Skip to content
ToolDeck

Sales & Lead Capture

LinkedIn Reply Filter

Filters your LinkedIn DMs by intent so you only open the messages that matter.

Saves about ~2 hrs per week

What it does

  • Classifies each incoming LinkedIn message by intent: sales, spam, collab, client
  • Tags high-value messages so they surface first
  • Drafts a suggested reply for the important ones
  • Produces a daily digest of who reached out and why

How it works

  1. New LinkedIn messages pulled on schedule
  2. AI classifies each by intent and value
  3. Top messages get drafted replies
  4. Daily digest hits your inbox or Slack

What's included

  • Message classification
  • Reply drafts
  • Daily digest

Set it up yourself

Follow these 4 steps in order. You will need accounts for: LinkedIn, Claude.

Step 1: Grab your messages

  1. LinkedIn's official API is heavily restricted — most people use a browser-side integration or export.
  2. For the DIY path use the guide's approach: a scheduled export of your message inbox (LinkedIn → Messaging → download) or a logged-in automation via a supported connector.
  3. Put messages into a 'Messages Inbox' sheet with columns: From, Date, Message, URL.
  4. Test the export format by running it once manually.

Watch out: Respect LinkedIn's terms — pulling messages with unapproved automation can get your account flagged; do not store sensitive content beyond what you need.

Step 2: Define the categories

  1. Buckets: sales-inbound (someone wants to buy), partnership, outreach-spam (copy-paste), existing-client, apply-to-work-here, other.
  2. Write one-line rules per category so the AI has a rubric.
  3. Add priorities: sales-inbound and existing-client = high; spam = low.
  4. Note any senders on your blocklist/ignore list.
  5. Save the rubric.

Watch out: The rubric is what the classifier uses — vague buckets ('interesting' vs 'sales') produce useless output.

Step 3: Add the classification prompt

  1. Build: 'Classify this LinkedIn message. Return JSON: {intent, priority, isSpam, draftedReply}.'
  2. draftedReply only for high priority — short, in your voice.
  3. Copy-paste detection: 'if the message is generic enough to apply to anyone, mark isSpam true'.
  4. Paste prompt + API key into the workflow.
  5. Test on 10 messages from your inbox and tune the rubric.

Watch out: The copy-paste detection is the sanity check for this whole flow — without it spam floods the top of your digest.

Step 4: Schedule the digest

  1. Set a daily schedule (e.g. 8am) that processes messages received since yesterday.
  2. Write high-priority items (with drafted replies) to the top of the digest.
  3. Send the digest to your inbox or Slack channel.
  4. Add a weekly rollup: count by intent, reply rates if you track them.
  5. Keep the run time fixed every day — routine beats surprise.

Watch out: The digest replaces your instinct to check the DM tab — trust it for a trial week before relying on it entirely.

Frequently asked questions

What does the LinkedIn Reply Filter automation do?

Filters your LinkedIn DMs by intent so you only open the messages that matter. Classifies each incoming LinkedIn message by intent: sales, spam, collab, client

How much time does LinkedIn Reply Filter save?

It is estimated to save ~2 hrs per week. Actual savings depend on your volume.

Which tools does LinkedIn Reply Filter use?

It works with LinkedIn and Claude.

What does LinkedIn Reply Filter include?

Message classification, Reply drafts, Daily digest.

Can I set up LinkedIn Reply Filter myself?

Yes. The do-it-yourself guide on this page walks through 4 steps, from getting access to testing the finished automation.