Hermes Second Brain: 10X Smarter AI (Free Setup)

A Hermes second brain made my AI agent 10X smarter overnight — and it's free to set up.

Most AI agents have one big problem: they forget you.

Every conversation starts from zero.

You re-explain who you are.

What you're working on.

What you've already tried.

The Hermes second brain setup fixes that — using two free tools that record your context automatically and feed it to your agents.

By the end of this post, you'll have:

A Hermes agent that knows you better than you know yourself.
Automatic memory from your screen and microphone (OMI).
A living wiki of all your knowledge (Obsidian).
All synced to OpenClaw, Claude, and any local agent.

🔥 Want my full Hermes second brain stack with the AIO setup? Inside the AI Profit Boardroom, I share my exact OMI + Obsidian + Hermes config with the prompts I use for memory recall. Plus a 2-hour Hermes course, weekly live coaching, and 3,000+ members already using this. → Get the stack here

The Two Free Tools That Power The Hermes Second Brain

Two tools.

Both free.

Both open source.

OMI — records your screen and microphone all day, generates memories automatically. Free for the use case I'm showing.

Obsidian — local vault that stores all your notes and memories. Fully free.

Together they create a "second brain" — an external memory that any AI agent can read.

Why Hermes Second Brain Specifically

Hermes is the agent I run because:

It's free, open source.
It runs locally — privacy matters when you're feeding it deep context.
It supports memory profiles natively.
It plugs into any LLM (cloud or local).

If you're brand new to Hermes, this is the setup that makes everything else click. I cover the broader install in How To Setup Hermes Agent.

Step 1 — Install OMI

Go to omi.com (or wherever OMI is hosted).

Download the Mac app.

Install like a normal app.

Launch it.

OMI runs in the menu bar. It can:

Record your microphone (for conversations).
Record your screen (for visual context).
Auto-generate memories.
Sync to apps like Obsidian.

Privacy note: you can disable any of these at any time. I keep screen recording on, microphone selective.

Step 2 — Install Obsidian

obsidian.md — download, install. Free forever.

Create a new vault. Mine is called "Memory Vault".

That's where OMI's memories will land.

Step 3 — Connect OMI To Obsidian

Inside OMI, go to Apps.

Find Obsidian.

Click connect.

Point at your Obsidian vault folder.

OMI starts writing memories into the vault automatically.

I had over 650 memories within a couple of weeks of setup.

That's the second brain.

Step 4 — Point Hermes At The Vault

This is the unlock.

Inside Hermes (terminal or workspace), give the command:

"Use my notes from the Obsidian vault for your memory, particularly from OMI."

Then point Hermes at the file path of your Obsidian folder.

To get the file path on Mac: open Obsidian → right-click the vault folder in Finder → Get Info → copy the file path.

Paste that into Hermes config.

Done.

Hermes now reads from your second brain on every prompt.

How To Get The Most From OMI

Watch this earlier walkthrough I did on the OMI + Obsidian combination — the same memory pattern, slightly different angle:

That covers the wiki side of things — how OMI structures memories Karpathy-style.

What Hermes Knows About You After This

After a couple of weeks of OMI running, Hermes knows:

What time you sleep.
What music you like.
What projects you're working on.
Who's in your team.
Your business goals.
Your tone of voice in conversations.
Specific decisions you've made and why.

That level of context turns a generic AI agent into your AI agent.

Every reply gets sharper.

Every recommendation gets more relevant.

Use With OpenClaw Too

The same vault works for OpenClaw, Claude, and any local agent.

Inside OpenClaw, give the same command:

"Use this file path for context and memory: [Obsidian vault path]"

OpenClaw pulls in the same memories.

I cover the broader OpenClaw setup in OpenClaw Computer Use and ClawX OpenClaw.

Use With Claude Desktop

Same trick.

Tell Claude to read from the Obsidian path.

It treats your vault as additional context for every conversation.

This pairs perfectly with the Claude Code SEO Agent workflow if you want personalised SEO automation.

What OMI Records (And What It Doesn't)

Be specific about privacy.

OMI captures:

Audio from your microphone (you choose when).
Screen captures (you choose what's excluded).
Memory summaries.
Tasks and goals it sees.

OMI does NOT capture:

Anything when you've disabled it.
Passwords (you can exclude).
Specific apps if you blacklist them.

I recommend excluding password managers + sensitive apps.

