agentos.guide › blog

How To Learn Building AI Agents: Seven Builds That Turn Into One Agent OS

By Julian Goldie · 7 October 2026 · agentos.guide

How to learn building AI agents is to build seven small agents in order, so that each one teaches you a single concept.

You start with a scheduled daily recap, then an inbox triage agent, a news radar, a second brain, a skill taught from a book, a kanban build line and finally your own Agent OS dashboard.

I have built and published every one of these, so the commands and the numbers below are real.

Here is the ladder first, then a quick-start for tonight and the way all seven fit into one system.

The seven builds, in the order I would learn them

Here is the whole ladder on one screen, with the room each build becomes in an Agent OS.

An AI agent is a model that can use tools, follow a goal and keep working without you typing each step.

You will not understand that sentence properly until one of your own agents delivers something while you are asleep.

That is why I teach agents as builds and not as theory.

The method inside my community is the same every time, which is one agent doing one job on a free model first.

Quick-start: get rung one running tonight

These five steps take you from nothing to a working agent in one evening.

The install is one line in your terminal.

curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash

Hermes Agent is free and open source, and it comes from Nous Research.

If you prefer buttons to a terminal, my one-click Hermes guide covers the easier install.

Rung 1: a scheduled daily recap

You learn the basic loop here, which is a prompt that runs, does some work and delivers a result.

You need Hermes, one model and somewhere to receive the message.

hermes cron create "0 6 * * *" "Read my task list and send me a short to-do recap for today" --name "Daily 6AM To-Do Recap" --deliver telegram

The schedule also accepts plain forms such as 30m and every 2h.

In my Herald OS video I showed three of my own automations, which are a daily 6am to-do recap, a daily Japanese practice session and a morning inbox triage.

You are done when the recap has arrived three mornings in a row without you opening the app.

The prompt is what usually breaks, because Hermes asks for a self-contained prompt and beginners write one that assumes a chat history.

Delivery is the other weak point, so connect Telegram with hermes gateway setup or use --deliver local until you have.

In an Agent OS this build is the clock, which is the part that starts work with nobody at the keyboard.

Rung 2: an inbox triage agent

You learn tools and permissions here, which decide what an agent can reach and what it is allowed to do.

Hermes ships a built-in skill called email-inbox-triage, and its own description says it prioritises threads and drafts replies safely.

That skill names himalaya and google-workspace as its related mail skills, so you connect your mailbox through one of those first.

Then you attach the skill to a morning job.

hermes cron create "0 7 * * *" "Triage the last 24 hours of my inbox, list the five threads that need me, and draft replies without sending anything" --name "Morning Inbox Triage" --skill email-inbox-triage --deliver telegram

You are done when its five priority threads match yours for five mornings and nothing has been sent without you.

The mailbox connection usually breaks first, and vague rules for what counts as urgent come second.

I would keep the agent on drafts until you trust it, because a drafting agent cannot embarrass you.

In an Agent OS this is still the clock, but now the clock is holding a tool.

🔥 Want the exact learning to build AI agents setup I run inside my Agent OS?

Inside the AI Profit Boardroom you get the installable Agent OS, step-by-step video tutorials, four coaching calls a week and 3,400+ members building real automations.

→ Get access here

Rung 3: a research and news radar

You learn live data, structured output and how to run an agent from a script.

The one-shot flag, hermes -z, sends a single prompt and prints only the final answer.

My radar is one hermes -z run with a strict prompt, fired by a scheduled job at 6am.

It returns six signals, and each one has a headline, a heat score out of 100, the real post and a content angle.

The top three are pushed to my phone with hermes send.

hermes -z "Return the six hottest stories in my niche today with a headline, a heat score out of 100, a link and one content angle" --provider xai-oauth --model grok-4

Only xAI models carry the live X search tool, which is why that run is pinned to one provider and model.

My Hermes Oracle AI guide shows the receipts, which are 49 mornings logged, 294 trends scored and 46 pages published from those trends.

You are done when six scored signals with working links arrive every morning for a week.

