05 · Use cases

Nine patterns that fit Harare

These are patterns, not case studies. We describe how each works, what it needs and the local constraint it has to respect. We do not name clients, and we do not invent results. When a pattern has been deployed for a Harare business that agrees to be named, it will appear here with a date.

Pattern 01

WhatsApp order desk

Who: Produce wholesalers, distributors, tuckshop suppliers (Mbare, Graniteside)
How it works: A business number receives text and voice-note orders. An assistant transcribes and structures them (item, quantity, unit, delivery window, currency), writes a row to a shared sheet, and sends a confirmation the customer can correct with one reply. A human confirms price before dispatch.
What it needs: A WhatsApp Business number, a sheet or simple order table, a person who owns exceptions.
Harare constraint: Voice notes in Shona and English mixed; USD and ZiG quoted in the same thread; the stall has no laptop and prepaid data — everything lives in the thread.

Pattern 02

Fiscalised quote-to-invoice

Who: VAT-registered wholesalers, professional firms, pharmacies
How it works: Quote assembled from a price list with a rate date; buyer name, address, TIN and VAT number captured before invoicing; the invoice is issued through the fiscal device (FDMS) and its QR code sent back on WhatsApp; payment reminders follow a cadence.
What it needs: A fiscal device or fiscalised invoicing software with an API or import path; a customer master with TINs.
Harare constraint: ZIMRA FDMS buyer-detail transmission (deadline 31 May 2025); only QR-coded fiscal invoices count for input tax; dual-currency pricing.

Pattern 03

Appointment, reservation and recall loop

Who: Clinics, practices, restaurants, bank branches, salons
How it works: Booking captured on WhatsApp or from an existing diary; confirmation, reminder and reschedule messages sent automatically; no-shows re-offered slots; recalls (6-month, annual) generated from the last visit date.
What it needs: A diary or PMS/booking system, consent records, a fallback human for anything clinical or complaints.
Harare constraint: Consent under the Cyber and Data Protection Act; the reminder engine must be cloud-hosted so it survives a power cut at the practice.

Pattern 04

WhatsApp catalogue and reservation

Who: Mall retailers (Sam Levy’s, Avondale), boutiques, electronics
How it works: Catalogue maintained in WhatsApp Business; enquiries answered from stock and price lists; reservations held for a set window; pickup or delivery confirmed.
What it needs: A maintained stock list, someone updating it, a policy for holds.
Harare constraint: USD pricing expected; POS integration varies by tenant; keep images small for prepaid-data customers.

Pattern 05

WhatsApp concierge and group-booking desk

Who: Business hotels, lodges, conference venues
How it works: Enquiries triaged (room, group, event); quotes assembled from rate cards; arrival details, directions and outage-time notices sent proactively; escalation to reservations staff with the thread attached.
What it needs: PMS or rate card, a reservations owner, a decision on which currency leads the quote.
Harare constraint: Conference peaks around HICC dates; delegates on roaming or prepaid data; USD rates with ZiG option.

Pattern 06

Shipment status desk and document checklist

Who: Freight forwarders, clearing agents (Msasa)
How it works: Customer asks "where is my container?" and gets the consignment stage from a tracking sheet; a per-consignment checklist mirrors the customs pack and chases missing documents.
What it needs: A consignment tracker (even a sheet), a checklist agreed with the clearing team.
Harare constraint: The bottleneck is paper at the border, not the message; do not promise ETAs the corridor cannot keep.

Pattern 07

Proof of delivery

Who: Couriers and last-mile fleets (Graniteside, citywide)
How it works: Driver photographs the delivery and sends it to a job thread; the sender receives confirmation automatically; exceptions (no-one home, refused) trigger a re-attempt flow.
What it needs: Drivers’ own phones, a job list, a dispatcher.
Harare constraint: Offline-first capture with later sync; prepaid mobile data; no app installs beyond WhatsApp.

Pattern 08

Parent communications desk

Who: Private schools, tutoring centres, universities (applicant triage)
How it works: Fee reminders on a cadence with USD amount, ZiG equivalent and rate date; notices and absence reports; an admissions FAQ that hands off to the registrar.
What it needs: Fee schedule, guardian consent, a registrar who owns exceptions.
Harare constraint: Two currencies; small messages for ≈3.8 GB/month families; minors’ data.

Pattern 09

Listing lead qualifier

Who: Estate agencies, developers (Borrowdale corridor)
How it works: Enquiry captured with budget, currency, area, timing and financing; qualified leads routed to an agent with a viewing slot proposed; unqualified leads nurtured with matching listings.
What it needs: A listings feed, agent roster, viewing diary.
Harare constraint: Buyers quote USD; leads cluster in the northern suburbs; agents are in cars on mobile data.

What these patterns are not

None of them is "an AI chatbot". Several use a language model to transcribe, structure or draft; all of them end with a person who owns exceptions. For the difference between an agent, a chatbot and a workflow, and what each costs, read the category explainer on aiagent.co.zw.

Implementation of the WhatsApp, CRM and booking patterns is what our publisher does for a living: AI Automated Solutions Zimbabwe. On this site the disclosure lives on services in Harare; the patterns are here so you can evaluate any provider against them. Filter by district and sector in the index.