Daily Slack Digest and Priority Matrix

The Scenario

You're a senior programme manager overseeing three concurrent projects. By the time you sit down on Monday morning, there are 200+ Slack messages across eight channels waiting for you. Buried in the noise are things you actually committed to doing, decisions that need your input, and deadlines someone mentioned in passing at 11pm on Friday.

You need a system. Not a summary — those are easy and mostly useless. You need something that identifies your actual obligations, classifies them by urgency and importance, and tells you exactly what to focus on when you open your laptop.

Today, you're going to build that system: a scheduled Cowork task that runs every morning, scans your Slack, and produces an Eisenhower matrix brief. Then you'll configure dispatch to send it to your phone, so you know your priorities before you even sit down.

What You'll Learn

By completing this tutorial, you'll be able to:

• Design a Slack scanning configuration that separates signal from noise
• Write Eisenhower matrix classification rules that Cowork can apply consistently
• Build a multi-phase prompt that scans, extracts, classifies, and formats in sequence
• Configure scheduled tasks for automated daily execution
• Set up dispatch delivery to a mobile device
• Calibrate classification accuracy through iterative refinement

Prerequisites

• Claude Desktop with Cowork enabled (Pro or Max plan — Max recommended for scheduled tasks)
• Slack connector installed and authenticated in Cowork
• A Slack workspace with active channels you participate in
• Familiarity with the Eisenhower matrix (urgent/important quadrant system)
• A mobile device with the Claude app for dispatch testing

Step 1: Map Your Slack Landscape

Before building the automation, catalogue what Cowork will be scanning. List:

• Your channels: Every channel you're a member of, categorised as high-priority (project channels, leadership channels) and low-priority (social, general, random)
• Your direct message contacts: Key stakeholders who send you action items via DM
• Your commitments language: How do people assign you work? ("Can you handle this?" "@yourname please review" "Assigning this to @yourname" "You mentioned you'd...")

Create a file called digest-config.md with this information. Structure it like this:

## High-Priority Channels
- #project-alpha
- #project-beta
- #leadership-updates

## Low-Priority Channels
- #general
- #watercooler

## Key Stakeholders (DMs to check)
- Sarah Chen (VP Engineering)
- Marcus Webb (Product Lead)

## Commitment Patterns
- Direct mentions: @myname
- Assignment language: "can you," "please review," "assigned to," "action on"
- Commitments I made: "I'll," "I will," "I can handle," "I'll take that"

Checkpoint: You've got a digest-config.md that maps your Slack landscape, prioritises channels, and documents how work gets assigned to you.

Step 2: Design the Eisenhower Classification Rules

The Eisenhower matrix has four quadrants:

You need to teach Cowork how to classify Slack messages into these quadrants. Create classification rules in your digest-config.md:

Q1 — Do First (Urgent + Important):

• Mentions of deadlines within the next 24 hours
• Direct requests from leadership with time pressure
• Client-facing commitments or blockers
• Messages containing words like "ASAP," "today," "urgent," "blocking"

Q2 — Schedule (Important + Not Urgent):

• Project milestones due this week
• Review requests without immediate deadlines
• Strategic discussions requiring your input
• Messages containing "this week," "when you get a chance," "before Friday"

Q3 — Delegate (Urgent + Not Important):

• Requests that someone on your team could handle
• Administrative tasks with deadlines
• Meeting scheduling and logistics
• FYI messages that need acknowledgement but not your personal action

Q4 — Eliminate (Not Urgent + Not Important):

• General announcements already covered elsewhere
• Social messages and reactions
• Threads you were tagged in but have no action required

Checkpoint: Your digest-config.md includes classification rules for all four Eisenhower quadrants with specific signal words and patterns.

Step 3: Build the Digest Prompt

Write the master prompt that ties everything together:

Read the configuration in digest-config.md for channel priorities, stakeholder list, and classification rules.

Phase 1 — Scan: Search my Slack workspace for all messages from the last 24 hours (or since 5pm yesterday if it's a Monday morning) in the high-priority channels and from key stakeholder DMs listed in the config.

Phase 2 — Extract: Identify every message where:

• I'm directly mentioned or tagged
• Someone asked me to do something (using the commitment patterns in the config)
• I committed to doing something ("I'll," "I will," etc.)
• A deadline was mentioned that affects my projects
• A decision was made that I need to know about

Phase 3 — Classify: Place each extracted item into the Eisenhower matrix using the classification rules in the config:

• Q1 (Do First): urgent + important
• Q2 (Schedule): important + not urgent
• Q3 (Delegate): urgent + not important
• Q4 (Eliminate): not urgent + not important

Phase 4 — Format: Produce the daily brief as daily-brief-[date].md with:

1. Today's date and a one-line summary ("5 items need your attention today")
2. Q1 items first, with source channel, sender, and the original message text
3. Q2 items with recommended scheduling (e.g., "Handle by Wednesday")
4. Q3 items with delegation suggestions (who on the team could take this?)
5. Q4 items as a collapsed list (acknowledged but not actioned)
6. A "Commitments Made" section listing anything I said I'd do

Only include items that genuinely require my attention. Don't pad the brief with noise to look comprehensive. If nothing urgent happened, say so.

Checkpoint: Your prompt has four phases (scan, extract, classify, format) and references the config file for all parameters.

