00:00If you are still typing frantic notes during every client call, you are playing the game the hard way.
00:06The most effective entrepreneurs operate differently.
00:10They hand the busy work over to an AI meeting co-pilot, allowing them to step back and actually direct
00:17the workflow.
00:18Today, we're building exactly that.
00:21A system that reads meeting transcripts, identifies follow-up requests, and automatically books the next meeting.
00:28You'll need four tools.
00:30Fathom to record, N8N as the hub, OpenAI to analyze, and Google Calendar to schedule.
00:37Setting up this architecture takes a few minutes, but once it is live, you have a permanent, scalable asset that
00:44reclaims hours of your management time every single week.
00:48Step one is securing your data feed.
00:51Open your Fathom preferences and turn on the auto-record toggle.
00:55This ensures every Zoom or Google Meet session is logged automatically.
01:00The moment you end that live call, Fathom generates a full transcript and a clean summary of your conversation.
01:07Now we need a place to catch that output.
01:10We will use N8N as the central nervous system to route this data.
01:15Open a workflow in N8N and add your first node.
01:19Search for the Fathom trigger and drop it onto the canvas.
01:23Before building the rest, you must hit test.
01:27Pulling in sample transcript data proves the connection works.
01:31If you skip this, you are building in the dark.
01:34A verified data feed is the absolute requirement for any reliable automation sequence.
01:41With the data secured, we must process it.
01:43An AI agent node will act as the cognitive center of this operation.
01:49Click and drag the output from your Fathom trigger directly into the input of the AI agent node.
01:55Next, we give the agent its instructions.
01:58Type, review the transcript to identify if a follow-up meeting is requested.
02:03If yes, extract the proposed date, time, and the attendees' email addresses.
02:09You will authenticate this node using your OpenAI API key.
02:13Treat this key like a bank password and never share it publicly.
02:17By giving it precise rules, you have taken a passive stream of text
02:22and turned it into an active decision maker capable of understanding human intent.
02:28Our agent can read, but it cannot act yet.
02:31To fix that, we introduce agentic tools.
02:35Plugins that let the AI reach outside of its own node to affect other software.
02:40Search for the Google Calendar tool and dock it straight into your AI agent.
02:45You will need to authorize N8N to communicate with Google servers.
02:50Click through the standard OAuth prompts to connect your account.
02:53Warning, pay close attention here.
02:55You must explicitly check the box that grants right access to create events,
03:00or the workflow will fail.
03:02Finally, map the original calendar invite data from your trigger into the agent's context.
03:08This ensures that when a speaker says,
03:10Let's meet with Sarah, the agent maps that name to her actual email address.
03:15Granting this right access officially bridges the gap between passive AI analysis
03:21and active autonomous delegation.
03:24The build is complete.
03:26Trigger the final end-to-end test to verify the system works.
03:30In the background, the AI evaluates the transcript,
03:34confirms the scheduling intent,
03:36and automatically executes the calendar tool.
03:39Just like that, the new meeting block appears on your schedule,
03:43perfectly populated with the correct attendees and time.
03:47True passive income and business scale
03:50come from replacing busy work with intelligent systems.
03:54If you are ready to build more leverage-driven workflows,
03:58subscribe to the channel,
03:59and I will see you in the next video.
04:01Peace be with you in the next video.
04:02Bye-bye.іт's
Comments