← News & Resources

Self-paced course · 15 days

Fifteen days to become an AI agent power user.

The model is the engine. The harness is everything that makes it useful: the loop that plans, acts and checks, the tools that reach your systems, the memory that knows you over time, the skills that repeat what worked, and the schedule that works while you sleep.

This hands-on course teaches all of it with Hermes Agent, the open-source agent harness from Nous Research. Every lab runs on your own work, so fifteen days from now an agent is doing part of your job.

Learning to master agents isn't rocket science, and we don't think it should be limited to companies and people with big budgets. We run this training for companies in person, where a week for a team easily costs $50,000, or $1,500 to $4,000 per person. So we made a lighter, self-paced version of our in-person course: most of the same material, at a fraction of the price.

Free for a limited time. A few weeks after you download it, we'll reach out to ask how it went.

How each day works

About 90 minutes a day, three parts every time.

01

Understand

What the feature is, how it works and what it changes, then one curated video.

02

Do

A 40–65 minute lab on your own work: steps, a done-when and a common failure.

03

Ship

One deliverable you keep. By day 15 these add up to a working agent setup.

The syllabus

Three weeks, from first install to a team of agents.

Week 1 · Days 1–5

Foundations

The harness, install, models, context and tools.

  1. 01What a harness actually isYou ship: your agent backlog: three ranked tasks to automate.
  2. 02Install and first sessionYou ship: a working install and a one-page command cheat sheet.
  3. 03Models and providersYou ship: your model policy: default, fallback and private.
  4. 04Context engineeringYou ship: SOUL.md and a context file for your main project.
  5. 05Tools and toolsetsYou ship: your toolset config and a research brief it produced.

Week 2 · Days 6–10

Capabilities

MCP, memory, skills, always-on hosting and scheduled jobs.

  1. 06MCP: your systemsYou ship: one connected system, scoped to read-only tools.
  2. 07MemoryYou ship: a memory audit: what it knows, what to correct, what to forget.
  3. 08SkillsYou ship: a skill for your first task that works on a fresh session.
  4. 09Always onYou ship: an agent you can message from your phone, locked to your account.
  5. 10Scheduled automationYou ship: your first task running unattended on a schedule.

Week 3 · Days 11–15

Operate at scale

Subagents, profiles, security, real workflows and self-improvement.

  1. 11Subagents and delegationYou ship: a reusable delegation template.
  2. 12Profiles and agent teamsYou ship: a two-profile team and one completed Kanban handoff.
  3. 13Security and governanceYou ship: a hardened config and a one-page risk register.
  4. 14Real workflowsYou ship: three tasks running, each with before-and-after minutes.
  5. 15Self-improvement and scaleYou ship: graduation evidence plus a 30-day roadmap.

Before day one

What you need.

A laptop
macOS, Linux or Windows, and comfort typing commands into a terminal. No server needed.
A model
Nous Portal, a subscription you already pay for, or an API key such as OpenRouter. At least 64K tokens of context.
A budget cap
Set a monthly spending cap with your provider before day 1. Local models cost nothing per task.

Run it with your team

Prefer to learn together?

The PDF is the self-paced version. We also run the course for teams, with live sessions and hands-on help setting up agents on your own systems.