🚀 Want a full second brain playbook? Inside the AI Profit Boardroom, I share my OMI privacy config, Obsidian vault structure, and Hermes prompts that actually use the memory effectively. Plus the 2-hour Hermes course + daily training drops. 3,000+ members. → Join here

The Karpathy Wiki Pattern

Andrej Karpathy talks about a personal wiki where every conversation links to other knowledge — a living document that gets smarter every day.

Doing this manually takes too long. Most people give up.

OMI + Obsidian + Hermes automates the whole thing.

You go about your day.

Memories accumulate.

The wiki grows.

Your AI agent gets smarter.

That's the loop.

Multi-Layer Memory Setup

Three layers worth knowing:

Layer 1 — Capture. OMI captures everything you say and do across devices.

Layer 2 — Structure. OMI converts captures into summaries, decisions, action items, memories.

Layer 3 — Agent context. AI agents query the structured knowledge via Obsidian or directly via OMI's MCP server.

The third layer is where it gets really powerful — see this earlier walkthrough on the agent context piece:

That's the broader Hermes framework for 2026 — second brain is one piece of it.

Setting Goals With OMI

OMI also tracks goals + tasks.

You can say goals out loud and OMI captures them.

Hermes reads these goals when it suggests actions.

So instead of asking "what should I do?", Hermes already knows your top priorities.

What If OMI Fills Up

OMI's free tier limits messages, not memories.

For the second brain use case, you barely talk to OMI directly.

You're using it as a recorder + writer to Obsidian.

So the free tier covers most users indefinitely.

Common Setup Issues

1. Hermes can't read the vault path.

Make sure the file path is exact.

Copy from Finder Get Info, not typed.

2. OMI not writing to Obsidian.

Check the apps section in OMI.

Re-authorise the connection if needed.

3. Hermes ignoring the memory.

Be specific in your prompt. Tell Hermes to USE the memory, not just read it.

Example: "Before answering, read my OMI memories from [path]. Use that context for every response."

Daily Reality

What it looks like once running:

8am — open Hermes. It already knows what I'm working on this week.
Throughout day — OMI captures meetings, decisions, screen context.
Evening — Hermes' suggestions are tailored to my actual day.

I don't re-explain anything.

The agent already gets it.

Why This Beats Other Memory Solutions

Most "AI memory" tools require manual input.

You write notes.

You tag them.

You re-read them.

OMI + Obsidian + Hermes is automatic.

You go about your day.

The memory updates itself.

Your agents get smarter without effort.

That's the right level of automation.

What I'd Tell Someone Setting Up Today

Three pieces of advice.

1. Start with Obsidian first.

Get familiar with note vaults before adding OMI.

2. Be selective with OMI recording.

Don't enable everything on day one. Add screen recording first, microphone second.

3. Test your Hermes prompts.

Try a few queries that should trigger memory recall.

If Hermes isn't using the vault, refine your prompt.

Daily Workflow

What my actual day looks like with this stack:

Wake up — OMI is running.
Throughout day — meetings, calls, decisions all get captured.
Evening — Hermes suggests next-day actions based on the day's memories.
Weekly — review Obsidian vault for patterns.

Used to spend hours catching up on what I'd done.

Now it's automatic.

🔥 Want my full Hermes second brain setup? Inside the AI Profit Boardroom, I share my exact OMI privacy config, Obsidian vault structure, Hermes memory prompts, and a 2-hour Hermes course. Plus weekly live coaching where I'll set this up with you on a screen-share. 3,000+ members. → Get the setup here

FAQ — Hermes Second Brain

Is the Hermes second brain free?

Yes — Hermes, Obsidian, and OMI's basic tier are all free.

How long does setup take?

20-30 minutes including OMI install and Hermes config.

Will OMI record everything I do?

Only what you allow. You control microphone, screen, and excluded apps.

Can I use this with OpenClaw too?

Yes — same vault works for OpenClaw, Claude, and any local agent.

How much memory can Hermes hold?

Practically unlimited via Obsidian vault.

Hermes pulls in only the relevant context per query.

Will my data leave my machine?

OMI processes locally by default. Memory stays in your Obsidian vault on your machine.

What happens if I uninstall OMI?

Your Obsidian vault stays. The memories are yours.

📺 Video notes + links to the tools 👉

🎥 Learn how I make these videos 👉

🆓 Get a FREE AI Course + Community + 1,000 AI Agents 👉