What is Woodhouse?
Woodhouse is first of all a learning experiment. A learning experiment that has the purpose of giving me knowledge and practical know-how of how AI works and can be used in real life.
The idea is to create a personal digital assistant who will have an AI intelligence, have access to a PC, my home network, some servers and cloud services. I call the assistant Woodhouse.
Being an AI assistant and given the resources it is up to me to find use cases and harvest the experiences from using AI to solve everyday tasks, projects and whatever may come our way.
There is no fixed end-game or final goal, as AI evolves so will this project.
To maximize my learning and experience harvest I have set up just one rule — Woodhouse does it all! That means whatever I need done or wish to complete, it should happen by me prompting Woodhouse to do it. If I need a webserver I should not install it before telling Woodhouse to use it. Woodhouse should download, install and configure it — after all he has his own PC and the knowledge of the whole internet, right?
You may be thinking that this sounds silly or far out… and yes, to both…. but this website, the one you are looking at now, is made entirely by Woodhouse — I have not once written anything other than instructions for what I would like. The code, the graphics, it is all Woodhouse's doing from my prompts. The development server we use to test is downloaded, installed and configured by Woodhouse. He also committed the code to GitHub before publishing it to the live server so you can see the result.
Still thinking this is silly? I do too… it is far out, but it is also fun and there are a lot of learnings for me while doing this. I will share some of my learnings and experiences on this page, mostly because forcing myself to write things down as articles really makes things stick to my mind, but maybe others will find inspiration to start their own learning journey!
Allan (and Woodhouse)
Current Integrations
- Telegram — primary interface for conversation and notifications
- Home Assistant — controls 16 lights via IKEA DIRIGERA hub, Google Home TTS
- Google Calendar — Boro fixtures sent as calendar invites
- Wazuh SIEM — home network security monitoring
- Dream Router 7 — syslog forwarding, threat intelligence
AI Models
Primary
Claude Sonnet 4-6
Conversation, reasoning, complex tasks
Research & Analysis
Gemini 2.5 Pro
Boro match analysis, web research
Background Tasks
Claude Haiku 4-5
Pulse nudges, Lighthouse log reviews
Sample Box One
This is placeholder content for the first box. Replace this with real content when ready.
Sample Box Two
This is placeholder content for the second box. Replace this with real content when ready.
Sample Full Width Box
This is placeholder content for the full-width box. Replace this with real content when ready.
Infrastructure
12 April 2026
Managing AI Costs
AI APIs are not free. Here is how I think about model selection to keep costs sensible without sacrificing quality.
Read more →
Tools
12 April 2026
OpenClaw & Obsidian
Using Obsidian as the project brain alongside OpenClaw — documentation, memory backup, and project management from anywhere.
Read more →
Configuration
12 April 2026
Rules & Behaviour
How you shape an AI assistant's personality and guardrails is more important than which model you pick.
Read more →
Automation
12 April 2026
The Jobs & Routines
From daily fitness nudges to security log reviews — how automated routines turn Woodhouse from a chatbot into something more.
Read more →