Automate Meeting Notes Summaries with Gemini AI and Slack Notifications

Scheduled trigger → fetch meetings from MeetGeek API → split and format → Gemini generates summary and action items → restructure to Slack Block Kit → send channel notification

Who

Teams needing automated daily meeting briefs, project managers, remote or hybrid teams using Slack

You get

Automatically fetch meeting data daily, generate summaries and action items with Gemini AI, and push formatted Slack Block Kit notifications to a designated channel

You need

n8n · slack

Steps

1. Configure schedule trigger and meeting list API

   In the n8n workflow, add a Schedule Trigger node and set the run interval (e.g., daily). Then add an HTTP Request node (named Get meetings list in the workflow), set URL to https://api.meetgeek.ai/v1/meetings/, choose predefinedCredentialType with nodeCredentialType httpBearerAuth, and enter your MeetGeek API bearer token.

2. Split meeting data and rename fields

   Use a Split Out node to split the meetings array into individual items. Then use an Edit Fields (Set) node to rename meeting_id to extract_meeting_id (value: {{ $json.meeting_id }}) for downstream reference. Use a Date & Time node to format timestamp_start_utc as yyyy-MM-dd while keeping other input fields.

3. Filter meetings from the past 7 days

   Add an AI Transform node with the following auto-generated code to filter dates: read all input items, compute today and 7 days ago, keep only items where formattedDate falls within the range. The node instructions are "Filter the formatted Date in the past week upto today".

   Execution prompt · Source text。AI Transform node instructions field

   Copy prompt

   Filter the formatted Date in the past week upto today

   Execution prompt · Source text。AI Transform node jsCode field (auto-generated by n8n from instructions)

   Copy prompt

   const items = $input.all();
   const today = new Date();
   const oneWeekAgo = new Date();
   oneWeekAgo.setDate(today.getDate() - 7);
   
   const filteredItems = items.filter((item) => {
     const itemDate = new Date(item?.json?.formattedDate);
     return itemDate >= oneWeekAgo && itemDate <= today;
   });
   
   return filteredItems;

4. Generate summary and action items with Gemini AI Agent

   Add an AI Agent node, set promptType to define, and paste the full prompt into the text field. Connect a Google Gemini Chat Model node as the language model for the AI Agent. Also add an HTTP Request Tool node (named summary in the workflow), set URL to =https://api.meetgeek.ai/v1/meetings/{{ $json.extract_meeting_id }}/summary/, authentication to httpBearerAuth, so the AI Agent can call it as a tool.

   Execution prompt · Source text。AI Agent node text field (full prompt when promptType=define)

   Copy prompt

   =# 📝 Meeting Notes Summarizer with Action Items
   
   You are an AI assistant that **summarizes meeting notes** and **extracts clear action items**.
   
   ---
   
   ## 🔹 Input Data
   - **meeting_id**: `{{ $json.extract_meeting_id }}`
   - **title**: `{{ $json.title }}`
   
   ---
   
   ## 🔹 Task Instructions
   1. Call the **Meeting Note Summary API** using the **HTTP Request node**, passing `meeting_id` as a path parameter.  
   2. For each API response, generate a summary for the corresponding meeting.  
   3. Provide a **concise summary** in **3–5 bullet points**.  
   4. Extract a structured **Action Items** list with:
      - **Task description**  
      - **Responsible person** (if mentioned, otherwise `Unassigned`)  
      - **Deadline** (if mentioned, otherwise `No deadline`)  
   5. Ensure each meeting summary **includes the meeting title** (`{{ $json.title }}`).  
   6. If the meeting notes are **empty or missing**, still return the JSON format but use:
      - `"summary": ["No summary available"]`  
      - `"action_items": []`  
   
   ---
   
   ## 🔹 Output Format (JSON)
   ```json
   {
     "title": "{{ $json.title }}",
     "summary": [
       "point 1",
       "point 2",
       "point 3"
     ],
     "action_items": [
       {
         "task": "Task description",
         "owner": "Person name / Unassigned",
         "deadline": "Date / No deadline"
       }
     ]
   }

5. Restructure output to Slack Block Kit

   Add a Code node (named Restructure to slack block in the workflow), input is the array of AI Agent outputs. The node iterates over each note, cleans and parses the JSON, and builds a Slack Block Kit blocks array with header (meeting title), section (summary bullets), section (action items with task/owner/deadline), and divider. Falls back to default text if parsing fails.

   Execution prompt · Source text。Code node (Restructure to slack block) jsCode field

   Copy prompt

   // Input: array of note-taker outputs
   const notes = $input.all().map(item => item.json);
   
   // Slack blocks array
   let blocks = [];
   
   notes.forEach((note, index) => {
     // Clean and parse JSON inside code block
     const cleaned = note.output.replace(/```json|```/g, '').trim();
     let parsed;
   
     try {
       parsed = JSON.parse(cleaned);
     } catch (err) {
       parsed = { title: `Meeting ${index + 1}`, summary: ["Could not parse notes"], action_items: [] };
     }
   
     // Meeting header with title
     blocks.push({
       type: "header",
       text: {
         type: "plain_text",
         text: parsed.title || `Meeting ${index + 1}`
       }
     });
   
     // Summary section
     if (parsed.summary && parsed.summary.length > 0) {
       blocks.push({
         type: "section",
         text: {
           type: "mrkdwn",
           text: "*Summary:*\n" + parsed.summary.map(s => `• ${s}`).join("\n")
         }
       });
     }
   
     // Action items section
     if (parsed.action_items && parsed.action_items.length > 0) {
       blocks.push({
         type: "section",
         text: {
           type: "mrkdwn",
           text: "*Action Items:*"
         }
       });
   
       parsed.action_items.forEach(item => {
         blocks.push({
           type: "section",
           text: {
             type: "mrkdwn",
             text: `• *Task:* ${item.task || "N/A"}\n   *Owner:* ${item.owner || "Unassigned"}\n   *Deadline:* ${item.deadline || "No deadline"}`
           }
         });
       });
     } else {
       // If no action items
       blocks.push({
         type: "section",
         text: {
           type: "mrkdwn",
           text: "_No action items available_"
         }
       });
     }
   
     // Divider
     blocks.push({ type: "divider" });
   });
   
   return { blocks };

6. Send Slack channel notification

   Add a Slack node (named Send a message in the workflow), set messageType to block, blocksUi to ={{ '{ "blocks": ' + JSON.stringify($json.blocks) + ' }' }}, channelId to your target channel (e.g., #meeting-notes), authentication to oAuth2 with Slack OAuth2 credentials. Set executeOnce to true to control send behavior. [First-test safety] Before the first test, point channelId at a test channel or a DM to yourself, or temporarily disconnect the Slack "Send a message" node; only reconnect the production channel after Block Kit output looks correct.

You are building an n8n workflow: scheduled fetch of meeting list from MeetGeek API, Gemini AI-generated summary and action items, then push as Slack Block Kit to a channel. First confirm you have an n8n instance, MeetGeek API bearer token, Google Gemini (googlePalmApi) credentials, Slack OAuth2 credentials, and the target Slack channel ID.