A spreadsheet has been the investor’s friend for thirty years. I built my first one in 1997, and I’ll be the first to admit I still keep a battered Excel file with macros that look like they were designed in a different geological era. So when AI-powered cash-flow modelling tools started turning up in real estate webinars two years ago, my reaction was somewhere between mild interest and active scepticism.

I’ve spent the last few months actually using six of these tools against real client scenarios. Some of them are useful. Some of them are spreadsheets with marketing budgets attached. Let me walk through what’s worth your time.

What these tools claim to do

The premise across the category is roughly the same. You feed in a property — either an address that gets enriched from public data, or manual entry — plus a few assumptions about your loan, your income tax situation, and your management approach. The tool then projects cash flow over the holding period, accounting for rental growth, vacancy assumptions, maintenance costs, depreciation, and tax outcomes.

The AI piece is where the tools differentiate. Some use AI just for natural-language input (“show me how this looks if rates go up 1% next year”) and traditional financial modelling underneath. Others claim to use machine learning to predict rent and vacancy outcomes based on suburb-level historical data. A few are starting to incorporate scenario modelling that simulates dozens of variable combinations and shows you probability distributions rather than point estimates.

What’s genuinely useful

Three capabilities have actually shifted how I look at deals when reviewing them with clients.

Scenario stress-testing at speed. The thing I do most often when reviewing an investment property is ask “what happens if interest rates go up half a percent, vacancy doubles, and the planned rent increase doesn’t land?” In a spreadsheet, that takes me a few minutes of careful clicking. With a decent AI tool I can verbally describe the scenario and get the answer in seconds. That speed lets me test more scenarios per deal, which is genuinely better diagnostics.

Sensitivity analysis on the variables you forgot. A spreadsheet only tests what you remember to test. The best of the AI tools will run sensitivity analysis on every input variable and flag which ones the outcome is most exposed to. I had a deal recently where the model surfaced that the deal’s outcome was hyper-sensitive to a 5% increase in body corp fees over five years, which was a variable I’d have ignored. That kind of “here’s what you didn’t think to ask” surfacing is where the AI genuinely beats my Excel habits.

Suburb-aware rental forecasts. Some tools pull granular rental growth and vacancy data from sources like CoreLogic and SQM Research and apply suburb-level historical patterns to forecasts. This is better than the 3.5%-per-year assumption that most investors plug in by default. It’s not perfect, but suburb-aware is meaningfully better than suburb-blind.

What’s still terrible

A few areas where these tools haven’t earned their keep yet.

Tax modelling. Every tool I’ve tested gets Australian negative gearing roughly right, depreciation about half right, and Capital Gains Tax modelling consistently wrong. The problem is that tax outcomes depend heavily on the investor’s broader financial position, which the tool doesn’t see. If the tool quotes you a tax outcome with specific dollar figures, treat that as a starting estimate and run the actual numbers with your accountant.

Maintenance and capex forecasting. AI tools that try to forecast maintenance costs based on property age and type sound clever in theory and produce nonsense in practice. The variability in actual maintenance costs from property to property is enormous, and machine learning struggles when the training data has high variance and low predictive structure. Most tools default to a percentage-of-rent assumption that’s no better than what a good spreadsheet does manually.

Settlement and acquisition costs. Several tools quietly underestimate stamp duty, conveyancing, lender fees, and the actual cost of getting the property into your portfolio. The numbers vary enough between states and lender types that the defaults are often wrong. Always check.

The output question: how to actually use them

Here’s where I diverge from the marketing material. These tools are useful as second opinions and scenario explorers. They aren’t useful as the primary source of truth for an investment decision. The reason is structural. A tool that gives you a single confident answer to “should I buy this property?” is hiding the uncertainty that’s actually in the data. The best tools surface the uncertainty. The worst ones bury it under a confident dashboard.

When I’m working with a client on a deal, I’ll often run the numbers through one of these tools to test scenarios, then independently work through the same scenarios in a spreadsheet I trust, then compare the answers. If the two diverge significantly, that’s the cue to dig into why. Sometimes the tool’s caught something I missed. Sometimes I’ve caught the tool fudging an assumption.

I’ve spoken to Team400 about how some of their consulting clients are building custom in-house versions of these tools, and the consistent feedback is that the off-the-shelf products are good enough for individual investors, but small property funds and SMSF managers usually benefit from something tailored to their specific portfolio structure. Worth thinking about if you’re operating at scale.

Tools worth a look

I won’t name them all, but a few specific category notes.

Tools built specifically for the Australian market are generally more useful than US-built tools with an Australia mode. Tax, depreciation, and rental yield norms differ enough that imported defaults will mislead you.

Tools that show you the working — variable assumptions, growth rates, vacancy assumptions — are vastly more trustworthy than tools that just give you a number. If you can’t audit the inputs, you can’t trust the output.

Tools that integrate with your accountant or broker are worth slightly more than tools that don’t. The exported model your accountant can actually read is worth more than the prettiest dashboard.

The honest verdict

If you’re a beginner investor doing one or two properties, a decent spreadsheet and good advice from your accountant will outperform any of these tools. You’ll learn more by building the model yourself, and the discipline of doing it manually catches errors that a slick UI will hide from you.

If you’ve got five or more properties or you’re moving fast on multiple opportunities, the speed advantage of the AI tools starts to outweigh the loss of intimacy with the numbers. The scenario-running is genuinely better than what you’ll do manually.

If you’re using AI cash-flow modelling as a substitute for understanding the numbers, you’re going to lose money. The tools don’t replace the discipline. They just make the discipline less tedious.

What I’m watching for the next twelve months is whether any of these tools start to integrate genuine portfolio-level optimisation, where the tool helps you understand which property to sell or refinance to improve overall portfolio efficiency. That’s where the AI piece could become really valuable. We’re not quite there yet, but the trajectory looks promising.

The spreadsheet isn’t dead. It’s just got more competition than it used to.