The Decision Maker's Second Brain

I was on a call with a consultant who manages 25 simultaneous projects. Not small ones: large-scale transformations, enterprise RFPs, multi-country implementations at a company with twelve billion dollars in annual revenue. His commercial team is 250 people. His job is to hold the thread across all of it.

He opened by telling me about a system he'd built. Not a productivity hack.

A system, engineered, layered, deliberate, that does for his memory what no single tool had ever managed to do.

He calls it his second brain.

Every client meeting he records with Meetily. The raw transcript lands on his device. From there, an automation he built picks it up, feeds it through a local AI environment, and converts it into a structured note, the kind of entry he'd want to read before walking into the next meeting with that client. That note goes into Obsidian(opens in new tab), where he's built a knowledge graph(opens in new tab): one node per client, one per project, linked by context, decisions, commitments, open threads.

He also maintains a database of action items pulled from those transcripts (not just his own, but items he's cascaded to his team), each with a deadline and an owner. A personal dashboard sits on top of it, showing him what's overdue, what's due this week, and what's slipping.

The whole system runs on his device. Nothing touches a cloud server he doesn't control.

I asked him why he'd built it around Meetily rather than one of the more polished, cloud-based AI meeting tools.

"The data doesn't leave the device," he said. "That matters when you're working on client projects under NDAs, when you're in commercial discussions that haven't closed, when what gets said in the room needs to stay in the room."

This is the constraint that most AI tools are not designed around. The people with the most sensitive conversations, the ones who most need to capture and organize what gets said in their meetings, are exactly the people who can't send that data somewhere they don't control. The more important the meeting, the more this matters.

What Meetily gives this consultant is not just transcription. It is the infrastructure layer on which a genuinely private second brain can be built. The transcript is the raw material. What you build on top of it is up to you.

The simplest version is the most accessible: every meeting captured, every transcript available, AI summaries generated locally using an open-source model running on your own machine. No internet connection required. No data in transit. You end every call with a document that tells you what was decided, what was agreed, and what you need to do next.

The more sophisticated version is what this consultant built. You change the folder where Meetily stores its output to a location your other tools can read: your Obsidian vault, a shared team drive, a connected database. From there, you attach whatever AI you trust most: a local model if you want zero external exposure, your own private cloud model if you have one, a large language model running behind your own API endpoint(opens in new tab) if you want more capability. You build the automation that turns a raw transcript into a structured note, an action item list, a project status update, whatever your specific workflow demands.

When I told him we were building an API(opens in new tab) and webhooks, he immediately understood what it meant. Right now he has a daily import job: a script he built himself that picks up raw transcripts and pushes them through his pipeline.

With the API, that becomes native. The transcript triggers a webhook. The webhook fires the automation. The whole pipeline runs without a daily import script holding it together, and every step of it stays within the infrastructure he controls.

He told me he'd started recommending Meetily to colleagues. Not because anyone asked him to. Because he'd watched what it did to his own cognitive load and wanted others to have it.

Twenty-five projects. One head. One system that actually helps carry the weight.

Meetily Workflows are built for leaders exactly like him.

Frequently Asked Questions

What is a second brain for meeting notes?

A second brain is a personal system for keeping what you learn outside your head, so you can find it when you need it.

For meeting notes, it means keeping conversations, decisions, and action items in one place, linked to the right client or project. You can look back at what was discussed, what was agreed, and who needs to do what.

How do I build a second brain like this with Meetily?

You can connect Meetily to Obsidian and an action-item tracker with a small script. Here’s how it works:

1. Meetily records the meeting and transcribes it on your own Windows or Mac computer. No bot joins the call.
2. When the summary is ready, Meetily sends your script a signed webhook, an automatic notice that carries the meeting ID but not the notes.
3. The script gets the meeting title, summary, and timestamped transcript from the local Meetily API. For a quick one-off, the meetily-pro CLI does the same in one command: meetily-pro meetings export <meeting-id> --format md.
4. An AI model of your choice, either a local model or a provider you pick with your own API key, turns the export into a structured note and a list of action items.
5. The script saves the note in your Obsidian vault and adds the action items to your tracker, each with an owner and a deadline.

Meetily must be running for this to work. The Meetily API is off by default, so turn it on and set up your own script.

The Obsidian recipe walks you through the setup. The Meetily API docs(opens in new tab) have the full reference, the Meetily CLI guide covers quick commands, and the meetily-workflows GitHub repository(opens in new tab) has example scripts.

Does Meetily work with Obsidian?

Yes. Meetily exports meeting notes as Markdown files that you save straight into your Obsidian vault, with no plugin needed.

A webhook tells your script when a summary is ready, and the script gets the note from the local Meetily API(opens in new tab) (or the meetily-pro CLI), so each note lands in your vault without copying and pasting. Meetily must be running, and your own script handles saving the file.

Follow the Obsidian recipe for the setup instructions.

Can I track action items from meeting transcripts automatically?

Yes. Meetily’s summaries include action items, and your script can fetch them from the local Meetily API(opens in new tab) as JSON, a structured format that’s easy for scripts to work with.

The script can then add those items to your own database or task tracker, with an owner and a deadline for each one. You can then use a dashboard to see what’s overdue, what’s due this week, and what’s slipping.

Is it safe to use AI note takers for confidential meetings?

It depends on where the recording, transcript, and summary go. Check that against your company's confidentiality policy and any client agreements, such as NDAs. Meetily records and transcribes on your own computer by default, and no bot joins the call.

Summaries made with a local model stay on your machine. If you use an AI provider with your own API key instead, the transcript text goes to that provider. See how each AI provider handles your data.

The Meetily API is off until you turn it on. By default it only listens on your own computer, and webhooks carry a meeting ID, not the notes.

Can Meetily summarize a meeting without an internet connection?

Yes. Once you’ve downloaded a local AI model, Meetily can generate meeting summaries on your own computer without an internet connection.

If you use an AI provider with your own API key instead, only the transcript text is sent to that provider.

See how to choose a summary model for more details.

Can I ask Claude or Cursor about my past meetings?

Yes. The Meetily MCP server, built on the same Meetily API, lets AI assistants connect to your meeting data.

You can connect Claude Desktop, Claude Code, Codex, Cursor, or Windsurf, choose what each one is allowed to do, and ask questions like “What did we decide last week?”

Most assistants send what they read to their own AI provider, so only connect the ones you already trust with your meeting notes.