• EDIT: for the the easiest possible introduction to get started using AI to build software applications, in just a few minutes, on any computer you already have, do this: https://aibynick.com/thread/29

If you're overwhelmed by where to start with all the current goings-on in AI, one the best entry points right now is to get to know Hermes agent.

Get an inexpensive VPS at Contabo or OVH for ~$50ish per year. If you're not comfortable with VPS, Linux, SSH, etc., ask Google, ChatGPT or any other LLM chat how to get started using a VPS and SSH, and take a look at https://com-pute.com/nick/linux_server_basics.txt

Open an account with Openrouter and set up an API key. This lets you use any of the hundreds of most popular commercial and open source LLMs, in your own locally hosted AI tools (those models are the 'brains' used by agentic AI harness applications such as Hermes).

Log into your VPS server with SSH, create a tmux session to run hermes, and run:

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

Have your OpenRouter API key ready, and I'd suggest choosing google/gemini-3.1-flash-lite-preview as the model to start with, when the installation asks. Google/gemini-3.1-flash-lite-preview is extraordinarily intelligent, it's extremely fast, and it's by far the least expensive choice among the frontier models, to serve as the brain of all your agentic tools. You don't need to pay for the costly fees associated with the Anthropic Claude models, or for any of the bigger, more expensive models, to do 99% of what you'll likely want to do when you're getting started with agentic workflows. Google/gemini-3.1-flash-lite-preview will give you fantastic results at 1/20 to 1/100 the cost of Claude and the expensive flagship models. You can always choose to use different models later, if you find that there's any task google/gemini-3.1-flash-lite-preview doesn't succeed at.

Once you have Hermes installed, just ask it any question you have about accomplishing anything, and learn to use Hermes by exploring how to build solutions for your needs and interests. It's a lot like using a chat interface, except Hermes can actually do work directly on your server. It can access files and the command line on your server, it can access all sorts of built-in tools and skills, it can build and deploy software which you direct it to create, it can interface with any service you give it access to, etc.

Be careful what you ask it to do, especially if you give it credentials for any sensitive accounts!

EDIT: for the the easiest possible introduction to get started using AI to build software, in just a few minutes, on any computer you have, do this: https://aibynick.com/thread/29

If you're overwhelmed by where to start with all the current goings-on in AI, one the best entry points right now is to get to know Hermes agent.

Get an inexpensive VPS at Contabo or OVH for ~$50ish per year. If you're not comfortable with VPS, Linux, SSH, etc., ask Google, ChatGPT or any other LLM chat how to get started using a VPS and SSH, and take a look at https://com-pute.com/nick/linux_server_basics.txt

Open an account with Openrouter and set up an API key. This lets you use any of the hundreds of most popular commercial and open source LLMs, in your own locally hosted AI tools (those models are the 'brains' used by agentic AI harness applications such as Hermes).

Log into your VPS server with SSH, create a tmux session to run hermes, and run:

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

Have your OpenRouter API key ready, and I'd suggest choosing google/gemini-3.1-flash-lite-preview as the model to start with, when the installation asks. Google/gemini-3.1-flash-lite-preview is extraordinarily intelligent, it's extremely fast, and it's by far the least expensive choice among the frontier models, to serve as the brain of all your agentic tools. You don't need to pay for the costly fees associated with the Anthropic Claude models, or for any of the bigger, more expensive models, to do 99% of what you'll likely want to do when you're getting started with agentic workflows. Google/gemini-3.1-flash-lite-preview will give you fantastic results at 1/20 to 1/100 the cost of Claude and the expensive flagship models. You can always choose to use different models later, if you find that there's any task google/gemini-3.1-flash-lite-preview doesn't succeed at.

Once you have Hermes installed, just ask it any question you have about accomplishing anything, and learn to use Hermes by exploring how to build solutions for your needs and interests. It's a lot like using a chat interface, except Hermes can actually do work directly on your server. It can access files and the command line on your server, it can access all sorts of built-in tools and skills, it can build and deploy software which you direct it to create, it can interface with any service you give it access to, etc.

Be careful what you ask it to do, especially if you give it credentials for any sensitive accounts!

If you're overwhelmed by where to start with all the current goings-on in AI, one the best entry points right now is to get to know Hermes agent.

