Every six months I sit down with three or four buyer’s agent friends and we compare what we’re paying for. It’s a useful exercise because the SaaS bills creep up quietly, and what looked indispensable in February often hasn’t been opened by July.

This winter’s audit was more revealing than usual. A few tools that had been default picks for years have dropped off entirely, and the AI category has finally settled into something practical rather than experimental.

The Categories That Matter Now

A working buyer’s agent in 2026 is running tools across roughly five categories: CRM, due diligence, comparables and valuation, communication, and pipeline tracking. The stack used to sprawl across eight or nine. Some consolidation is happening, but not in the obvious places.

CRM is still where most agents start. AgentBox and Box+Dice remain the heavyweights for sales agents who happen to also do buying work. Pure buyer’s agents have mostly moved to Notion-based setups or one of the newer purpose-built tools like BuyersHub. The reason is simple — sales CRMs treat your “client” as a vendor by default, and the workflow assumptions never quite fit a buy-side relationship.

Due Diligence: The Quiet Revolution

This is where I’ve seen the most change in the last twelve months. PropTrack and CoreLogic are still the foundation — most agents I know subscribe to at least one — but the second layer of tools has shifted hard.

Suburb-level analysis tools that bundle CoreLogic data with planning overlays, school catchment maps, and infrastructure pipelines are now table stakes. The standout right now is the integration between CoreLogic and the state planning portals, which finally lets you check zoning, heritage overlays, and proposed amendments in one workflow rather than three. That has saved me probably 15 hours a month.

The thing that surprised me is how many agents still aren’t using satellite-imagery time-series tools to check past flooding, vegetation changes, or unapproved structures. These have been available for a couple of years and they’re cheap. If you’re doing investment-grade due diligence on regional properties, not using them is leaving money on the table.

Comparables and Valuation

I’ll be blunt — most automated valuation models are still not good enough to trust on borderline properties. They’re fine for ballpark, terrible for unusual stock. The two-bedroom apartment in a uniform high-rise? AVMs nail it. The Federation cottage in a heritage conservation area? Not even close.

What’s working is using the AVM as a sanity check on your own comp set rather than as the primary number. The agents getting in trouble are the ones who started outsourcing the actual comp work to the model. Vendors and buyers both notice when your justification thins out.

There’s a smaller wave of tools that pull recent off-market sales from agent networks. These are useful but require relationships to be valuable — the data is only as good as the agent who entered it.

Communication: Where AI Finally Stuck

The communication layer is where AI has stopped being theatre. A year ago everyone was experimenting with AI-drafted vendor emails and most of them were terrible. They had a tone problem and the legal exposure made everyone nervous.

What’s actually working in 2026 is much more targeted. AI is great at:

• Summarising contract amendments in plain English for your client
• Drafting first-pass shortlists from your weekly inspections
• Producing inspection notes from voice memos
• Translating strata minutes into the three things your client actually needs to know

What still doesn’t work well is AI-drafted offers or anything legally consequential. The agents getting burned are using AI to summarise contracts of sale and missing the unusual clause. Always read it yourself.

For more involved automation builds — connecting your inspection notes to your CRM and triggering follow-up sequences without losing the personal feel — a few of the agents I know have brought in an AI consultancy to wire it up properly. The DIY no-code stuff works until it doesn’t, and the cost of a broken automation in this industry is a missed offer.

Pipeline and Diary Management

Calendar tools haven’t changed much. Calendly is still ubiquitous. The newer arrivals like Reclaim have a fan base but I haven’t seen them stick widely. The thing that has changed is voice-to-CRM workflows — driving between inspections and dictating notes that end up cleanly in the right client file is a productivity gain I didn’t expect to care about as much as I do.

What Got Dropped This Cycle

A few categories have quietly died:

• Standalone email templating tools — superseded by CRM-native templates or AI drafts
• Generic project management tools — Notion or the CRM swallowed them
• Most virtual staging platforms — clients pushed back on the perception of misrepresentation
• Mass-text SMS tools — too spammy for buyer-side work

The buyer’s agent stack in mid-2026 is leaner than it was in 2024. It’s also more dependent on a handful of foundation platforms. Whether that’s a good thing depends on whether you trust PropTrack, CoreLogic, and your CRM vendor to behave themselves on pricing over the next few renewals. History suggests they won’t.

What I’d Build If Starting From Scratch

For a new buyer’s agent setting up today, my honest recommendation is a CRM (BuyersHub or Notion if you’re frugal), one full data subscription (PropTrack OR CoreLogic, not both for the first year), a transcription tool, and a single AI assistant for drafts and summaries. Total monthly cost around $450. Everything else can be added when you can articulate the specific problem it solves.

The agents who get into trouble are the ones who buy software hoping it’ll create the workflow. It doesn’t. The workflow has to be there first.