← Back to case studies Case Study

AI Operating System for Consulting Work

A live example of applied AI proficiency: agents, scheduled monitoring, browser automation, knowledge capture, and coding workflows used to run real work.

Project: Personal AI operating system and consulting delivery stack
Tech: OpenClaw, Claude Code, Codex, n8n, Playwright, Python, Notion, Discord
Role: Strategy, delivery, and hands-on implementation

The Results

This is the way I now work: AI agents handle monitoring, drafting, analysis, coding support, and knowledge capture while I keep judgement, prioritisation, and accountability in the loop.

24/7
Scheduled checks for services, inboxes, dashboards, and operational drift
Multi-agent
Subagents, browser automation, coding agents, and workflow tools coordinated around tasks
Human-led
Automation designed around review, approval, and real-world usefulness

The Challenge

Most AI claims are vague. The practical challenge is turning AI from a chat window into a dependable work system: one that remembers context, checks live state, escalates appropriately, and helps ship actual deliverables.

What mattered

The system has to be useful under messy real-world conditions: changing priorities, private data, brittle websites, partial information, and tasks that span code, operations, clients, and personal admin.

The Solution

I use OpenClaw as an agentic command layer, Claude Code and Codex for implementation, n8n for workflow automation, Playwright/browser automation for web tasks, and structured memory/Notion/Discord loops for continuity.

Implementation Highlights

Agentic monitoring

Service health, email triage, Kanban, calendar, trading, and client dashboards are checked on schedules with stateful deduping.

AI-assisted delivery

PRDs, code changes, testing, browser checks, and deployment tasks are accelerated with Claude Code, Codex, and subagents.

Knowledge loop

Decisions, tasks, context, and durable learnings are captured into Notion, memory, and Obsidian rather than disappearing into chat history.

OpenClaw Claude Code Codex n8n Playwright Python Notion Discord

Impact

The result is a consulting workflow that is faster, more observant, and more systematic: AI does the repetitive scanning and drafting while human judgement stays focused on what matters.

Before

Ad hoc
AI as isolated prompts and manual follow-up

After

Systematic
Agents, schedules, memory, and delivery workflows

Key Takeaways

AI proficiency is operational, not theatrical

The signal is not knowing tool names; it is using them safely and repeatedly to reduce friction in real work.

The best systems keep humans in control

Approvals, audit trails, and escalation points matter as much as automation speed.

Want AI That Actually Fits Your Work?

I help teams move from scattered AI experiments to practical, governed workflows that save time and improve delivery.

Get in Touch