Build an AI Employee for Your Creative Life

Most creators use AI like a vending machine — prompt in, output out. Hermes gives you an operator: memory, tools, scheduled routines, and a workspace that remembers what you're building.

Start Installing → Copy Prompts
Hermes Agent for Creators livestream Why AI Matters
A creator does not need more outputs. A creator needs an operation.

What You'll Build

Six concepts that turn Hermes from a chatbot into a creative operating system. Click any card to go deeper.

Concept 01

AI Employee

A persistent agent on your computer that remembers your projects, uses tools, and talks to you through Telegram — not a one-off chat session.

Onboard your agent → Concept 02

Creative OS

A folder structure that gives your agent a room to work: Projects, Scripts, Research, Inbox. For an agent, a folder is a workplace.

Build your home base → Concept 03

Daily Brief

One scheduled morning ritual: active project, smallest next action, one idea worth revisiting. Hermes stops being purely reactive.

Schedule your brief → Concept 04

Voice Keeper

Turn your taste into infrastructure. Hermes studies your writing and speaking samples and saves a reusable voice profile for future drafts.

Preserve your voice → Concept 05

Mission Control

A local dashboard showing active projects, content pipeline, deadlines, and today's brief — the same map for your agent and your brain.

Build your dashboard → Concept 06

Verification Loop

Agents say they're done — you verify. A short checklist before you trust packaging, research, files, or outreach Hermes produces.

Build your checklist →

Install Checklist

Work through these in order. Check them off as you go.

  • Install Hermes with the official script
  • Run hermes setup, hermes doctor, and hermes status
  • Choose a model/provider (see quick guide below)
  • Connect Telegram via BotFather token
  • Enable core tools: file, terminal, web, browser, skills, memory, cron
  • Paste the onboarding prompt — tell Hermes who you are
  • Create your Creative OS folder structure
  • Schedule your first daily creative brief
  • Assign one real project with a brief and memory file
  • Teach Hermes one reusable skill
  • Set safety defaults: manual approvals, secret redaction

Full Setup Walkthrough →

Model & Provider Quick Choices

Best Quality

Anthropic API — if you're comfortable paying for API costs. Best reasoning and writing quality for creative work.

Practical Default

OpenAI Codex OAuth / GPT-5.5 — where available. Strong all-around choice for creators already in the OpenAI ecosystem.

Budget / Experiment

Nous Portal — sign up, pick a tier, run hermes model and select Nous Portal. Good on-ramp for testing agent workflows.

Free-Tier Caveat

DeepSeek V4 via Nous Portal free tier — currently showing as a free option, but model availability changes. Verify live in the portal before promising it. Great for "try before you pay."

The model matters, but the workflow matters more. A perfect model with no memory, no files, and no routine is still just a fancy chatbot.

The Three-Layer Ladder

Layer 1

ChatGPT / Claude

  • Ask for output
  • One-off drafts
  • Useful, but disposable
Layer 2 — You Are Here

Hermes Agent

  • Remembers your work
  • Has tools & files
  • Schedules jobs
  • Runs workflows
Layer 3 — Your Habit

Verification Loop

  • You spot-check outputs
  • Match work to project rules
  • Open files & test links
  • Done ≠ "agent says done"

See the verification checklist →

Start Here Routes

New to Hermes?

Install, configure, connect Telegram, and onboard your agent in one pass.

Setup Walkthrough

Ready for workflows?

12 creator use cases: podcast prep, YouTube packaging, author operating room, and more.

Creative Use Cases

Want copy-paste prompts?

Onboarding, daily brief, project brief, skill builder — all ready to paste into Hermes.

Prompt Library

Security-minded?

Boundaries, backups, and judgment — not panic, not recklessness. Start safe.

Safety Guide

Verification Loop

When Hermes starts doing real work, the question shifts from can it do this? to how do I know it did it right? Run this loop yourself before you ship, publish, or send.

Done is not when the agent says done. Done is when you have checked the work against your standards.

Quality checklist (run every time)

  • Read the deliverable out loud — does it sound like you?
  • Compare against your project brief or MEMORY rules
  • Open every file path and link the agent referenced
  • Spot-check facts, names, dates, and claims
  • Confirm required fields are present (titles, tags, sources, next steps)
  • Ask: would I stake my reputation on this as-is?

Where to apply it

YouTube packaging

Does the title/description match the video? Any clickbait drift? Thumbnail prompt and tags complete?

Podcast / guest prep

Are sources real? Questions serve your audience? Risks and angles flagged honestly?

Research & drafts

Weak sources removed? Claims traceable? Voice and continuity match the project?

Files & dashboards

Artifacts open locally? Routes and paths correct? Nothing broken behind a "done" message?

Install Hermes first. Build one workflow. Then make verification a habit — not an afterthought.

Safety & approvals →