Why AI is Replacing Virtual Assistants in Real Estate

Nineteen Months With a Virtual Assistant

Most articles about AI replacing virtual assistants are written by people who never hired one. I did — for about a year and a half, start to finish. This is what happened in between, and what it taught me about which work actually needs a person.

The Hire Was Good

The easy version of this story is "I hired badly." I didn't. She was the strongest candidate across multiple rounds of interviews, and it wasn't close. She sent a personalized intro video. She tailored her pitch to my actual listings. And when she started, she delivered: content creation, Canva graphics, social media work I couldn't have produced myself. For the first few months, hiring her felt like the smartest move I'd made since buying my first property.

The flaw was in the arrangement, not the person, and it was there from day one. I hired a VA because my full-time job left no time to run my rentals. But that same job is what kept me from managing her. Her working hours sat entirely inside my work schedule. I'd book check-ins and cancel half of them because something at my job caught fire. The reason I needed help was the exact reason I couldn't supervise the help.

The Collapse

The work was good, then thinner, and eventually output fell off a cliff — months where barely anything shipped. When I added meetings it rallied, which told me something I didn't want to hear. The work happened when I watched, and I was not available to watch.

At the end of the year I paid the full annual bonus anyway. Part of me wasn't sure it had been earned. I paid it because a year had been served and because I knew some of the management failure was mine. I still think that was right.

The Incentives That Went Nowhere

At the one-year mark I tried to fix a structural problem with money. I offered a choice: keep her salary and add performance incentives, take a small raise, or take a lower base with bigger upside. She kept the salary and took the incentives.

I offered a per-deal bonus. None were secured. I offered bonus revenue for screening and vetting tenant leads, and the leads sat there. I could not tell, in the moment, whether the status reports matched the work — and checking would have cost me the hours I was trying to buy back. A full year in, she was still asking questions that were answered in the knowledge base I'd built at her own request.

Eventually I offered a last choice: reduced hours or parting ways. Reduced hours were chosen, and the output did not recover. The arrangement ended.

It Wasn't a Bad Hire. It Was the Wrong Kind of Work.

I resisted this conclusion for months, because blaming her would have been easier. But look at the actual task list: check the inbox constantly. Answer the same rental question for the fortieth time. Chase every lead within minutes, at any hour, indefinitely. Log the expense, update the sheet, send the reminder, write the status report.

Nobody is good at that job for eight hours a day, because it isn't a job that rewards being good. It rewards being awake, being fast, and never getting bored, and humans are bad at all three. The work she was genuinely great at, design and content with a point of view, filled maybe four hours of my week. I was paying for forty.

Run the Numbers

Here's the market as of August 2026, from published pricing. Wishup, one of the larger managed VA services, lists its entry plan at $1,999 a month and a US-based assistant at $5,400 a month. Hiring directly overseas is cheaper: typical rates run $5 to $12 an hour depending on country and skill, which works out to roughly $900 to $2,000 a month full-time. That was my range.

The sticker price was never the real cost, though. The real cost was my time. Writing task briefs. Reviewing work I couldn't verify. Building tools she requested and then didn't use. Auditing status reports after I stopped trusting them. Every one of those hours came out of the same small budget of evenings that made me want help in the first place. If managing the help takes five hours a week, you haven't bought time back. You've traded one job for a worse one. If you want to run this math on your own numbers, our ROI calculator does the arithmetic.

What AI Actually Fixes

The case for AI on this category of work isn't intelligence. It's three boring properties. It's awake: a rental inquiry that arrives at 11:42pm gets answered at 11:43pm, not on Monday. It's fast: the best-known research on lead response, the Lead Response Management study run by Dr. James Oldroyd on InsideSales.com data, found that responding within five minutes instead of thirty makes you 21 times more likely to qualify a lead. And it's auditable: everything it does leaves a record. It cannot mark a lead "reviewed" without reading it, because reading it is the same operation. I found my VA's unopened leads by accident. Software doesn't get to be invisible.

What AI Doesn't Fix

I built an AI product, and I still keep humans, including me, on this list.

Showings. Someone has to unlock the door, and I think it should be the owner, because people tell you things in person they'll never type. Physical work: inspections, walking a unit with a contractor, driving a neighborhood. Judgment with real money attached: which applicant, which offer, whether the roof can wait a year. Relationships: your lender, your agent, the neighbor who texts when something looks off. And genuinely creative work. My VA's best graphics were better than anything I've generated with AI since. If your VA is great at that, keep her for that, at the hours that work actually fills.

If You're Deciding Right Now

Sort your tasks into two piles. Pile one is repetitive, time-sensitive, and verifiable: lead follow-up, inquiry responses, rent reminders, expense logging, reporting. Pile two is physical, creative, judgment, and relationships. Hire humans for pile two, and only pile two. Pile one should not need a person, and paying a person for it is how you end up where I was: paying someone to be bored, then spending my evenings checking whether the boredom won.

That's the honest lesson from nineteen months. It wasn't a bad hire. It was work that should never have needed a person, assigned to a person, managed by someone with no time to manage. Every part of that failure was structural, and swapping in a different VA wouldn't have changed it.

It's also why I built Doughy. It does pile one for landlords and investors: answers rental inquiries, follows up with leads by email, keeps the books current, and flags the decisions that need you, stopping at limits you set. It's in early access, and I'll be straight about the edges: capabilities like AI calling and texting roll out in stages, so you can't have all of it today. What you can have today answers in a minute at any hour, and it has never once marked a lead reviewed without reading it.