karrari was running five vans out of two depots on a shared google sheet, three whatsapp groups and a whiteboard in the dispatcher's office. every new client doubled the coordination cost. a sick driver cost the morning.

they didn't need a shiny app — they needed the whiteboard to be accurate, the whatsapp to be searchable, and the sheet to stop breaking on monday mornings.

• a dispatch console for the office team — drag-and-drop jobs onto routes, live driver positions, one-click reassign.
• a driver pwa — no app-store install. route-of-the-day, proof-of-delivery photos, signed receipts.
• a customer portal — self-service booking, live tracking link, automated eta sms (via twilio).
• a weekly ops report that actually gets read, because it answers questions the ops manager was already asking.

two weeks in the dispatch office before we wrote a line of code. sarah sat behind the dispatcher; christian rode a route. we timed every handoff, mapped every decision, and asked "what would break this?" until we stopped being surprised.

then twelve weeks of weekly friday demos. nothing went to production that hadn't been used — in demo form — by the real ops team in the real office for at least three days.

the numbers after one quarter in production:

4.8×

drops dispatched per hour vs the spreadsheet baseline.

31h / wk

returned to the ops team — one full-time role redeployed to customer success.

zero

missed pickups in the first quarter post-launch (previously ~3 a week).

6 clients

onboarded in q2 2025 without adding ops headcount.

karrari dispatch v2 — currently in build. adding auto-routing (or-tools), predicted eta based on 18 months of our own telemetry, and a lightweight accounting tie-in so the ops manager stops also being the bookkeeper.