Most of the AI automation sold into Cape Town was designed somewhere else. That is
not a complaint about quality — a lot of it is good software. It is that the assumptions
underneath it were made for a market where people email, where the power stays on, where one
language is enough, and where GDPR is the law that matters. None of those four hold here, and
the gap shows up in the places that cost money.
What we build, and who for
AI agents that answer your enquiries — on WhatsApp, on the phone, on your website and in
your inbox — within seconds, at any hour, in your own words. They answer what they can,
ask what they must, book the job in your diary, and hand you anything that needs a person. We
also build the unglamorous half: the quoting, the invoicing, the follow-up that never happens
and the re-typing between systems that eats a week.
The clients are small and mid-sized businesses across the city and the province — the CBD
and Foreshore, the Southern Suburbs, the Northern Suburbs and Durbanville, the Atlantic
Seaboard, Somerset West and the Winelands, and up the West Coast. Trades, professional
services, practices and hospitality, mostly. If your enquiries arrive faster than people can
answer them, that is the problem we work on.
Why a Cape Town business loses the enquiry
Rarely on price. Almost always on silence. The message lands at twenty to eight on a weeknight
or halfway through a Saturday, nobody sees it until Monday, and by Monday the customer has
three quotes and a date booked with whoever replied first. Nothing about that was a pricing
decision.
Two local things sharpen it. The first is the channel: this is a WhatsApp market, and a
WhatsApp left unanswered for two hours reads as a business that is closed or does not want the
work. The clock a customer runs on WhatsApp is not the clock an email runs on. The second is
power. Load reduction still cuts supply across overloaded suburbs through the morning and
evening peaks, and the evening peak is exactly when domestic enquiries arrive. An agent hosted
properly keeps answering when your office cannot.
We have written up what the response-time research
actually says — including what it does not claim, and why most of it describes a
market that is not this one.
What it costs
It is on the page, in Rands, like everything else we sell. AI reception starts at R2 450
a month, the full front desk at R5 900, and custom builds are scoped after a paid
discovery phase rather than guessed at. The full price list is published
because anybody hiding their pricing is telling you something.
Before any of that, there is a week-long audit that counts what
actually reaches you and what gets answered, and puts a Rand figure on the gap using your
numbers rather than ours. It is the cheapest thing we sell and it regularly ends the
conversation, because a business already answering nine enquiries in ten inside ten minutes
does not need us.