Setup Guide

From zero to
running overnight

Getting AI Nightstand running is a straightforward process. Here's everything you need to know — from choosing your hardware to your first morning briefing.

Prerequisites

What you need

AI Nightstand runs on any machine capable of hosting a local LLM. You don't need a powerful GPU — a modest mini PC or even a spare laptop will handle most overnight tasks. The key ingredients are Ollama for model inference and a scheduler to trigger the agent at bedtime.

Minimal
Raspberry Pi 5 / Old Laptop
Runs small models (1–3B parameters). Good for email triage and calendar prep. Slower processing — start the agent earlier in the evening.
Recommended
Mini PC (16GB RAM)
Handles 7–13B models comfortably. Fast enough for full overnight workflows including news curation and draft generation. Beelink, MinisForum, or similar.
Power User
Home Lab / GPU Server
Runs 30B+ models. Handles complex reasoning, longer documents, multiple simultaneous tasks. Ideal if you already have an Ollama stack running.

Installation

Five steps to your first morning briefing

01

Install Ollama

Ollama handles local model inference. Install it on your host machine — Linux, macOS, or Windows. Takes about two minutes.

curl -fsSL https://ollama.com/install.sh | sh
02

Pull a model

Choose a model appropriate for your hardware. Mistral 7B is a solid all-rounder for most overnight tasks. Llama 3.1 8B is excellent for writing and summarization.

ollama pull mistral
03

Configure your agent

Edit the AI Nightstand config file to define your data sources, tasks, and output format. Tell it which email accounts to check, which news feeds to monitor, and how you want your briefing structured.

nano ~/.ainightstand/config.yaml
04

Schedule the agent

Use cron (Linux/Mac) or Task Scheduler (Windows) to trigger the agent at your preferred time. Most users set it to run around 11 PM so the briefing is ready by 6–7 AM.

crontab -e  # add: 0 23 * * * ainightstand run
05

Read your briefing

Open your morning briefing via browser, terminal, or email — your choice. Review drafted replies, scan the news digest, check your agenda. Your day starts with clarity.

open http://localhost:8080/briefing

Optional — Remote Access

Access your briefing anywhere

Want to read your morning briefing on your phone before you even get out of bed? Pair AI Nightstand with Caddy as a reverse proxy and Cloudflare Tunnel for secure remote access — no port forwarding required, no data leaving your network unencrypted. Your briefing stays on your server; you just read it remotely.

Ready to go deeper? We can help.

Hardware Guide Get Setup Help