Step 4: Run the First Manual Test

Point Cowork at your working folder and run the prompt manually. Don't schedule it yet — test it first.

Review the output:

• Scan coverage: Did Cowork check all the channels in your config?
• Extraction accuracy: Open Slack and verify the items it found. Did it miss anything obvious? Did it include things that should've been filtered out?
• Classification accuracy: For each item in the matrix, ask yourself: "Is this in the right quadrant?" Count the misclassifications.
• Formatting: Is the brief readable in 5 minutes or less? Is the Q1 section immediately actionable?

Keep a tally: items correctly classified, items misclassified, items missed, items that shouldn't have been included.

Checkpoint: You've run the digest manually, verified accuracy against actual Slack messages, and counted classification hits and misses.

Step 5: Calibrate the Classification Rules

Based on your test results, update digest-config.md:

Common calibrations:

• Too many Q1 items: Your urgency signals are too broad. Tighten them — "ASAP" is genuinely urgent; "when you can" isn't.
• Missing commitments: Add new commitment patterns you didn't anticipate — "I'll circle back," "let me check on that," "I'll loop in [person]."
• Wrong quadrant: If strategic project updates keep landing in Q3, refine the importance criteria to account for project context, not just the message tone.
• Noise in Q4: If Q4 is full of dozens of items, add exclusion rules — "Ignore messages that are purely emoji reactions or single-word replies."

Run the prompt again after calibration and compare the results.

Checkpoint: Classification accuracy has improved measurably after calibration. Your before/after hit rates are documented.

Step 6: Set Up the Scheduled Task

Once the digest is producing reliable results, schedule it to run automatically. In Cowork, set up a scheduled task:

Schedule this task to run every weekday at 7:00am:

[paste your refined prompt here]

Verify the schedule:

• Confirm the task is set for weekdays only (not weekends, unless you work weekends)
• Confirm the time is correct for your time zone
• Confirm the output file will be saved with the correct date in the filename

Checkpoint: The scheduled task is configured and set to run on weekday mornings.

Step 7: Configure Dispatch Delivery

The brief is only useful if you see it before you start working. Set up dispatch to send the daily brief to your phone:

After producing the daily brief, send a dispatch message to my mobile device with the following:

• Subject: "Daily Brief — [date]"
• Body: The full text of the Q1 (Do First) section only
• Attachment note: "Full brief saved as daily-brief-[date].md on your desktop"

The idea is that the dispatch notification shows you only the urgent items. You can read the full brief when you get to your desk.

Test the dispatch by running the workflow manually and confirming the notification arrives on your phone.

Checkpoint: Dispatch delivers the Q1 summary to your phone. You've verified the notification contains the correct content.

Step 8: Validate Over Multiple Days

The real test is consistency. Let the scheduled task run for at least 3-5 business days without intervention. Each morning:

1. Read the dispatch notification on your phone
2. Open the full brief on your desktop
3. Spot-check 2-3 items against actual Slack messages
4. Note any classification errors or missed items

After a week, compile your observations:

• Average accuracy rate across the week
• Most common misclassification pattern
• Whether the brief saved you time versus your manual process
• Any days where the scheduled task failed to run (and why)

Add these observations to a validation-log.md file.

Checkpoint: You've got 3-5 days of automated briefs with accuracy notes, and a validation log documenting reliability and improvement areas.

Expected Output

Your deliverable is a running daily automation:

• digest-config.md — channel map, stakeholder list, classification rules
• daily-brief-[date].md — at least one (ideally several) generated briefs
• A configured scheduled task running every weekday morning
• Dispatch delivery to your mobile device
• validation-log.md — multi-day accuracy tracking

This system replaces your 45-minute Monday morning Slack trawl with a 5-minute phone notification. More importantly, it catches things you'd have missed — commitments buried in threads, deadlines mentioned in channels you rarely check, and action items assigned to you while you were offline.

The Strategic Value Beyond Time Savings

The Eisenhower matrix isn't just a productivity tool — it's an accountability system. When you know exactly what's urgent and important each morning, you stop reacting to whoever shouts loudest and start choosing where to spend your attention.

The "Commitments Made" section is particularly valuable for managers. It holds you accountable to your own promises — the "I'll look into that" responses you type in Slack at 4pm and forget by next morning. When the digest surfaces those commitments the following day, you either act on them or consciously decide to renegotiate. Neither option involves the silent default of letting things slide.

Over time, the digest also reveals patterns in how you spend your attention. If Q1 items are always dominated by the same project, that project may need more dedicated resources. If Q3 (delegate) items pile up week after week, you may not be delegating effectively. The digest isn't just a daily tool — it's a weekly and monthly mirror.

Extension Challenges

1. Weekly rollup — Build a second scheduled task that runs on Friday at 4pm, summarising all the week's briefs into a "What happened this week" document. Compare items from Monday's Q2 (Schedule) with Friday's actual outcomes.

2. Team digest — Modify the workflow to produce a team-level brief that consolidates action items across all team members, not just yours. This requires scanning for commitment patterns from each person on the team.

3. Sentiment layer — Add a sentiment analysis step that flags emotionally charged messages — frustrated clients, overwhelmed team members, escalation language. Include a "Mood Check" section in the brief that alerts you to interpersonal issues before they become problems.