Eight weeks to a real AI operating system
Kopi Bandung
We deployed AI OS across store ops, supply chain, and customer comms. Every team uses it. Every day.
The challenge
The problem
Kopi Bandung grew from 3 to 22 outlets in 18 months. Stock orders were done manually on WhatsApp each morning across all outlets. Customer reviews piled up in inboxes nobody checked. Shift reports took managers an hour each and were inconsistent. The founder had tried two AI tools before us. Neither one lasted past the first week.
What we built
How we solved it
Week one was an operational audit across five outlets. We mapped three workflows that were taking the most time and had clean enough data to automate: stock ordering, shift reporting, and review responses. Stock ordering became an agent that checks inventory each morning and sends outlet managers a pre-filled order with one-tap approval. Shift reports became a 10-minute structured input that the AI turns into an ops summary. Review responses became a single queue one person clears in 20 minutes instead of three scattered inboxes. Month two we added supply chain forecasting and a customer win-back agent.
The results
What happened after
96% of the team uses the system every day. Admin time across the ops team dropped 11 hours a week. Stock-outs fell 38% because orders stopped coming from memory and guesswork. The founder opened three more outlets in the same quarter.
Read another project.
Want results like this?
Book a fit call and we will scope your first build.
We will show you the first three things we would build and what each one returns.