13 AI systems that replaced the work of 20+ roles
Cemre Akkaya · Head of Marketing & Revenue, Desertcart (cross-border ecommerce, 60+ markets)
The situation
Running commercial operations across 60+ markets and 15 currencies, the same problems kept eating margin and time every month: manual reporting, no CRM, reactive support, fragmented data. The default answer is hiring. I built instead.
What I built (over 18 months, while running the P&L)
- Ads reporting: 10 Google Ads accounts across 6 currencies. Three people spent 2 days building the monthly report. My pipeline produces it in 40 seconds. Cost: $0.
- Multi-currency analytics: daily revenue attribution with FX normalisation across 15 currencies. Finance went from manual month-end views to a daily dashboard.
- B2B CRM: full pipeline and account management for an 11-market division, built in a weekend against a $100K/6-month vendor quote.
- AI support agent: monitors every ticket across 4,300+ SKUs, spots product and ops issues, fires 9 proactive alert types. Caught 3 supplier fulfilment failures in its first month.
- Lifecycle email: abandoned cart, replenishment, recommendation flows, fully automated, runs without intervention.
The result
13 systems shipped · 20+ roles of manual work replaced · all still running in production · built by a commercial operator, not an engineering team
Why it matters for your business
Most companies between €3-30M have five of these problems and no time to fix them. I find the manual process that costs the most, build the system once, and hand it over running.