I Cancelled ChatGPT Business. Here's the AI Stack I Actually Run Now.

25 June 2026  |  By Timothy, CPA — Managing Director, Professional Financelink (PFL)
Multiple AI model subscriptions finance work routing Australia 2026

I Cancelled ChatGPT Business. Here's the AI Stack I Actually Run Now.

Three tools, not three subscriptions. I cut the most expensive one and I'm not missing it.

A Few Days Ago, I Cancelled ChatGPT Business

It wasn't a small decision — I'd been on it for a while, and it was genuinely useful. But sitting down to actually total up what I was paying across AI subscriptions, the ChatGPT Business cost stood out as the one doing the least to justify itself relative to everything else in the stack. With Claude already doing most of my heavy writing and analysis work, and Gemini sitting there for free as part of my existing Google Workspace subscription, ChatGPT Business had become the expensive middle option I kept paying for out of habit rather than need.

It's also been a genuinely disruptive few weeks for the AI tools market generally, which made it a reasonable moment to reassess. Anthropic's Claude Fable 5 launched, then was disabled worldwide — for every user, not just those outside the US — after the US government issued an export control directive on 12 June citing national security authorities. Because the order applied to foreign nationals regardless of location, and Anthropic couldn't reliably separate users by nationality in real time, the company disabled both Fable 5 and the underlying Mythos 5 model for all customers globally. Anthropic has said it's working to restore access and disputes the rationale, but hasn't confirmed a firm return date as I'm writing this. Separately, Google held back Gemini 3.5 Pro at I/O in May with a "next month" promise, and it's now landing through June. When the ground is moving that fast, it's a decent prompt to ask which subscriptions are actually earning their keep.

3 → 2
Paid AI subscriptions in my stack, down from three after cancelling ChatGPT Business.
12 Jun
Date of the US export control directive that led Anthropic to disable Fable 5 and Mythos 5 worldwide.

Why I Stopped Picking a "Best" Model

Early on, I tried to find the single best AI tool and standardise on it for everything — drafting, analysis, research, writing this blog. That approach made sense when the gap between providers was small. It stopped making sense once I noticed how differently each tool actually performed on the specific, repeated tasks I do every week — and it became a genuinely expensive habit once I added a third paid subscription on top without being disciplined about what it was actually replacing.

The pattern I've settled into isn't about chasing whichever model tops a benchmark this month. Benchmarks change constantly and most of them measure coding or reasoning puzzles that have little to do with what a finance manager actually does day to day. What's mattered more is noticing, task by task, where I keep getting better first-draft output with less correction needed afterward.

The Stack I'm Actually Running Now

This is genuinely just my own observed pattern, not a benchmark claim — your mileage may differ depending on your workflow and the specific tasks you're doing.

  • Claude — primary, paid: Long-form writing, including this blog, plus Korean-language work and anything where tone and structure matter for something published under my own name. This is where most of my actual finance writing and analysis happens day to day.
  • Gemini — secondary, included in Google Workspace: Large document and spreadsheet analysis, and anything that benefits from sitting directly inside Docs, Sheets, or Gmail. I'm not paying extra for this on top of my existing Workspace subscription, which makes it a genuinely free second opinion rather than another bill.
  • Gemma 4 — backup, planning a local setup: I'm working on getting this running locally as a backup option for lower-stakes tasks, mainly to have something that doesn't depend on a subscription or an internet connection at all. Still in the planning stage rather than daily use, but worth setting up properly.
  • Fact-checking and adjudication: I run drafts through more than one model before publishing anything with numbers or legal claims in it, and I've noticed each tool has different blind spots — on at least a few occasions, one has flagged accurate recent figures as implausible, which I traced back to training data lag. It's a useful reminder that "the AI disagrees" isn't the same as "the AI is right" — always check against a primary source when it matters.
📌 I've written before about treating AI fact-checking as a genuine cross-verification process rather than accepting any single tool's corrections at face value — that discipline matters more as the number of tools in your stack grows, not less.

Why ChatGPT Business Was the One to Go

I'd been using ChatGPT Business partly with an eye toward building AI agents on top of it — automating some of the repetitive finance and admin work that eats into a week. The catch I hadn't fully reckoned with: ChatGPT Business has a hard two-seat minimum, even for a one-person operation — OpenAI's own pricing confirms standard seats start at two, full stop. That worked out to roughly AU$70 a month for capacity I mostly wasn't using, just to get access to the agent-building features. Once I actually sat down and worked out what those features were costing me versus what I was getting from them, the maths didn't hold up — the same agent work was something I could build more directly, and far more cheaply, using Claude instead, without needing a separate platform with a forced minimum seat count.

For reference, Claude Pro runs at roughly AU$30 a month (it's billed in USD at $20, converted) on a single seat with no minimum. The two-seat minimum on ChatGPT Business alone made it well over double the cost of the tool that was actually doing the work I needed — and that's before counting the fact that I was the only person using it.

