleanup group supported by the selected service.

Reviewing that view does not approve every later step. Creating an external draft, sending it, accepting a connection, archiving a thread, muting a sender, unsubscribing, and deleting are different actions. Strawberry can carry your approved choice into the real inbox, then verify what changed.

Follow important work beyond the inbox

Some inbox items reveal a wider coordination problem: a promise made in a meeting is absent from the task system, or a response depends on a decision recorded elsewhere. That is the point to move from inbox triage into an open-loop audit rather than stretching one communication queue into a view of the whole company.

After a few reviewed passes, Strawberry can remember the priorities, exceptions, voice, delegation rules, and definition of done that actually work for this queue. If there is a useful cadence, a Routine can prepare the next priority view and proposed replies while keeping sends and provider changes reviewable.

Official Strawberry skill
Copy skill file Download skill file

Triage My Inbox

Help the user bring a communication queue under control. Success means the important work has been handled or clearly parked—not that every provider must show an empty screen.

Because Strawberry works in the browser, use the inbox where the work already lives. The user can see the same messages, profiles, threads, and controls, inspect what you are doing, and take over when useful. Connected apps can make this faster, but email is not the only inbox and an app connection is not the only way to begin.

1. Choose the inbox and what done means

Infer what is already clear, then confirm only details that could change the work:

  • the inbox, account, and time window;
  • whether the user wants orientation, replies, cleanup, delegation, or a combination;
  • important people, topics, commitments, and exceptions; and
  • which actions may be proposed or applied after review.

Start with one inbox or a small representative sample. Work across multiple inboxes only when the user asks for a combined view. Even then, keep the accounts, source states, and available actions clear rather than pretending every provider works like email.

2. Work inside the real queue

Use an approved connection or the visible, logged-in website. For email, inspect the relevant full thread and sent history. On LinkedIn, distinguish messages, connection requests, InMail, and the states the site actually exposes. For another queue, learn its own meaning of unread, pending, assigned, closed, muted, or archived before proposing changes.

Bring in related calendar, meeting, project, file, CRM, Slack, or other context only when it helps understand a sender, promise, decision, or response. Suggest a connection when it would materially improve the result; otherwise continue from the browser, uploads, or details the user provides.

If the selected queue is primarily customer support and the work depends on SLAs, refunds, escalations, help-center policy, or specialist support judgment, use the appropriate Customer Support workflow instead.

3. Find what really needs attention

Adapt the sorting to the source and the user's work. Look for:

  • important incoming work and messages that need a reply;
  • decisions, approvals, introductions, or promised follow-ups;
  • work that can be delegated or routed to someone else;
  • recurring noise and low-value items the user may want to reduce; and
  • ambiguous, private, sensitive, or consequential items that need the user's judgment.

Do not treat unread as important, an unanswered message as an obligation, or a connection request as something to accept by default. Check whether a thread was resolved elsewhere before flagging it. When a commitment reaches beyond this inbox into meetings, tasks, projects, or other systems, use strawberry/operations/close-open-loops for the wider audit.

4. Give the user a useful control view

Lead with the messages that could change what the user does next. A useful first result may include:

  • a short priority view with the reason each item matters;
  • replies drafted in the right channel and the user's established voice;
  • clear delegation or follow-up suggestions; and
  • a reviewable cleanup group for archive, label, mute, unsubscribe, accept, decline, delete, or another action the selected service actually supports.

Keep the visible result compact. Separate the current provider state, your recommendation, and the action that would be taken. For a large inbox, calibrate on a varied sample before expanding.

5. Apply only the approved actions

Show the account, people or records, channel, action, and scope before making consequential changes. One approval may cover a clearly defined batch; ask again when the scope or expected effect changes.

Use the controls and concepts of the selected service rather than translating email labels into LinkedIn or another provider. If an action is unavailable, say what can be done instead. After applying an approved batch, verify the result and report anything that stayed unresolved.

Drafting, creating an external draft, sending, accepting or declining a connection, archiving, muting, unsubscribing, and deleting are distinct actions. Do not treat approval of the priority view or reply wording as approval of all later steps.

6. Make the next pass faster

Use feedback from a real batch to remember useful priorities, exceptions, voice, delegation rules, and what counts as done for this inbox. Do not turn one unusual day into a permanent rule. When the method should help others, offer to save the accepted approach as a team skill or use strawberry/operations/set-up-shared-team-workflow for shared ownership.

When the queue has a useful cadence, a Routine may check the agreed inbox and time window, prepare a short priority view and reply suggestions, and surface proposed cleanup or delegation. Keep sends and provider changes reviewable unless the user explicitly approves a narrower recurring action. Stop when the account or inbox is unclear, a person or thread cannot be matched reliably, private context could cross a boundary, or the next action needs the user's judgment. Skip the Routine when on-request help is enough.