Live search sometimes returns an empty sweep for no clear reason, and one retry on empty fixes most of those mornings.

One-shot mode also skips approval prompts, so keep that prompt narrow and read-only.

In my Agent OS this build has its own screen, where I can scroll back through every morning's trends.

Rung 4: a second brain with Obsidian

You learn memory here, which is how an agent stops answering like a stranger.

Obsidian is a free download, and Hermes has a built-in obsidian skill.

You make a new vault, hand Hermes the full folder path and ask it to build the folders with an index note in each.

You must give the real path, such as ./SecondBrain, because I skipped that once while filming and Hermes built the folders in its own workspace.

Then you write one short about page that covers who you are, what the business does and how you like to sound.

My vault held 614 notes when I measured it for the Hermes second brain guide, which is far too much to paste into a chat.

So the agent reads the about page first, then today's notes, then the project pages it needs.

You are done when a fresh chat answers a question about your business correctly and names the notes it read.

In my demo it listed all 11 notes it used.

Dumping every document into the vault is the common mistake, and so is letting several agents write into the same files.

In an Agent OS this build is the memory, and it is the piece almost everyone skips.

Rung 5: a skill taught from a book or SOP

You learn what a skill is, which is knowledge or a procedure the agent loads only when it needs it.

You use the /learn command inside a Hermes chat.

/learn ~/Documents/your-book-or-sop.pdf

Hermes reads the document once and writes a lean index, a reference file per chapter, a glossary and a cheat sheet.

After that, every question goes from the index to one chapter to the answer.

I tested it on Sun Tzu's The Art of War as an epub, and the walkthrough is in my /learn guide.

The engine's own testing on real books reports 24 to 51 times fewer tokens than pasting the book into the chat.

You are done when a brand new session answers three questions you already know the answers to, with no upload.

An old build is the trap that caught me, because book mode only ships in newer versions, so run hermes update first.

Your own onboarding SOP is a better first document than a huge book.

In an Agent OS this build is the skills shelf, and it grows every time you teach the agent something.

Rung 6: a kanban build line with several agents

You learn orchestration here, which means turning a big job into tickets and letting separate workers finish them.

Each worker is a Hermes profile with its own model and rules.

hermes profile create builder
hermes kanban boards create my-builds
hermes kanban create "Build a landing page for my newsletter" --assignee builder --max-runtime 30m
hermes kanban watch

The dispatcher runs inside the gateway now, so hermes gateway start is what gets workers picking up tickets.

In my Hermes Command Desk guide three tickets moved to done without me, which were a landing page, a content kanban app and an analytics dashboard.

Those three builds ran on free models, so my bill for them was $0.00.

You are done when three tickets reach the done column and you have opened each finished build yourself.

My first three dispatches all died instantly, and my kanban build line guide lists the three config traps behind that.

The first is a worker that boots a whole gateway and dies on a port conflict, the second is a provider name Hermes does not recognise, and the third is a base URL that gets silently ignored.

In my Agent OS this build is the ticket board, where I drop a card and a worker picks it up.

Rung 7: your own Agent OS dashboard

You learn system design here, which is how memory, agents and a task board share one home.

Claude Code is the builder, Obsidian is the memory, Hermes handles long jobs and Tailscale gives you phone access.

npm install -g @anthropic-ai/claude-code

You open a folder, type claude and describe the dashboard you want in plain English.

Claude builds it and gives you a local address, which is usually localhost:3000.

Then you look at it, say what is wrong and let it fix that.

That describe, review and refine loop is the entire skill, and every feature on my own dashboard was built with it.

My free Agent OS build guide gives the exact prompts in five steps, which are the dashboard, the memory vault, the agents, the kanban board and access from anywhere.

You are done when one page shows your agents, your memory and your board, and it opens on your phone.

Installing every agent on day one is the big mistake, so start with Claude Code plus one more.

Automating too early is the other one, and the order I teach is manual first, then assisted, then automated.

🔥 Want the exact learning to build AI agents setup I run inside my Agent OS?

Inside the AI Profit Boardroom you get the installable Agent OS, step-by-step video tutorials, four coaching calls a week and 3,400+ members building real automations.