Get an inexpensive VPS at Contabo or OVH for ~$50ish per year. If you're not comfortable with VPS, Linux, SSH, etc., ask Google, ChatGPT or any other LLM chat how to get started using a VPS and SSH, and take a look at https://com-pute.com/nick/linux_server_basics.txt

Open an account with Openrouter and set up an API key. This lets you use any of the hundreds of most popular commercial and open source LLMs, in your own locally hosted AI tools (those models are the 'brains' used by agentic AI harness applications such as Hermes).

Log into your VPS server with SSH, create a tmux session to run hermes, and run:

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

Have your OpenRouter API key ready, and I'd suggest choosing google/gemini-3.1-flash-lite-preview as the model to start with, when the installation asks. Google/gemini-3.1-flash-lite-preview is extraordinarily intelligent, it's extremely fast, and it's by far the least expensive choice among the frontier models, to serve as the brain of all your agentic tools. You don't need to pay for the costly fees associated with the Anthropic Claude models, or for any of the bigger, more expensive models, to do 99% of what you'll likely want to do when you're getting started with agentic workflows. Google/gemini-3.1-flash-lite-preview will give you fantastic results at 1/20 to 1/100 the cost of Claude and the expensive flagship models. You can always choose to use different models later, if you find that there's any task google/gemini-3.1-flash-lite-preview doesn't succeed at.

Once you have Hermes installed, just ask it any question you have about accomplishing anything, and learn to use Hermes by exploring how to build solutions for your needs and interests. It's a lot like using a chat interface, except Hermes can actually do work directly on your server. It can access files and the command line on your server, it can access all sorts of built-in tools and skills, it can build and deploy software which you direct it to create, it can interface with any service you give it access to, etc.

Be careful what you ask it to do, especially if you give it credentials for any sensitive accounts!

If you're overwhelmed by where to start with all the current goings-on in AI, one the best entry points right now is to get to know Hermes agent.

Get an inexpensive VPS at Contabo or OVH for ~$50ish per year. If you're not comfortable with VPS, Linux, SSH, etc., ask Google, ChatGPT or any other LLM chat how to get started using a VPS and SSH, and take a look at https://com-pute.com/nick/linux_server_basics.txt

Open an account with Openrouter and set up an API key. This lets you use any of the hundreds of most popular commercial and open source LLMs, in your own locally hosted AI tools (those models are the 'brains' used by agentic AI harness applications such as Hermes).

Log into your VPS server with SSH, create a tmux session to run hermes, and run:

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

Have your OpenRouter API key ready, and I'd suggest choosing google/gemini-3.1-flash-lite-preview as the model to start with, when the installation asks. Google/gemini-3.1-flash-lite-preview is extraordinarily intelligent, it's extremely fast, and it's by far the least expensive choice among the frontier models, to serve as the brain of all your agentic tools. You don't need to pay for the costly fees associated with the Anthropic Claude models, or for any of the bigger, more expensive models, to do 99% of what you'll likely want to do when you're getting started with agentic workflows. Google/gemini-3.1-flash-lite-preview will give you fantastic results at 1/50 to 1/100 the cost of Claude and the expensive flagship models. You can always choose to use different models later, if you find that there's any task google/gemini-3.1-flash-lite-preview doesn't succeed at.

Once you have Hermes installed, just ask it any question you have about accomplishing anything, and learn to use Hermes by exploring how to build solutions for your needs and interests. It's a lot like using a chat interface, except Hermes can actually do work directly on your server. It can access files and the command line on your server, it can access all sorts of built-on tools and skills, it can build and deploy software which you direct it to create, it can interface with any service you give it access to, etc.

Be careful what you ask it to do, especially if you give it credentials for any sensitive accounts!

If you're overwhelmed by where to start with all the current goings-on in AI, one the best entry points right now is to get to know Hermes agent.

Get an inexpensive VPS at Contabo or OVH for ~$50ish per year. If you're not comfortable with VPS, Linux, SSH, etc., ask Google, ChatGPT or any other LLM chat how to get started using a VPS and SSH, and take a look at https://com-pute.com/nick/linux_server_basics.txt

Open an account with Openrouter and set up an API key, which lets you use any of the most popular commercial and open source LLMs in your own tools.

Log into your VPS server with SSH, create a tmux session to run hermes, and run:

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash

Then just ask it any question you have about anything, and learn to use Hermes. It's a lot like using a chat interface, except Hermes can do work directly on your server, it can access files and the command line on your server, it can access all sorts of built-on tools and skills, it can build and deploy software which you direct it to create, it can interface with any service you give it access to, etc.

Be careful what you ask it to do, especially if you give it credentials for any sensitive accounts!