For most South Africans, buying a home is the largest financial commitment they will ever make. It is also one of the least transparent. Listings sites show what is for sale, but they offer little insight into what a suburb is actually doing – whether prices are rising or stagnating, how long properties are sitting on the market, what the rental yields look like, or whether a new development across the road is about to flood the area with supply.
A small Cape Town start-up called FindHomes is trying to close that information gap. It describes itself as building “the property intelligence layer South Africa never had”.
Co-founded by Adrian Bunge and Jo Eyre, FindHomes has spent the past 18 months building a search and analytics platform that combines natural-language queries, image recognition, suburb-level investment data and an AI agent that produces on-demand market reports over WhatsApp. The company recently announced it is actively raising capital after a long bootstrapped run.
Bunge, the CEO and chief technology officer, comes from a family of property investors and has 12 years of software engineering experience across financial services and start-ups. He authored a guidebook called The Eight Laws of Property Investing in South Africa and is building the platform’s technical stack himself.
Eyre, the chief growth officer, brings a global marketing background spanning Epic Games’ Quixel and Unreal Engine divisions, the scaling of streaming service iflix to more than a million African users, and regional growth at Opera Mini.
Visual data
The origin story is personal. Bunge’s mother bought an apartment in Rondebosch seven years ago, at the peak of the market. Demand has since softened and several new developments have come online. To this day, the apartment cannot be sold for what she paid. “Totally avoidable mistake had we been able to access the data,” Bunge said in an interview with TechCentral this week. “We just didn’t know what we didn’t know.”
The diagnosis, that property buyers are flying blind in a market that hides the data they most need, is driving the product road map. FindHomes’ search interface accepts natural-language queries. A buyer can specify a 15-minute drive to a particular school, and the system, using Google’s location API, will surface only properties within that envelope.
Read: From app idea to board game hit
The platform also indexes the photographs in each listing, so a query like “kitchen with a mountain view” returns properties where the visual data – not just the agent’s written description – confirms the feature.
Beyond search, FindHomes generates analytical reports at the suburb level. These cover days on the market, the ratio of mandates that sell versus those that expire, months of supply, the frequency and depth of price cuts, and price-per-square-metre trends. The data is drawn from deeds records and the company’s own listings monitoring. A separate set of tools predicts rental yields by matching sales prices against rental data, addressing what Bunge calls the holy grail of property investing.
The most distinctive product is Lola, an AI agent that runs over WhatsApp and is currently in early access for Plus subscribers. Users can prompt it to compare suburbs, analyse individual listings against comparable sales or assess new-build developments. The choice of WhatsApp as the interface, Bunge said, was deliberate: it requires no app download and is the channel estate agents themselves prefer.
Avoiding hallucination is, by his admission, the company’s hardest engineering problem. FindHomes addresses it by routing the agent through real data sources rather than allowing it to generate from context, by getting it to execute code for any calculation, and by passing the output through a second smaller model for verification.
“These are multimillion-rand investments for people,” Bunge said. “Although we’re not giving investment advice, we really don’t want somebody to make a mistake.”
The data foundation is built largely on scraping estate agency websites. Bunge claims coverage sits at roughly 95%, and the company offers agents a takedown link if they want their listings removed. A recent partnership with geospatial data firm AfriGIS will add flood-risk overlays, with lightning-risk data planned for an eventual Gauteng expansion. For now, FindHomes operates only in the Western Cape.
Pricing
FindHomes offers a free tier alongside a paid Plus subscription:
- The free tier includes AI-powered search, commute-based filtering, price-per-square-metre analysis and customisable property alerts.
- The Plus tier, at R3 000 for three months – R1 000/month – unlocks historical sales data for all Cape Town suburbs going back to 2022, rental and yield forecasts on every for-sale listing, one-on-one expert guidance, and early access to Lola.
The team is bootstrapped and has just opened its first capital raise. Bunge said he is open to angel investment but is looking for backers who can offer industry introductions and mentorship rather than simply writing cheques. The company was recently a top-10 finalist at the Innovation City Startup of the Year Awards and presented in the Africa Tech Week pitching den.
The harder question is whether an intelligence-led product can hold ground against incumbents that already own the listings traffic and could, in principle, build similar AI features.
Read: R85-million for SA start-up reinventing the stethoscope with AI
FindHomes’ bet is that trust – accuracy, transparency, the explicit framing of property buyers rather than agents as the primary customer – is structurally difficult for portal businesses to replicate.
If the bet pays off, the company may end up doing something more interesting than competing on features. It may end up changing how the market itself behaves. – © 2026 NewsCentral Media
Get breaking news from TechCentral on WhatsApp. Sign up here.
