Video summary

6 Hermes Agent use cases I promise will change your life

Main summary

Key takeaways

Technology

Summary of the video (6 Hermes Agent use cases)

The creator explains how Hermes Agent (AI agent software) can act like a 24/7 “AI employee.” They stress that most people won’t get the full benefit unless they use the product’s features correctly. They then share six practical workflows they claim improved their process—each centered on running autonomous tasks, iterating prompts, and connecting multiple tools/services.

1) Long-running tasks with /goal (unlocked via meta-prompting)

  • The creator describes /goal as an underrated way to get Hermes to handle very long tasks (they claim over 24 hours).
  • Warning: If you use /goal too generically (e.g., “build me an app”), results are often mediocre.
  • Solution: Use meta prompting—ask the AI to generate the best possible /goal prompt tailored to your specific project.

Example workflow

  1. Provide full details for the project (example: a 3D third-person shooter in Godot).
  2. Ask Hermes to generate an optimized /goal prompt.
  3. Hermes runs it and asks clarifying questions first.

Integration idea

  • After Hermes generates code, the creator uses other tools (mentions Claw Code / Codeex and Claude coder tooling) to help finish and polish outputs.

When to use

  • Complex apps, detailed changes, and long-form deliverables.

2) Start each day with a built-in Hermes Kanban board (autonomous task routing)

Hermes introduced a Kanban board feature that can route tasks to Hermes agents/sub-agents and run them.

How it works

  • Enter tasks into the board—especially triage—and Hermes auto-assigns and executes them.

How to access

  • Type “Hermes dashboard” in the terminal → open the provided URL → go to Kanban.

Morning routine described

  • Write a to-do list on paper.
  • Move Hermes-relevant items into the Kanban board, placing them in triage.
  • Example Hermes tasks:
    • Script a YouTube video about Hermes use cases
    • Generate tweets
    • Produce a report on trending AI tools
  • Non-automatable tasks (e.g., paying a credit card bill) are handled by the human.

Outcome

  • Hermes moves work forward (from triage into execution) and completes tasks autonomously, while the user handles human responsibilities.

3) Technical research & competitor app breakdowns via browser/console automation

For builders, Hermes can run full technical research on competing apps.

Claimed capabilities

  • Control a browser, click through pages, inspect the console, examine assets, and extract:
    • The tech stack
    • Notable features

Example

  • “Open creatorbuddy.io and do a full research and technical breakdown.”

What you get

  • A detailed report that may include:
    • High-level architecture overview
    • How the product works
    • Features to emulate
    • Payments/cost information and analytics/events collected (as described by the speaker)

Follow-on idea

  • Convert the report (e.g., into Markdown) and feed it into other coding agents/tools to recreate or improve similar functionality quickly.

4) Build and maintain a personal “memory wiki” (automated diary/knowledge base)

The creator recommends building a website that acts as a structured, clickable memory store for Hermes.

What it should include

  • Topics discussed with the agent
  • Daily logs (what was worked on)
  • Links to revisit prior conversations and outputs by topic or date

Purpose

  • Acts as an automated diary
  • Strengthens the agent’s memory system
  • Helps Hermes reference past work when starting new projects

Setup approach

  • Use a “copy/paste” prompt asking Hermes to build a memory wiki site with subjects and daily logs.

Key emphasis: structured access to past interactions helps Hermes retrieve context more effectively.

5) “General computer administrator” using Tailcale across devices

This workflow aims to let Hermes operate across multiple devices (phones/tablets/desktops).

Setup

  • Install Tailcale (described as free; not sponsored).
  • Connect all devices to a private network.

Result Hermes can function as a controller/dispatcher for tasks across machines, including examples like:

  • Fetching files from one device to another (e.g., getting a document to your phone)
  • Deploying or using workloads across devices (e.g., running a local LLM on a Mac Studio and using it from a MacBook)
  • Testing locally served apps from different devices by bridging “local host” via the Tailcale network

Framed as a “superpower” that makes multiple devices feel like one accessible environment.

6) A daily morning priority prompt for self-improvement & personalization

A proactive personalization step: Hermes messages you every morning (example time: 9:00 a.m.).

What Hermes asks

  • “What is your number one priority today?”

Then Hermes:

  1. Proposes tasks aligned to that priority
  2. Updates/improves its memories about you

Claimed benefit

  • Better context → better recommendations and work quality
  • Ongoing personalization over time

The creator provides a prompt template meant to be run daily, stressing consistency.

Closing / demo result

  • The creator references a game being built from the earlier /goal run.
  • Quick gameplay summary:
    • Stealth-like combat where enemies can detect through walls
    • Looting (ammo/bandages)
    • Gameplay loops such as shooting → looting → picking up items
  • Conclusion: The game isn’t perfect, but the agent-generated structure/assets can be iteratively improved (e.g., using an image generation model to create game assets).

Main speakers / sources

  • Speaker: The YouTube creator (single primary narrator; no other named individuals directly speaking in the subtitles).
  • Referenced partner/source: HubSpot (promotes a free “AI agents cheat sheet” guide).
  • Named tools/services mentioned: Hermes Agent, Tailcale, Claude Coder / Codeex / Claw Code, ChadGPT image generation model (as stated), and GoDot / Godot.

Original video