Product Design & Automation
AUTOMATING ORDER MANAGEMENT FOR A GROWING RESTAURANT
Emanuel Cocina was handling every order manually on WhatsApp. I designed a conversational system that automates the full flow, from menu browsing to order confirmation.
Year :
2025
Company :
TRIUM
Products :
WhatsApp Automation
Scope :
Conversational UX
Problem :
Emanuel Cocina is a restaurant in Bogotá with a growing customer base. Every order came through WhatsApp, and every order was handled manually: confirming availability, taking the order, calculating totals, coordinating delivery or pickup. No system, no history, no way to scale.
As volume grew, so did the friction. Missed messages, inconsistent pricing, no record of what sold or who the regulars were. The channel driving the business was also becoming its biggest operational bottleneck.
My role :
I identified the problem directly through conversations with the client, proposed the solution and designed the full conversational architecture, state logic, branching flows and the operations dashboard. API integration and technical development were handled by my co-founder and developer partner.
The solution :
A WhatsApp-native ordering system with two entry points: customers who message the number directly, and customers arriving from the restaurant's Treinta link. Both flows converge into the same ordering experience, adapted to where the user is coming from.
The system handles menu selection, cart management, delivery modality, differentiated pricing and full order confirmation, without the customer knowing they're talking to a bot.
Key design decisions:
Dynamic menu connected to Treinta in real time. Sold-out items don't appear. The menu the customer sees always reflects what's actually available.
Daily opening flow. Every morning, Emanuel activates the restaurant by sending "abrir" on WhatsApp. The bot asks for the daily sides, juice, and dessert sequentially. Those details get injected into every customer's welcome message, eliminating the most frequent inbound question.
Conditional branching by modality. Instead of a linear form, the WhatsApp Flow branches based on the customer's choice: delivery, pickup, or table reservation. Each path captures only the relevant data. Three screens instead of five.
Fixed packaging fee. Regardless of how many items are in the order, the packaging cost is a flat charge. A small decision that eliminated a consistent source of confusion and distrust at checkout.
Clean human handoff. If the bot fails to understand a customer twice in a row, it transfers automatically to Emanuel without exposing that anything went wrong. The customer always sees the same number throughout the conversation.
Operations Dashboard :
Beyond the conversational flow, I designed a dashboard where Emanuel manages the business: daily order panel, full order history, customer directory with frequency and spend data, and sales reports broken down by channel and modality.
The goal was to give the restaurant a real operational record, not just a messaging thread.
Current Status :
The conversational flow is built and tested in sandbox. The WhatsApp Flow form is complete. Waiting on Meta's production number approval before full launch.
What I Learned ::
Designing for a system that acts before the user makes decisions requires mapping more failure states than happy paths. The hardest part wasn't building the flow when everything worked. It was deciding what the system does when it doesn't understand, and making sure that moment never feels like a failure to the customer.
More Projects
Product Design & Automation
AUTOMATING ORDER MANAGEMENT FOR A GROWING RESTAURANT
Emanuel Cocina was handling every order manually on WhatsApp. I designed a conversational system that automates the full flow, from menu browsing to order confirmation.
Year :
2025
Company :
TRIUM
Products :
WhatsApp Automation
Scope :
Conversational UX
Problem :
Emanuel Cocina is a restaurant in Bogotá with a growing customer base. Every order came through WhatsApp, and every order was handled manually: confirming availability, taking the order, calculating totals, coordinating delivery or pickup. No system, no history, no way to scale.
As volume grew, so did the friction. Missed messages, inconsistent pricing, no record of what sold or who the regulars were. The channel driving the business was also becoming its biggest operational bottleneck.
My role :
I identified the problem directly through conversations with the client, proposed the solution and designed the full conversational architecture, state logic, branching flows and the operations dashboard. API integration and technical development were handled by my co-founder and developer partner.
The solution :
A WhatsApp-native ordering system with two entry points: customers who message the number directly, and customers arriving from the restaurant's Treinta link. Both flows converge into the same ordering experience, adapted to where the user is coming from.
The system handles menu selection, cart management, delivery modality, differentiated pricing and full order confirmation, without the customer knowing they're talking to a bot.
Key design decisions:
Dynamic menu connected to Treinta in real time. Sold-out items don't appear. The menu the customer sees always reflects what's actually available.
