Hire-a-VA math: when AI replaces the VA hire

The first solo freelancer I knew who hired a virtual assistant was thrilled for about three months. By month six, she was paying $1,200/mo for someone she had to keep re-training, who couldn't context-switch fast enough on her busiest weeks, who needed her to write briefs that took longer than just doing the task.

By month nine, she fired the VA, automated half the work, and started doing the other half herself in the time saved.

That was 2021. I think most freelancers who hired VAs between then and now have had a version of the same arc.

The math

A solo VA contract in the US/AU runs about $1,200–$2,000/mo for 20–30 hours of work. That's $50–$100/hour of agency time, against your $150–$250/hour billable rate. The arbitrage is real — if you can give them 20 hours of work that genuinely substitutes for your time, you're up.

But here's what actually happens:

• You spend 4 hours/week briefing them.
• They do 12–15 hours of actual output (the rest is interrupted by clarifications).
• You spend 2 hours/week reviewing and re-doing about 20% of it.

Net: ~12 hours of usable output for ~6 hours of your time. A 2x leverage, costing $1,500/mo.

That 2x leverage is real, but the band where it works is narrow. It exists for tasks where (a) the rubric is clear, (b) the consequence of an error is small, and (c) the volume is high enough to justify the briefing tax. Email triage and basic invoice prep fit. Drafting proposals doesn't. Anything involving judgment doesn't.

What AI changes

The narrow band where VAs were a good leverage point has been collapsing because AI does the same work without the briefing tax, the review tax, or the $1,500/mo:

• *Reading a prospect's email and extracting client/project/budget/timeline.* Same work a VA would do. AI does it in two seconds.
• *Drafting a proposal from a scope.* Same work. AI drafts faster, and you edit the draft directly.
• *Drafting a follow-up email for an unpaid invoice.* Same work. AI does it in the moment you'd otherwise be writing the brief.
• *Categorizing time entries for invoicing.* Same work. AI categorizes as you log.

For all of those, the briefing tax is zero — the AI sees the same data you see. The review tax is small. The cost is in cents per task, not hundreds per month.

What AI doesn't change

What a VA can do that AI can't:

• *Phone calls.* AI doesn't pick up the phone.
• *Strategic relationship work.* AI doesn't take the client to lunch.
• *Embodied tasks.* AI doesn't ship the swag, hand-deliver the contract, or visit the print shop.
• *Persistent memory across a long project.* A great VA carries the project in their head for months. AI can be given context, but the gap is real — it's shrinking, not closed.

If your VA mostly does the first list — extraction, drafting, categorizing — AI eats most of that. If your VA does the second list — phones, embodied tasks, relationships — they're irreplaceable.

When the VA hire still makes sense

• Your client volume requires a real human relationship layer that you don't have time for.
• You're at the income tier where 20 hours/week of admin freed up turns into $5,000+ of billable work — not extra hobbies.
• You've reached the ceiling of what AI can do for your specific workflow (it happens — niche industries with unstructured inputs).

If you're spending $1,500/mo on a VA whose work is "send the proposal," "follow up on the invoice," "write the welcome doc" — that math doesn't hold up in 2026.

The 2026 setup that actually works

For solo freelancers at $80–$300k/year, the kit I see hold up:

• *AI* for extraction, drafting, categorizing.
• *You* for judgment, relationships, taste.
• *A part-time bookkeeper or accountant* for tax-specific work (not a generalist VA).
• *Nothing else.*

The VA layer has compressed. That's not a bad thing — it's $1,500/mo back in your pocket, plus the leverage of the AI on top.

— Jhayden