← Back to Articles

NZ AI Engineer: Why I Cancelled ChatGPT After 2 Years ($690 Stack Revealed)

The exact tools, costs, and ROI behind my monthly AI spend

I spend $690 a month on AI tools. For some, that sounds excessive. For a business owner, it's leverage. One tool alone costs $200 a month, and it's the reason I finally cancelled my ChatGPT subscription after two full years.

In this post, I break down every single tool in my stack, the ROI on my business (and personal life), and a 4-step framework you can use to build your own stack without falling for shiny object syndrome.

The Daily Driver: Cursor

Harkness AI slide showing Google Workspace tools: Gemini Pro, Meeting Transcripts, NotebookLM, and Smart Email
Google Workspace integrations providing daily productivity tools

Cost: ~$106/month + usage

Cursor is my most used tool and a non-negotiable for coding. It's a VS Code fork with an embedded chat that edits code in real time. I use Gemini for UI design and GPT-4o for backend logic, swapping models as needed.

Best Bang for Buck: Google Workspace

Cost: ~$30/month

  • Gemini Pro integrated across the suite
  • Google Meet transcripts with summaries straight to Drive
  • NotebookLM for flashcards, reports, and even audio podcasts from docs
  • Smart email drafting and priority inbox in Gmail

The Heavy Hitter: Claude at $200/Month

Slide highlighting MCP Connections feature with red border emphasising native tool integrations
Claude's MCP connections enabling seamless tool integration

Claude replaced ChatGPT for me. It's my hub for personal automation, content, and complex reasoning. Four reasons:

  1. Claude Skills: Markdown "skills" keep reusable system instructions without copy/paste.
  2. Native MCP connections: Directly acts in tools (e.g., draft and send invoices in Xero, search Gmail) instead of copy-paste.
  3. Claude Code: Great for big architecture passes; I still use Cursor for tight edits.
  4. Artifacts: Render code, mini-apps, and sites inside chat when you don't want to deploy.

The Core Support Stack

  • Notion ($41.87/mo): Everything management - proposals, content calendar, scripts - with AI assists.
  • Make ($27.91/mo): Automation glue; aligns with what most Kiwi clients already use.
  • Canva ($20/mo): Design, decks, carousels; fast branded outputs.

Development Infrastructure

  • Supabase ($131/mo): Backend DB + auth (Google/Microsoft logins).
  • Vercel ($106/mo): Web deploys for projects.

The Math: Does $690/Month Make Sense?

If I bill $100/hour and save ~3 hours/day: 90 hours/month → $9,000 value. Even at $50/hour, that's $4,500 saved. Subtract $690 and it's still strongly net positive. It's not an expense; it's buying back time.

How to Build Your Own Stack (4-Step Guide)

Slide showing 4-step process: Identify Problems, Find Solutions, Trial Period, Stay Focused
Framework for building a custom AI and automation tech stack
  1. Identify problems first: Find the bottlenecks before picking tools.
  2. Find specific solutions: Best-in-class per bottleneck, not one "magic" tool.
  3. 30-day trial: Commit to configuring and learning; don't judge in 2 days.
  4. Avoid shiny object syndrome: Once it works, stop shopping and execute.

Ready to work smarter? Here’s how I help businesses across New Zealand implement these workflows and stacks.

Ready to move faster with AI?

Let’s Build Your AI Advantage

Book a free strategy call and I’ll show you exactly where AI can save you time, cut costs, and unlock growth in your business.

Book a Free Call
Not sure where to start?See how I help →
15+ NZ businesses helped
Free 30-min consultation
No obligation
Stay ahead on AI

Join Blake’s AI Briefing

Practical AI insights, tools, and updates for NZ businesses, delivered straight to your inbox. No spam, unsubscribe anytime.

Subscribe Free