Daily opening flow. Every morning, Emanuel activates the restaurant by sending "abrir" on WhatsApp. The bot asks for the daily sides, juice, and dessert sequentially. Those details get injected into every customer's welcome message, eliminating the most frequent inbound question.
Conditional branching by modality. Instead of a linear form, the WhatsApp Flow branches based on the customer's choice: delivery, pickup, or table reservation. Each path captures only the relevant data. Three screens instead of five.
Fixed packaging fee. Regardless of how many items are in the order, the packaging cost is a flat charge. A small decision that eliminated a consistent source of confusion and distrust at checkout.
Clean human handoff. If the bot fails to understand a customer twice in a row, it transfers automatically to Emanuel without exposing that anything went wrong. The customer always sees the same number throughout the conversation.
Operations Dashboard :
Beyond the conversational flow, I designed a dashboard where Emanuel manages the business: daily order panel, full order history, customer directory with frequency and spend data, and sales reports broken down by channel and modality.
The goal was to give the restaurant a real operational record, not just a messaging thread.
Current Status :
The conversational flow is built and tested in sandbox. The WhatsApp Flow form is complete. Waiting on Meta's production number approval before full launch.
What I Learned ::
Designing for a system that acts before the user makes decisions requires mapping more failure states than happy paths. The hardest part wasn't building the flow when everything worked. It was deciding what the system does when it doesn't understand, and making sure that moment never feels like a failure to the customer.
More Projects
Product Design & Automation
AUTOMATING ORDER MANAGEMENT FOR A GROWING RESTAURANT
Emanuel Cocina was handling every order manually on WhatsApp. I designed a conversational system that automates the full flow, from menu browsing to order confirmation.
Year :
2025
Company :
TRIUM
Products :
WhatsApp Automation
Scope :
Conversational UX
Problem :
Emanuel Cocina is a restaurant in Bogotá with a growing customer base. Every order came through WhatsApp, and every order was handled manually: confirming availability, taking the order, calculating totals, coordinating delivery or pickup. No system, no history, no way to scale.
As volume grew, so did the friction. Missed messages, inconsistent pricing, no record of what sold or who the regulars were. The channel driving the business was also becoming its biggest operational bottleneck.
My role :
I identified the problem directly through conversations with the client, proposed the solution and designed the full conversational architecture, state logic, branching flows and the operations dashboard. API integration and technical development were handled by my co-founder and developer partner.
The solution :
A WhatsApp-native ordering system with two entry points: customers who message the number directly, and customers arriving from the restaurant's Treinta link. Both flows converge into the same ordering experience, adapted to where the user is coming from.
The system handles menu selection, cart management, delivery modality, differentiated pricing and full order confirmation, without the customer knowing they're talking to a bot.
Key design decisions:
Dynamic menu connected to Treinta in real time. Sold-out items don't appear. The menu the customer sees always reflects what's actually available.
Daily opening flow. Every morning, Emanuel activates the restaurant by sending "abrir" on WhatsApp. The bot asks for the daily sides, juice, and dessert sequentially. Those details get injected into every customer's welcome message, eliminating the most frequent inbound question.
Conditional branching by modality. Instead of a linear form, the WhatsApp Flow branches based on the customer's choice: delivery, pickup, or table reservation. Each path captures only the relevant data. Three screens instead of five.
Fixed packaging fee. Regardless of how many items are in the order, the packaging cost is a flat charge. A small decision that eliminated a consistent source of confusion and distrust at checkout.
Clean human handoff. If the bot fails to understand a customer twice in a row, it transfers automatically to Emanuel without exposing that anything went wrong. The customer always sees the same number throughout the conversation.
Operations Dashboard :
Beyond the conversational flow, I designed a dashboard where Emanuel manages the business: daily order panel, full order history, customer directory with frequency and spend data, and sales reports broken down by channel and modality.
The goal was to give the restaurant a real operational record, not just a messaging thread.
Current Status :
The conversational flow is built and tested in sandbox. The WhatsApp Flow form is complete. Waiting on Meta's production number approval before full launch.
What I Learned ::
Designing for a system that acts before the user makes decisions requires mapping more failure states than happy paths. The hardest part wasn't building the flow when everything worked. It was deciding what the system does when it doesn't understand, and making sure that moment never feels like a failure to the customer.