Frequently asked questions

Yes, scanning needs a free account, and every account gets 1 free teaser scan a day. No card required.

Photograph a For Sale board on your phone and Guru reads the agency, agent, reference and any QR code off it, then uses your coordinates to find the matching listing nearby. If the property is not advertised online you still get a street brief instead of nothing: what is being asked on that road, how much stock is around you, and how much of it is on show or under offer. Snap results are never link-shareable, because a snap contains a photo of a private home and where you were standing.

Any public Property24 or Private Property listing works best: paste the link and go. Guru will also attempt other public listing pages.

Teasers arrive in under a minute. A full report runs deep market research and typically takes a few minutes, and the page updates live while Guru works.

Every report has its own shareable link. Anyone with the link sees the property and the estimate; the paid market-intelligence widgets stay locked for everyone except the report's owner.

Yes, and a re-run is not just the same report again. Guru analyses the property from scratch with fresh data, then compares the result against your previous run: the report names what changed since last time, and labels the new estimate as consistent with the old one, a justified change, or pulled back to stay consistent. A re-run costs the same credits as the original.

Yes, and it is connected to your account rather than guessing. Ask what your last scan concluded, whether an estimate has finished, how your portfolio is performing, or what the market looks like in a suburb, and it answers from your live data and our own record of what we have concluded there before. It can also file a bug report or a feature request for you.

No. Reports are AI-generated estimates built from real market data. They are not a formal, sworn or bank valuation and are not prepared by a registered property valuer. See our AI Disclaimer.

The same way an analyst would: comparable sales, suburb price levels, official growth indices, and computer vision over the photos for condition and finish. The asking price is never allowed to become the answer: Guru derives its estimate from that evidence, then the server measures it against what the seller is asking and, if the two are within 2% of each other, blends it back toward a size-based baseline the AI never touched and prints the reason on the report.

Because they measure different things. Asking is what the seller hopes for; Guru's estimate is what the evidence supports. The gap between them is your negotiation room: the report quantifies it.

Confidence is earned from the data: how many genuine comparables were found, whether they are sales or asking prices, how tightly they agree, and whether the independent verifier signed off. Thin data means a lower score, shown honestly.

Blurred widgets are locked previews: they show the kind of intelligence a full report contains, with representative (not real) data behind the glass. Subscribing or spending credits unlocks the real thing.

Up to four independent lenses that each produce their own value (the closest comparable, a real sold anchor, suburb floor-rate maths, and your purchase price compounded at suburb growth), shown converging (or not) on Guru's number.

No, and this is the part worth understanding. The server owns the figures and the AI writes the explanation. Price per m², the independent baseline, the confidence range, bond repayments, transfer duty, total acquisition cost, yield and growth projections and the negotiation score are all computed in ordinary code, reproducible and covered by tests. The AI researches the market, reads the photos and explains what it found. Where the two disagree, the code wins and the report records the correction.

Yes. Every estimate is written to a ledger with the shape of the evidence behind it, and when the property later sells, is relisted or is withdrawn, that outcome is scored against the prediction: how far off, in which direction, and whether the real price fell inside the published range. Those scores are what set future confidence ranges, so in time the band on your report reflects where prices actually landed for estimates built on evidence like yours. Today most reports still use a conservative default: a category of evidence needs at least 30 scored outcomes before measurement is allowed to speak, and it can only ever widen a range, never narrow one.

Never. Every comparable is labelled as either a confirmed sale or a price someone is asking, and the label travels all the way from the source to the card you read. Asking evidence is also discounted toward what properties actually fetch before it is allowed to move your estimate, because a listing price is an opening bid and a transfer price is a fact.

No. Sectional title and freehold are different markets that happen to wear similar prices, so a flat is never used as evidence about a house (or the reverse). Guru also refuses to mix a land rate with a floor rate, because they measure different things, and it will use fewer comparables rather than pad the sample with irrelevant ones.

Guru steps down to the strongest evidence it does have and says which route it used, so a thin-data suburb still gets an answer, carrying only the confidence that route actually earns. Where even that is not possible, the report says "not enough data yet" rather than echoing the asking price back at you as though it were a valuation.

Three places: the official StatsSA Residential Property Price Index (monthly since 2010), registered average sale prices per suburb, and archived portal pages going back to around 2011. Every historical dot on a growth chart names and links its source.

The shape comes from the official index; the levels are anchored to real recorded figures. Where a stretch is index-estimated rather than recorded, the chart draws it dashed and the caption says which data tier shaped it.

Current homes on the market divided by the suburb's real monthly sales pace. Under about 4 months favours sellers; over about 6 favours buyers. Guru computes it from registered sales volumes, not guesses.

Yes, two ways: it sharpens your own estimate, and it joins anonymised suburb statistics, but a single owner's report never shapes a suburb curve. Several owners must corroborate a period before Guru trusts it.

Credits are the billing unit for AI reports: a full scan costs 20 credits and a full home value estimate 25. Your plan grants credits every month; top-up packs (40 credits for R99, 120 credits for R249, 300 credits for R549) add more whenever you need them.

Plan credits refresh (and reset) with every monthly cycle, even on annual plans. Top-up credits never expire and are only spent after your plan credits run out.

The discounted intro price applies once, on your first-ever subscription. If you cancel and later re-subscribe, the standard price applies from day one.

Anytime. Upgrades take effect immediately: you pay only the prorated difference for the remainder of your cycle. Downgrades are scheduled for your next renewal and never charge you anything.

Self-serve from your subscription page, anytime. Your plan stays active until the end of the paid period. New subscriptions also carry a 7-day statutory cooling-off period. See the Refund Policy.

A home value estimate runs the full pipeline (vision over your photos and documents, deep research, verification) with no listing to start from. It is our heaviest report, so it is part of a subscription or paid with credits.

Building plans, an offer to purchase, rates and levy statements. Guru reads them with vision AI and uses what it finds to sharpen the estimate.

Free accounts track up to 2 properties in My Properties; portfolio-tier plans lift the cap. Tracked properties get monthly auto-revaluation from real area appreciation, no AI cost to you.

It is an opt-in listing monitor on the Guru and Portfolio plans. Switch it on from a full scan report and Guru re-checks that listing every 3 days for price cuts, price increases, relistings and delistings, then emails you when the seller blinks. Pause or remove it any time from your Watching page.

Report and chat processing uses third-party AI providers (currently OpenAI, with Anthropic configurable), including photos and documents you upload for vision analysis. Our providers are contractually restricted from training models on our customers' data.

No. Suburb-level statistics are aggregated and anonymised; your account data, reports, photos and documents are yours. The Privacy Policy has the full picture, including your POPIA rights.

From your profile page: account deletion is self-serve and removes your data as described in the Privacy Policy.

Yes, the Guru chat answers up to 15 messages a day on free accounts and 50 for subscribers. It can file bug reports and feature requests for you, and can hand you to a human anytime.