If you have read Claude vs ChatGPT for VAs, you already know we pay for both. Here is the wider tool stack underneath, what each thing costs in AUD, and which tools have been kicked off the list this year.

The standard per-VA stack (AUD/month)

Tool	Cost	What it does	Time saved/week
Claude Pro	$20	Writing, reasoning, code-adjacent	3-5 hrs
ChatGPT Plus	$20	Images, current-web, ecosystem	1-2 hrs
1Password Teams (per seat)	$12	Credential security	(security, not time)
Loom Starter (per seat)	$9	Async video updates	1 hr
Telegram	$0	Daily briefings + escalation	(process)
Per-VA standard total	$61		5-8 hrs/wk saved

At a $15/hr blended VA rate, 5-8 hours/week saved is $325-520 of recovered productivity per month. The stack pays back 5-8x.

Optional add-ons

Tool	Cost	When to add
Perplexity Pro	$30/mo	Specialist research VAs
Notion Plus	$14/mo	When team docs outgrow Google Docs
Grammarly Business	$20/mo	If your industry voice is unusually formal
Otter.ai Pro	$13/mo	High-volume meeting recap workflows
Make.com Core	$14/mo	Cross-tool automations beyond the free tier

We add these per-placement when the workload justifies them. Default: don’t.

Per-client tools (you pay, VA uses)

These are scoped seats on the client’s existing systems, not stack tools. Examples:

Shopify staff seat (free with most plans)
Xero advisor seat ($0-15/mo, depends on plan)
HubSpot/Pipedrive seat (per client’s plan)
Klaviyo (no per-user cost on most plans)
Helpdesk (Gorgias, Zendesk, Help Scout)
Practice management software (Cliniko, LEAP, etc)

Each is provisioned with the minimum permissions needed. Trust accounts, billing settings, and customer payment data are always scoped out unless the role explicitly requires write access.

What we kicked off the list in 2026

Notion AI ($10/mo) — Replaced by Claude Pro. The Notion AI features were 70 per cent overlap with Claude and 30 per cent worse. Killed.

Grammarly Premium ($20/mo) — Replaced by Claude with a “proofread, fix, return” prompt. Same job, no extra subscription.

Jasper ($49/mo) — Marketing-bro tone defaults that we always had to strip. Wrong fit for the warm-practical AU voice we use.

Calendly Premium ($16/mo) — Free Calendly tier plus a discipline of single-event-link-per-person works fine for most placements. Premium added cost without adding output.

Canva Pro ($16/mo per seat) — We use the free tier on the VA’s account plus client-provided Pro seats where needed.

Individual GPTs / Claude Projects per client — Adds noise. Better to keep the VA fluent on the base models with strong prompt libraries.

What about Microsoft Copilot, Gemini, etc

Microsoft Copilot: useful if your client is already deep in Microsoft 365 and you can get it as part of their existing subscription. Standalone not worth it for our use cases.

Gemini Advanced ($30/mo USD): comparable to ChatGPT Plus. We default to Claude+ChatGPT because the prompt library is calibrated for those two. If a client mandates Google Workspace, we will run Gemini alongside Claude.

Perplexity Pro: worth it for specialist research VAs only. Default placements don’t need it.

The two-tool rule

The biggest mistake we made in 2024 was over-tooling. We had VAs running Claude + ChatGPT + Perplexity + Jasper + Notion AI + Grammarly + Otter and rotating between them constantly.

The output got worse, not better. Context-switching between tools cost more time than each tool saved.

We pulled back to “deep on two, working on three more, none others”. Output quality went up. VA training time on new placements halved.

The rule for any new tool we consider adding: it must save the VA at least 30 minutes a week, the saving must hold up after 4 weeks of usage (not just the novelty), and it must not require switching context away from the existing tools mid-task.

Most tools fail rule 2 or 3.

Security stack (non-negotiable)

Three things every placement must have on day one:

1Password Teams seat ($12/mo) — Every credential lives here. No shared logins. Ever. When a VA offboards, access is revoked centrally.

A separate device for client work — VA uses their own laptop, but client work runs in a separate browser profile or a dedicated work browser (we recommend Brave or Edge for clean profile isolation). No personal browser extensions on the client side.

MFA on every account — All client accounts that support it. Authentication codes in the VA’s 1Password, not in SMS where possible.

This stack is roughly $12-15/mo per VA, before factoring in the productivity gains. We treat it as a fixed cost, not optional.

What this means for hiring

The stack maths is part of the placement quote. We bake the $60/mo standard stack into the agency margin so clients see one all-in number, not a list of subscriptions to manage.

If a client wants to bring their own tools (their Claude Pro account, their existing Loom seats), we credit it back. Most clients prefer the consolidated bill.

What’s next

For the comparison between the two AI tools, see Claude vs ChatGPT for VAs.

For the working Claude Code playbook, see How we run our VA team on Claude Code.

For the daily morning briefing that ties the stack together, see Building a Telegram briefing bot for your VA.

For a placement quote that includes this stack, book a discovery call.