A deep dive into my hybrid cloud architecture for personal AI automation

AI and automation concept
The future of personal productivity is agentic AI

For the past few months, I have been running an experiment: what happens when you give an AI agent full access to your digital life and ask it to help you run it?

The answer? It is transformative.

I am not talking about asking ChatGPT to write emails. I am talking about an always-on AI assistant that:

  • Monitors my inbox and creates actionable tasks automatically
  • Prepares my daily meeting briefings before I wake up
  • Books tennis courts at my preferred times without me lifting a finger
  • Manages content across X, YouTube, and Xiaohongshu
  • Analyses junior tennis tournaments to help plan my son’s competitive season

This is not science fiction. This is my actual setup, running 24/7 on infrastructure I control.


Why I Built This

In my professional work, I architect enterprise AI and hyper-automation solutions. The same principles that drive multi-million dollar investment decisions can transform personal productivity too.

Digital transformation dashboard
From enterprise AI to personal productivity

The gap between “AI tools exist” and “AI actually does things for you” is massive. Most people are still copy-pasting between ChatGPT and their apps. I wanted something different: an AI agent that acts on my behalf.


The Architecture: Hybrid Cloud Meets Home Network

Here is where it gets technical.

My setup runs on OpenClaw – an open-source AI agent framework that lets you build personal AI assistants with real tool access. But the magic is in how I have architected it:

Cloud infrastructure
Hybrid cloud architecture combining accessibility with privacy

Technical Architecture Diagram

+===========================================================================+
|                      PERSONAL AI COMMAND CENTRE                           |
+===========================================================================+
|                                                                           |
|    +-------------------------------------------------------------------+  |
|    |                    CLOUD LAYER (Oracle Cloud VM)                  |  |
|    |                                                                   |  |
|    |    +--------------+     +--------------+     +----------------+  |  |
|    |    |   OpenClaw   |     |    Claude    |     |  External APIs |  |  |
|    |    |   Gateway    |     |   Opus 4     |     |                |  |  |
|    |    |  (Always On) |     |  (AI Brain)  |     | - Telegram     |  |  |
|    |    +------+-------+     +--------------+     | - Microsoft    |  |  |
|    |           |                                  | - X/Twitter    |  |  |
|    |           |                                  | - YouTube      |  |  |
|    |           |                                  +----------------+  |  |
|    +-----------|------------------------------------------------------+  |
|                |                                                          |
|                |  [TAILSCALE VPN - Encrypted Mesh Tunnel]                |
|                |  * No ports exposed on home router                       |
|                |  * Military-grade WireGuard encryption                   |
|                |                                                          |
|    +-----------v------------------------------------------------------+  |
|    |                     HOME NETWORK LAYER                           |  |
|    |                                                                   |  |
|    |  +------------------+        +---------------------------------+ |  |
|    |  |   Linux Server   |        |       n8n Workflow Engine       | |  |
|    |  |                  |        |                                 | |  |
|    |  | - Local Files    |        |  [Email Scanner]                | |  |
|    |  | - MCP Servers    |        |  [Meeting Prep]                 | |  |
|    |  | - Private Data   |        |  [ToDo Creator]                 | |  |
|    |  +------------------+        |  [Social Publisher]             | |  |
|    |                              +----------------+----------------+ |  |
|    |                                               |                  |  |
|    |  +--------------------------------------------v---------------+ |  |
|    |  |                   Automation Targets                        | |  |
|    |  |                                                             | |  |
|    |  |   [Tennis Booking]     [Tournament Analyser]                | |  |
|    |  |   [Social Publisher]   [Email-to-Task Pipeline]             | |  |
|    |  |   [YouTube Uploader]   [Calendar Sync]                      | |  |
|    |  +-------------------------------------------------------------+ |  |
|    +-------------------------------------------------------------------+  |
|                                                                           |
+===========================================================================+

Why Hybrid?

Cloud Gateway (Oracle Cloud Infrastructure)

  • Always reachable from anywhere in the world
  • Handles Telegram integration for real-time messaging
  • Runs the AI brain (Claude Opus 4) via API
  • Manages external API connections and authentication

Home Network (via Tailscale VPN)

  • Local file access – private data never leaves my network
  • n8n workflows run on-premises for sensitive operations
  • Custom MCP servers for specialised tool integrations
  • Zero exposure of home network to the public internet

The Secret Sauce: Tailscale – My home server connects to the cloud gateway through Tailscale’s encrypted mesh VPN. No ports exposed on my home router, military-grade encryption between nodes, and it works seamlessly when I am travelling.