→ Get access here

How the seven builds become one Agent OS

The ladder is not seven separate toys.

Rungs one to three are the clock and the eyes, so work starts on time and the system can see what is happening outside.

Rungs four and five are the memory and the skills, so every agent knows your business and your processes.

Rung six is the workforce, so a ticket can be finished while you do something else.

Rung seven is the house that holds all of it behind one address.

In my own Agent OS, Hermes sits behind thirteen doors of one dashboard, and every exchange saves itself to my Obsidian vault.

You do not need thirteen doors to start, because the first two rungs already give you an agent that works every morning.

Two tracks: the no-code harness route and the code route

There are two honest ways to climb this ladder.

A harness is the finished shell around a model, so the loop, the tools, the memory and the scheduler already exist.

An SDK gives you the parts, and you assemble the loop yourself.

These are the frameworks I confirmed are live today on their own documentation pages.

• The OpenAI Agents SDK is OpenAI's framework for building agents in code.
• The Claude Agent SDK is Anthropic's SDK, and it sits in the Claude Code docs.
• LangGraph is the agent framework from LangChain.
• The Agent Development Kit is Google's framework.
• smolagents is the agents library from Hugging Face.
• CrewAI is a framework for crews of agents.

I teach the harness route, so I am naming these from their own docs and not from builds I have published with each one.

My advice is to climb the first five rungs on the harness first, because each concept carries over when you rebuild the same projects in code.

How to debug an AI agent

Debugging an agent is mostly reading what it actually did, and Hermes has a command for each layer.

After that, you change one thing at a time and run the job by hand before you schedule it again.

Most of my own failures came down to a vague prompt, a missing connection or the wrong model for the tool.

A seven-week plan

In week one, install Hermes and get the daily recap arriving for three mornings.

In week two, connect your mailbox and run the triage agent on drafts only.

In week three, build the radar and let it run for seven days.

In week four, make the Obsidian vault and write the about page.

In week five, run /learn on one SOP and test it with questions you know.

In week six, create a board and ship three tickets.

In week seven, build the dashboard that holds the first six.

Where to go next

How to learn creating AI agents is the sibling guide in this series. How to learn AI agents covers the wider path before you build anything. The best AI agent course in 2026 is my ranked list if you want lessons alongside the builds. The best Hermes Agent use cases gives you more build ideas once the ladder is finished.

FAQ

What is the best way to learn building AI agents?

Build small agents in order instead of studying more theory. Start with one scheduled daily recap, then add an inbox triage agent, a news radar, a second brain, a learned skill, a kanban build line and finally a dashboard that holds them all.

Do I need to code to learn building AI agents?

No. The first six builds use Hermes Agent commands and plain English prompts. The code route, where you use an agents SDK such as the OpenAI Agents SDK or the Claude Agent SDK, is for people who want agents inside their own product.

What should I build first?

Build a scheduled daily recap first. One hermes cron create command with a schedule and a self-contained prompt gives you an agent that runs every morning, and it teaches the core loop of prompt, run and deliver.

How do the seven builds fit into an Agent OS?

The recap, triage and radar are the clock and the eyes. The second brain and the learned skill are the memory. The kanban build line is the workforce. The dashboard is the house that holds all of them behind one address.

How do I debug an AI agent?

Run hermes doctor to check the install, hermes logs to read the errors, and hermes cron runs or hermes kanban show to see what a run actually did. Then change one thing at a time and run the job by hand before you schedule it again.

Final word on how to learn building AI agents

Build rung one tonight and add one rung a week after that.

By week seven you will have a working Agent OS, and you will know from experience how to learn building AI agents.

Related guides on agentos.guide

→ How to Build Your Own AI Agent OS

→ The Hermes Command Desk: one AI employee in every room of the Agent OS

→ Hermes AI + Obsidian: Build The Ultimate AI Second Brain

→ The Book Brain Engine: Hermes can learn any book now

📺 Video notes + links to the tools 👉

🎥 Learn how I make these videos 👉

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

← all guides on agentos.guide