That's not a verdict on ChatGPT generally — if I'm honest, it's probably still the most well-rounded all-rounder of the three for everyday tasks, in my own experience. But "well-rounded" and "worth a premium business subscription on top of two other tools" are different questions, and for my specific workload, the second one came back no.

The Case for Running More Than One Tool At All

Cutting a subscription doesn't mean collapsing back to a single tool. There's a reasonable amount of research suggesting cross-checking across models genuinely helps — a 2026 CHI (Computer-Human Interaction) study on how people actually use multiple AI models in everyday life found that users develop deliberate primary and secondary tool hierarchies that shift depending on the task, rather than defaulting to one model for everything. That matches my own experience: the value isn't in having more tools, it's in having tools that genuinely disagree with each other often enough to catch things a single model would miss.

The actual cost of running multiple tools isn't the dollar figure — it's attention: remembering which tool you're in, re-explaining context across platforms, and the discipline required not to let "I have multiple AI tools" become an excuse for sloppier prompting because surely one of them will catch the gap.

That's the argument for starting with one tool, getting genuinely fluent in it, and only adding a second once you can clearly articulate what specific task the new one does that the first one doesn't — and being equally willing to cancel a tool once it stops clearing that bar.

⚠️ A note on AI tools and data privacy: if you're running real organisational data — financial figures, employee information, client details — through any AI assistant, check each platform's training and retention settings individually. They differ significantly between providers, and the default isn't always the most private option.

What I'd Actually Recommend

  • Start with one tool and get properly fluent before adding a second — the productivity gain from depth usually beats the gain from breadth early on
  • Check what's already included free in subscriptions you're paying for elsewhere — Gemini riding along on Google Workspace is the easy win most finance teams are leaving on the table
  • Before paying for a premium tier to access agent or automation features, check whether you could build the same thing more cheaply using a tool you already pay for
  • Never trust a single AI tool's fact-check on numbers or legal claims that matter — cross-verify, and check primary sources yourself when something feels off
  • Review your AI subscriptions periodically with the same discipline you'd apply to any other recurring business cost — cancel what's no longer earning its place
Knowing which AI tool actually fits a given finance workflow — and building the discipline around using it safely with real organisational data — is exactly the kind of practical AI automation guidance PFL provides. Find out more at professionalfinancelink.com.au.
About the author: Timothy is a CPA with 20+ years in finance leadership across NFP, NDIS and SME organisations, and Managing Director of Professional Financelink (PFL), which provides senior-level outsourced finance, management reporting, and AI automation for Australian NFP, NDIS, and SME organisations. Learn more at professionalfinancelink.com.au.
Sources:

Comments

Post a Comment

Popular posts from this blog

Google Gemma 4 Just Launched — And It Might Solve Finance's Biggest AI Privacy Problem

Image
Date: 16 April 2026  |  By: Timothy, CPA — Managing Director, Professional Financelink (PFL) ⚠️ AI privacy notice: This post discusses the use of AI tools for finance tasks. When using any AI tool — including local models described here — never input identifiable participant data, staff personal details, board-in-confidence material, or any information subject to confidentiality obligations without understanding exactly where that data goes and how it is handled. The boundary matters, regardless of which tool you use. There's a tension that most finance professionals who use AI haven't fully resolved yet. On one hand, AI tools are genuinely useful for finance work — drafting, summarising, reviewing contracts, analysing data. On the other hand, finance sits at the centre of an organisation's most sensitive information: payroll data, board papers, funding agreements, audit findings, legal correspondence. Putting that inform...
Read more

Claude vs Gemini for Australian Finance: An Honest Comparison After 12 Months of Using Both

Image
Date: 29 March 2026  |  By: Timothy, CPA  |  Managing Director, Professional Financelink (PFL) A note on data and confidentiality: The observations in this post are based on my personal use of these tools across various finance contexts over the past 12 months. In all cases where I've used AI tools with real financial data, I've either tested with anonymised data, synthetic test data, or information already in the public domain, and utilised temporary chats where possible so the data doesn't get saved anywhere. No confidential information has been used in any testing. If you're using AI tools in a professional finance context, I'd strongly recommend making this a standing practice in your own workflow — it's both good professional ethics and sensible risk management. One more thing worth flagging: if you're using any AI tool with work-related data, make sure you've either subscribed to an enterprise plan or disabled ...
Read more

Why NFP Boards Are Finally Talking About AI — And What the Finance Team Should Do Before They Ask

Image
Date: 13 April 2026  |  By: Timothy, CPA — Managing Director, Professional Financelink (PFL) Note: The scenarios in this post are based on real experiences — mine and those shared by colleagues across the NFP sector. Details have been modified slightly to protect confidentiality, and I've used a first-person perspective throughout for readability. It's happening more often now. A board member reads something about AI in the AFR or hears a presentation at a sector conference, and suddenly there's an agenda item: "What is our position on artificial intelligence?" For the finance team sitting in the room, this moment can go one of two ways. Either you're prepared — with a clear view of what AI is already being used in the organisation, what the risks are, and what governance gaps need addressing — or you're scrambling to answer a question you haven't fully thought through yet. I've been in both s...
Read more