What It Actually Does (Daily Operations)

Productivity and automation
Automated workflows handle the mundane while I focus on what matters

Morning Routine (Automated)

6:00 AM – Email scanner cron job runs:

  • Scans inbox for overnight emails
  • Identifies actionable items (bills, deadlines, responses needed)
  • Creates Microsoft To Do tasks automatically
  • Summarises key emails in Telegram

6:30 AM – Meeting prep runs:

  • Pulls today’s calendar events
  • Generates briefing notes for each meeting
  • Identifies any prep work needed
  • Sends summary to my Telegram

All before I wake up.

Tennis Automation

My son plays competitive junior tennis. Managing his season is a logistics challenge – until I automated it.

Court Booking Bot

  • Monitors local tennis club availability
  • Books preferred courts on our regular training day
  • Handles authentication and session management

ITF Tournament Analyser

  • Scrapes ITF junior tournament schedules
  • Cross-references with player UTR ratings
  • Identifies optimal tournaments for his level
  • Tracks potential opponents and match history

Social Media Orchestration

Social media management
One AI orchestrating multiple platforms

I maintain presence on multiple platforms. The AI helps coordinate:

Platform Automation
X (Twitter) Drafts thread ideas, schedules posts, monitors engagement
YouTube Automated video uploads, metadata generation, thumbnail suggestions
Xiaohongshu Custom MCP server for Chinese market publishing

Email to Action Pipeline

This is the workflow I am most proud of:

Email arrives
      |
      v
AI scans content
      |
      v
Classifies: Bill? Deadline? Action required? FYI?
      |
      v
If actionable --> Creates Microsoft To Do task
      |
      v
Files email to appropriate folder (FINANCE, WORK, FAMILY, etc.)
      |
      v
Sends Telegram summary if urgent

No more inbox overwhelm. No more forgotten bills.


The Tech Stack

Technology stack
Built on open-source and self-hosted tools
Component Technology Purpose
AI Gateway OpenClaw Agent orchestration and messaging
AI Brain Claude Opus 4 Reasoning and task execution
Cloud Infrastructure Oracle Cloud Always-on gateway hosting
VPN Tailscale Secure home network access
Workflow Automation n8n (self-hosted) Complex automation pipelines
Messaging Telegram Real-time AI interaction
Email Integration Microsoft Graph API Inbox access and To Do creation
Tool Extensions MCP Protocol Extensible custom tool integrations

5 Lessons Learned

Lessons learned
Hard-won insights from building AI automation
  1. Start with one workflow, then expand – I began with email scanning. Once that was solid, I added calendar, then tennis booking, then social media. Each addition builds on battle-tested infrastructure.
  2. Keep private data local – My emails, files, and personal info never leave my home network. The cloud gateway only sees what it needs to. Defence in depth.
  3. Hybrid beats pure cloud – Running everything in the cloud is simpler, but you lose control. Running everything locally is more secure, but limits accessibility. Hybrid gives you both.
  4. The AI is only as good as its tools – Claude is brilliant, but it cannot do anything without tools. The real work is building the integrations, the MCP servers, the n8n workflows that give it hands and feet.
  5. Document everything – My setup has extensive markdown documentation. When something breaks at 2 AM, future-me needs to understand what past-me was thinking.

The Bigger Picture

Future of AI
The future of personal computing is here

We are at an inflection point in personal computing. The smartphone put a computer in everyone’s pocket. Agentic AI is putting a personal assistant in everyone’s life.

But here is the thing: you do not have to wait for Big Tech to give you this. The tools exist today:

  • OpenClaw is open source
  • Claude’s API is accessible
  • n8n is free to self-host

The question is not whether you can build this. It is whether you will.

For me, the answer was clear. As someone who works on enterprise AI transformation professionally, I needed to live it personally. To understand the possibilities and pitfalls firsthand.

And honestly? I cannot imagine going back.


Want to Learn More?

I am happy to share more details about specific components:

  • OpenClaw setup and configuration
  • n8n workflow patterns for personal automation
  • MCP server development for custom tools
  • Tailscale architecture for hybrid deployments

Drop a comment or DM – let us build the future together.


Jack Hui is a digital transformation leader with expertise in enterprise AI and hyper-automation strategy. He holds an MBA from Melbourne Business School and is passionate about applying enterprise technology principles to personal productivity.

Oh hi there 👋
It’s nice to meet you.

Sign up to receive awesome AI content in your inbox, every time.

We don’t spam! Read our privacy policy for more info.